<%BANNER%>

Locating Ethnic Context: Mother's Characteristics and Child Mortality in Trinidad and Tobago


PAGE 1

LOCATING ETHNIC CONTEXT: MOTHERS CHARACTERISTICS AND CHILD MORTALITY IN TRINIDAD AND TOBAGO By KUNIKO CHIJIWA A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS UNIVERSITY OF FLORIDA 2003

PAGE 2

Copyright 2003 by Kuniko Chijiwa

PAGE 3

To my brother and my sister

PAGE 4

ACKNOWLEDGMENTS My deepest appreciation first goes to my committee members, Dr. Barbara Zsembik and Dr. Charles Wood. Thanks go to my supervisor Dr. Zsembik for being instrumental in the development of all aspects of this study. She opened my eyes to vital issues in studying ethnicity in relation to health and quality of life. Especially, she led me to the world of data philosophy. Dr. Wood originally stimulated my interests in data analysis and its application to racial and ethnic studies. He also facilitated the basis of this study. I especially owe him for his patience in teaching me the enjoyment of data analysis. I also owe a debt to Trinidadians following my initial research in Trinidad for my thesis in Latin American Studies. This time, I especially owe Dr. Beni N. Balkaran of the Mt. Hope Hospital, Dr. Robert Lee of the Caribbean Epidemiology Center, and Dr. Victor Coombs for providing information on childrens and mothers health and child mortality; and Ms. Elizabeth Welsh of the Ministry of Health and Ms. Raynette Pierre of the Central Statistical Office for providing vital statistics. My special thanks go to Rajnie Ramlakhan, Esther Langoo and her family members for their sincere and everlasting friendship, which nourished me richly while I was in Trinidad. I would like to express my special gratitude to Macro International Inc. for allowing me to use the dataset from the Demographic and Health Survey, Trinidad and Tobago 1987. This survey established the framework of this thesis. iv

PAGE 5

Finally, I am very grateful to my family who supported me financially and emotionally, understood me, and allowed me to be selfish. I thank my best friend for providing constant moral support and encouragement; and my cats who reminded me to relax from time to time. v

PAGE 6

TABLE OF CONTENTS Page ACKNOWLEDGMENTS.................................................................................................iv TABLE OF CONTENTS...................................................................................................vi LIST OF TABLES.............................................................................................................ix LIST OF FIGURES.............................................................................................................x ABSTRACT.......................................................................................................................xi CHAPTER 1 INTRODUCTION........................................................................................................1 Background...................................................................................................................2 Significance of the Study of Child Mortality........................................................2 Ethnic Relations, Social Allocation, and Study of Child Mortality in TT............4 Practical Significance...................................................................................................6 Theoretical Significance.............................................................................................11 Conceptual Framework for Child Survival.................................................................15 2 LITERATURE LEVIEW...........................................................................................19 Trinidad and Tobago...................................................................................................19 Defining Differences between Ethnicity and Race.............................................19 Ethnic Context in Trinidad and Tobago..............................................................22 Slavery to collective identity........................................................................23 Ethnicity and class consciousness................................................................27 Properties of Child Mortality......................................................................................31 Difference between Infant Mortality and Child Mortality..................................31 Determinants of Child Mortality.........................................................................33 Socioeconomic and Demographic Variables on Child Mortality........................34 Income..........................................................................................................34 Maternal education.......................................................................................35 Marital status and residential characteristics................................................35 Intervening Health Care Variables in Relation to Socioeconomic Variables.....37 Prenatal care.................................................................................................38 Preventive health care..................................................................................39 vi

PAGE 7

Breastfeeding................................................................................................40 Type of place where the child is born..........................................................42 Infant and Child Mortality in Trinidad and Tobago............................................44 Hypotheses..................................................................................................................47 3 RESEARCH DESIGN AND METHODS..................................................................49 Data and Sample Size Analyses.................................................................................49 Measures.....................................................................................................................52 Child Mortality....................................................................................................53 Demographic Factors...........................................................................................55 Ethnicity.......................................................................................................55 Type of place of residence...........................................................................55 Marital status................................................................................................56 Socioeconomic Factors........................................................................................57 Maternal educational attainment..................................................................57 Quality of life...............................................................................................57 Health Related Factors........................................................................................60 Prenatal care.................................................................................................61 Type of place where the child is born..........................................................61 Quality of preventive health care for child...................................................62 Breastfeeding................................................................................................64 Procedures of Data Analysis.......................................................................................65 4 DATA ANALYSIS....................................................................................................67 Ethnic Differentials in Child Mortality and Its Determinants....................................67 Demographic Characteristics...............................................................................67 Socioeconomic Characteristics............................................................................70 Child Health Care Practices.................................................................................71 Influence of Maternal Characteristics on Child Mortality..........................................73 Influence of Demographic and Socioeconomic Factors......................................73 Health-related Proximate Factors........................................................................76 Multivariate Logistic Regression Models...................................................................79 Ethnic Influence on Child Mortality....................................................................79 Urban-Rural Setting, Marital Status, and Scio-economic Influences..................80 Influence of Health-related Factors on Child Mortality......................................82 5 CONCLUSION...........................................................................................................84 Ethnicity in the Analysis of Child Mortality..............................................................84 Locating Ethnic Context in Trinidad and Tobago......................................................90 LIST OF VARIABLES AND VALUES OF THE DEMOGRAPHIC AND HEALTH SURVEY IN TRINIDAD AND TOBAGO 1987 DATA..........................................97 vii

PAGE 8

LIST OF REFERENCES.................................................................................................106 BIOGRAPHICAL SKETCH...........................................................................................114 viii

PAGE 9

LIST OF TABLES Table page 1-1 Cause of infant mortality, 1988..................................................................................7 1-2 Cause of child mortality, 1988...................................................................................7 1-3 Causes of infant mortality, 1997................................................................................8 1-4 Cause of child mortality, 1997...................................................................................8 3-1 Distribution of women 15 to 49 by ethnic and type of place of residence in 1987 and 1990, Trinidad and Tobago...............................................................................50 3-2 Variable descriptions................................................................................................54 3-3 Rotated component matrix for 8 variables of household composition.....................59 4-1 Characteristics of mothers with children born in last 10 years by ethnicity............68 4-2 Proportions and odds for mothers who have lost at least one child and correlations between ethnicity and maternal characteristics........................................................75 4-3 Probability of having lost at least one child controlling for demographic, socioeconomic, and health care factors (Logistic Regression)................................81 ix

PAGE 10

LIST OF FIGURES Figure page 1 Conceptual framework for the analysis of the effects of socio-economic and socio-cultural factors on child survival..............................................................................17 2 Infant and under 5 mortality in Trinidad and Tobago and Latin American and Caribbean, 1960 to 2000..........................................................................................44 x

PAGE 11

Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts LOCATING ETHNIC CONTEXT: MOTHERS CHARACTERISTICS AND CHILD MORTALITY IN TRINIDAD AND TOBAGO By Kuniko Chijiwa May 2003 Chair: Barbara A. Zsembik Major Department: Sociology The objective of this study is to examine ethnic differentials in child mortality in Trinidad and Tobago. Child mortality is considered as an outcome of mothers demographic and socio-economic characteristics and quality of health care for a child. The discriminatory perceptions that incite ethnic conflicts over socioeconomic and political allocations between the two major ethnic groups, African and East Indian, were contextualized within Trinibagonian history reflecting the conditions of social inequality experienced by each ethnic group. In this study, child mortality serves as a variable to ascertain the socio-economic and cultural differences between African and East Indian for capturing a unique social stratification system in this ethnically polarized society. The Demographic and Health Survey in Trinidad and Tobago is employed to determine whether ethnicity differentiates child mortality. The data analysis consists of multivariate logistic regression models using nine explanatory variables that are divided into three clusters; demographic factors, socioeconomic factors, and health care factors. xi

PAGE 12

The analysis is organized in order to empirically test hypotheses that are derived from theoretical perspectives of Colemans social system considering the child survivorship framework originated by Mosley and Chen. Colemans social system theory is concerned with the balance and dynamics of competing interests in and controls over scarce resources between groups, which contribute to the construction of the social structure. In Trinidad and Tobago, ethnicity is considered as a major motive to pursue the common social and economic benefits through which we can locate the status of individuals and groups in the social structure. Logistic regression analysis on child mortality indicates that after controlling for all demographic, socioeconomic, and health care factors, ethnicity is statistically significant; and confirms that African mothers have more than twice the higher risk of child loss that East Indian mothers have. The beneficial impacts of health care factors are found. No interaction terms between ethnicity and the three health care factors are significant, which means that each health care factor works independently from ethnicity. Thus, better health care associates with decrease of child mortality universally for all women in this nation. The inclusion of health care factors, however, widens the child mortality gap between them, implying that the ethnic gap in child mortality cannot be attributed to health care factors. Therefore, Africans continue to be disadvantaged in child mortality. There must be continued efforts toward improving the socio-economic status and quality of health care of Africans. The analysis also implies that there may be misspecifications beyond factors associated with socioeconomic standings and quality of health care that may be culturally influencing their health practices and behaviors. xii

PAGE 13

CHAPTER 1 INTRODUCTION The objective of this study is to examine ethnic differentials in child mortality in the Republic of Trinidad and Tobago (henceforth called TT). Child mortality is considered as an outcome of mothers overall quality of life and as a variable, which can be used as an analytical category to ascertain economic and cultural differences between the two major ethnic groups, African and East Indian. This thesis utilizes a cross-sectional survey, the Demographic and Health Survey in Trinidad and Tobago, 1987 (henceforth called the TTDHS), to determine whether ethnic backgrounds differentiate child mortality among mothers in TT. In light of the social and economic context, the study focuses on specifying the different dimensions of the concept social inequality and on analyzing the way in which various factors affect child survivorship. If we control for the socioeconomic determinants and health care factors of child mortality, the question raised is, whether the children of women who belong to the disadvantaged ethnic group in child mortality continue to experience higher death rates than those born to the advantaged ethnic group in child mortality. In the case where ethnic background continues to be statistically significant after controlling for key social economic factors and health care factors, the results may suggest that the higher child mortality group is subject to additional disadvantages beyond factors associated with socioeconomic standings and health care factors; there may be cultural elements influencing their health practices and behaviors. 1

PAGE 14

2 For the objective of this study, the following specific aims were set: to determine whether the ethnic background of the mother significantly influences the probability of child mortality; to assess whether ethnic background differentiates the association between socioeconomic status and child mortality; to assess whether ethnic background differentiates the association between health care practices and child mortality; and to measure the overall magnitude of ethnic influence on child mortality in TT. Achieving these aims may improve our understanding of the characteristics of ethnic differences in well-being and health in TT. In the multi-ethnic communities within a nation like TT, it is important to improve and renovate health care systems taking into consideration cultural and behavioral characteristics in order to reduce the inequalities in quality of health. The quality of health care for a particular ethnic group may be largely influenced by socioeconomic status, by accessibility to facilities, and by availability of the kind of medical care the ethnic group prefers. Background Significance of the Study of Child Mortality Studies of race and ethnic differences in health have attracted the interests of researchers in analyzing health-related needs and problems in order to provide necessary and accessible health services for people from various ethnic backgrounds due to the changing population composition. Racial or ethnic background is a social construct in any kind of society and is associated with mothers absolute health status during the perinatal period, which may directly or indirectly affects fetuses, neonates, infants, and children. Because some ethnic groups are comparatively more likely to fall into disadvantaged social categories, in that they are more likely to be poor, unmarried, and

PAGE 15

3 less educated, they will have higher mortality which implies that observed mortality differences are primarily compositional in nature (LeClere et al. 1997). Because of the extreme dependency, childrens health and survival chances rely on their parents. Therefore, child mortality is considered as an important indicator of the well-being of a population (Birdsall 1980, Eberstein 1989, Wood and Lovell 1992, Hummer et al. 1999). Quality of child health care and maternal education as indices for socioeconomic status have been found to be highly correlated with infant and child health and mortality (Cramer 1987, Hogue et al. 1987, Mangold and Powell-Griner 1991). Income, which is closely related to occupation and education, exercises an important effect on the ability to obtain medical provisions (Hobcraft, et al. 1984). Thus income is an influential factor in consideration of individual socioeconomic status and is an important determinant of child mortality (Gortmaker 1979, Mosley and Chen 1984, Hummer 1993). A higher incidence of poor pregnancy outcomes and child mortality among women from disadvantaged socioeconomic backgrounds has been indicated in previous research on determinants of child mortality. Socioeconomic variables also serve as indicators of knowledge and make-up of medical services. Therefore, it is likely that people of higher socioeconomic status groups will be better able not only to afford drugs and other expensive care but also have access to valuable information on pregnancy and child bearing. They are also likely to have better housing and are more likely to be connected to water supplies and sewage systems. Education expands the role of the socioeconomic variable by disseminating knowledge on medical sanitary requirements. This knowledge can range from simple elements of child health care involving cleanliness and sterilization to more complex knowledge of

PAGE 16

4 what drugs and vaccinations are required; and the ability to find and use services (Hobcraft et al. 1984). Hence, socioeconomic status, specifically income, can be considered as a pivot of human resources, however, the mere existence of differentials by income, does not constitute a satisfactory explanation of the association between these income differences and child mortality (Gortmaker 1979). Because of the situation of children, especially new born babies, depending entirely on their parents, child morality is considered as a sensitive measure of their parents quality of life associated with different socioeconomic environments. Child mortality is often regarded of an extreme case of poor child health, which mirrors the parents life, in that it is an aggregate of realities derived from the parents circumstances and everyday choices. Although child mortality has been reduced by substantial improvements in parents standard of living, educational attainment, and access to medical care during the recent decades, differences persist. There are a variety of latent factors related to child mortality. To reduce such inequality within a multi-cultural society, we need a better understanding of the mechanisms of the relations between ethnicity and mothers characteristics; and of the possible reasons why specific ethnic groups continue experiencing lower quality of life and higher mortality. This understanding will help provide general health services to entire nations and the accessible health services to specific ethnic groups. Ethnic Relations, Social Allocation, and Study of Child Mortality in TT The southernmost island formation in the Caribbean Archipelago, TT is the nation composed of twin islands. An oiland gas-rich republic of 1.16 million people, TT is evenly split along ethnic lines with slightly less than half its people of East Indian descent (40.3%) and the same number of African descent (39.6%). The distinct ethnic identity of

PAGE 17

5 each has created a modern culturally bi-polarized society within the nation. The political and economic rivalry between them has been extensively studied by researchers. Many of them found evidence of differences between the two ethnic groups in terms of their aspirations, income distribution, social mobility, political behavior, and occupational placement -each of which has been formed in its history of development (Harewood and Henry 1985, Henry 1988, Selwyn Ryan 1991 1999, Center for Ethnic Studies 1993, Yelvington 1995). While such differences in economic superiority and social mobility between the two groups have attracted a great deal of scholarly attention, differences in health care practices derived from or largely influenced by their respective cultures which encourage or restrain their acceptance of modern medical practices, have not been well studied. Accessibility to health care services is concerned with allocational issues within a nation, since it is considered both a universal right for the entire nation and a privilege for a people who can afford a considerably higher quality of medical care for severe injury and prolonged illness. Health care is a fundamental right when one considers that societies have a duty to preserve life and also promote quality of life. Health care is a combination of intertwined social structure, which reflects economic, social and political inequalities, and the allocation of scarce resources. Health is a condition of physical, mental and emotional well-being (Fuligni and Brooks-Gunn 2000, Singer and Ryff 2001) that all people should enjoy, hence, health should not be a privilege for the few, but like education, should be universal; everyone should have access to it, regardless of age, religion, ethnicity, culture, nationality, or social class. In this respect, health care assumes social and economic importance as various groups within society jostle for access to this

PAGE 18

6 vital resource, which inevitably appears to be scarce. However, the right to something that is a scarce commodity in many developing countries is what creates dissatisfaction and tension among social collectives (Marshall and Mahabir 2000). A study of health requires investigating a complex combination of socioeconomic and cultural characteristics; physical, practical, and behavioral. The cultural variations can significantly predict child survivorship, which is disaggregated by social indicators, such as place of residence, household income, and parents level of educational attainment. Cultural variation may characterize the socioeconomic differences between the two major ethnic groups by examining the differences in child mortality. The results from these efficient indicators, which have been historically established by interactions between the two ethnic groups, provide valuable insights into an ethnically polarized societys stratification system. Practical Significance Table 1 presents the causes of infant mortality and child mortality in 1988 and 1997. In both 1988 and 1997, any incidence that occurs during the perinatal period was ranked the number 1 cause for infant mortality, followed by congenital anomaly and pneumonia in 1988, infectious and parasitic disease in 1997. The trend of both the rates for infant mortality and child mortality, which have continuously shown a downward curve since 1950s, appears to be reaching a low point. Table 1-1 and 1-3 provide list of cause of infant mortality which show that problems during the perinatal period remain the top cause; moreover, the problem gains strength as a driving factor behind infant mortality: 62.8% in 1988 and 69.6% in 1997. Informal interviews with medical doctors and epidemiological researchers in Trinidad, summer 2002, reveal current major concerns of infant mortality in TT in terms

PAGE 19

7 Table 1-1. Cause of infant mortality, 1988 Causes of DeathMaleFemaleTotalCertain Conditions Originating in the Perinatal Period12781208Congenital Anomalies242448Pneumonia17724Infectious and Parasitic Diseases7411Signs, Symptoms, and Ill-defined Conditions6410Injury and Poisoning3710Accidents and Adverse Effects178Diseases of the Circulatory System325Other Protein-Calorie Malnutrition202Malignant Neoplasms (Leukaemia)011Nutritional Marasmus101Anaemias101Influenza011Appendicitis011Total192139331 Source: Deaths Report 1988. Central Statistical Office, 1990. Table 1-2. Cause of child mortality, 1988 Causes of DeathMaleFemaleTotalInjury and Poisoning13922Accidents and Adverse Effects11920Congenital Anomalies21012Infectious and Parasitic Diseases7310Diseases of the Circulatory System516Pneumonia426Malignant Neoplasms (Leukaemia)235Anaemias-33Bronchitis, Emphysema, and Asthma123Signs, Symptoms, and Ill-defined Conditions213Other Protein-Calorie Malnutrition2-2Meningitis112Certain Conditions Originating in the Perinatal Period1-1Total 514495 Source: Deaths Report 1988. Central Statistical Office, 1990.

PAGE 20

8 Table 1-3. Causes of infant mortality, 1997 Causes of DeathMaleFemaleBothCertain Conditions Originating in the Perinatal Period11996215Congenital Anomalies252348Infectious and Parasitic Diseases4711Pneumonia6511Injury and Poisoning336Accidents and Adverse Effects336Signs, Symptoms, and Ill-defined Conditions325Diseases of the Circulatory System123Malignant Neoplasms (Leukemia)1-1Diabetes Mellitus-11Nutritional Marasmus-11Chronic Liver Disease and Cirrhosis1-1Total 166143309 Source: Population and Vital Statistics Report 1997. Central Statistical Office, 1990. Table 1-4. Cause of child mortality, 1997 Causes of DeathMaleFemaleBothInjury and Poisoning8412Congenital Anomalies549Accidents and Adverse Effects639Infectious and Parasitic Diseases527Malignant Neoplasms (Leukemia)347Diseases of the Circulatory System134Pneumonia134Signs, Symptoms, and Ill-defined Conditions2-2Homicide112Meningitis-11Bronchitis, Emphysema, and Asthma-11Direct Obstetric Deaths-11Total 322759 Source: Population and Vital Statistics Report 1997. Central Statistical Office, 1990.

PAGE 21

9 of ethnic differences. One concern was the fundamental transformation of dietary habits, which has caused serious health related problems, especially the potential for diabetes among women during pregnancy. The other is sexual activity, which has played a major role in increasing HIV/AIDS cases in TT. Recent research conducted by the Caribbean Epidemiology Centre reported that approximately 3% of new born babies have been infected with HIV/AIDS at birth and some 5% of the causes of death among children are related to HIV/AIDS. Mother-to-child transmission of HIV/AIDS has become a major problem in TT with up to three infants being infected everyday, via this route, assuming an HIV prevalence of between 2 and 3 % among pregnant women. An estimated 1,806 adults and children were newly infected with HIV/AIDS during 2000 (Caribbean Epidemiology Center 2001). These facts are not temporal social phenomena, but rather represent the inherent socio-cultural differences between the two ethnic groups. It is believed rapid dietary habit transitions are largely due to the Americanization of food culture that affects especially the people of East Indian descent. Higher prevalence of HIV/AIDS appears among the people of African descent. Due to the relatively lower infant and child mortality rate, the study of infant and child mortality has drawn little attention in TT. Few, if any, comparative studies on ethnic differences in terms of infant and child mortality in TT have been conducted. Although the TTDHS collected ethnic data, researchers have not attempted to use ethnicity for determining differences on health issues between African and East Indian women. The national statistical data on infant and child mortality comparing ethnic groups does not even exist, despite the fact that ethnic issues and allocations in every term are always prime interests among Trinibagonians. TT has a relatively more

PAGE 22

10 prosperous economic environment compared to other Caribbean nations. Oil production has in fact contributed to facilitating development of medical institutions and health development programs thus, lowering the infant and child mortality rate. With the pride of not being a developing county, the TT government no longer refers child mortality as an indicator of the level of quality of life among the nation, instead, they aim to expand and upgrade the public health facilities in delivering high quality health care to every citizen. As a small and oil-rich island, TT nations shall achieve their aim in the near future. However, infant and child mortality rates are widely regarded among researchers as variable indicators of the physical well-being of children as much literature has insisted (Birdsall 1980, Eberstein 1989, Wood and Lovell 1992, Hummer et al. 1999). To understand the differences in child survival associated with ethnicity is important for health policies and interventions. These differentials identify the highest risk group across ethnic groups, indicating the need to overcome the socioeconomic inequalities between ethnic groups, and the need for development health programs to take the inequalities into consideration. Given the numerous issues associated with ethnic strife over socioeconomic and political allocations, health care allocation is not exception to this; health care services should be considered a limited package of resources. Curiously, ethnic context has rarely appeared as one of the features of the country profile in research on child health and mortality (Harewood 1978, Heath et al. 1988, Marshall and Mahabir 2000, UNICEF 2003) as already mentioned. Hence this study will be a springboard for advancing our awareness of ethnic differences in terms of health, health care, and child mortality as accumulations of and confounding effects of

PAGE 23

11 socioeconomic and cultural factors in TT, and also, for more complete research on the association between ethnicity, child mortality, health behaviors, and quality of life. Theoretical Significance In the extant research of health and child mortality, socioeconomic status has been used extensively as an explanatory variable that typically measures the extent to which socioeconomic background is related with health. It is also used as a control variable in looking at other correlates of child health and mortality. Hence, taking into consideration the significance of socioeconomic influence on child mortality, and for the purpose of determining whether ethnic background overwhelms the relationships between child mortality and socioeconomic level, social stratification theory is considered first as providing theoretical guidance. The concept of social structure, social class, and socioeconomic status are central to the study of child mortality in social sciences. The theorists of the stratification school take a structural-functional approach. While structural functionalism considers a society composed of interdependent elements such as culture, personalities, and social systems, the theory of social stratification gives more weight to socioeconomic activities than to culture and personalities. Structural-functional analysis of social stratification is concerned primarily with the roles played by such socioeconomic activities, which maintain social structure. Heavy rewards in valued goods are given to motivate individuals to perform important social functions, with the heaviest reward being given to those occupying positions of functional importance in the society for which qualifications in the society were relatively rare (Persons 1940, Davis and Moore 1944). Although functional stratification theorists take cultural variations and personalities into consideration in the frame of functions, they mainly hold that the cohesive and

PAGE 24

12 integrative power of socioeconomic class linkages horizontal lines as it were surpasses the divisive power of the vertical lines, which divide one ethnic group from another (Braithwaite 1960). Structural functionalists insist that there must be a certain minimum of common shared values if the unity of the society is to be maintained. Hence, a structural-functional approach suggests that after controlling for social strata as measured by education and economic status, ethnic differences will not be statistically significant. Therefore, structural functionalism, in the same manner as traditional social stratification theory, chiefly attempts to describe the structure of social stratification based on the differences among people in terms of such criteria as wealth, income, occupation, education, descent, property, and prestige, and to specify the processes by which the social system is generated and maintained (Cuff et al. 1998). In their view, social system can be held together by a consensus on economic norms and values in spite of distinct cultural and ethnic diversity. It is important that child mortality research involves investigating how levels of inequality and variation in social context affect health outcomes. Also, in multi-ethnic society, socioeconomic measurements may need to capture more of the social context than the indices of income, education or occupational position can provide. Social context is derived from such factors as community, networks, and environment that child mortality research appears most interested. The variables of socioeconomic status in explaining the difference of child health outcome, and child mortality may be described more inclusively when they involve the social context influenced by cultural context, no matter what the degree of influence is. In this sense, a structural-functional approach has the same weakness as traditional stratification theory; both overlook cultural norms in

PAGE 25

13 quality of life and quality of health. Hence, the traditional stratification theory and the structural functional approach are not very useful in explaining the societies composed of many ethnic groups who devise their empirical measures either by determining the distributional characteristics of social stratification systems, or by identifying the positions of individuals, families, or other social groups in such systems (Oakes and Rossi 2003). Culture affects our perceptions and experiences of health and health care in many ways. Health care within any group can be affected by a multitude of cultural variables; some very basic, some more complex. It is a measure of human flexibility with diverse ways and means of meeting human needs (Loustaunau and Sobo 1997). Hence, Oakes and Rossi suggest that it is better to start with the question, what would be an ideal socioeconomic status measure. Such a measure is described by Nock and Rossi (1979), cited by Oakes and Rossi (2003: 7-8) that, socioeconomic status is that dimension of stratification which translates the objective distribution of social resources into meaningful perceptions of relative desirability. This concept holds that ethnic and cultural collectivity are elements of diversifying ways by which people share and distinguish the perceptions and meanings, and means of meeting human needs related to health and well-being. In reality, measurements of socioeconomic status are almost entirely represented by education and income, as well as occupational position, which are obtained from census type data due to its availability. Therefore, it may be useful to consider Colemans social system theory defining the value and the role of social capital in the creation of human capital (1988, 1990), which is also based on social stratification theory. The central idea

PAGE 26

14 of Colemans notion concerns how positions constituting social structure emerge, and how persons are motivated to occupy such positions. Coleman sets forth three ideas. The first is material capital, which refers to owned materials such as household composition and income that are tangible and analyzable. The next is human capital, which refers to inherited physical appearance and ability as well as education, skills, abilities, and knowledge one may acquire with ones investment. The last is social capital, which includes obligations to and from others, information channels, norms, and reputation effects. To possess social capital, a person must be related to others, and it is in the potential of those relationships where social capital lies (Oakes and Rossi 2003). Portes (1998) explains that, there is growing consensus that social capital stands for the ability of actors to secure benefits by virtue of membership in social networks and other social structures. Coleman emphasizes these networks and functions as a necessary condition for the rational action paradigm: Just as physical capital and human capital facilitate productive activity, social capital does as well. For example, a group within which there is extensive trustworthiness and extensive trust is able to accomplish much more than a comparable group without that trustworthiness and trust (1988: S101). The trustworthiness, in his term the role of closure, refers to obligations, expectations, and social norms, which are largely influenced by a persons cultural perception and his experience. The existence of sufficient ties among a certain number of people guarantees the observance of norms. The collective perception of inequality emerges from the balance and dynamics of interests and control over scarce resources. For understanding the interrelationship between the socioeconomic structure and ethnicity in a multi-cultural nation, it is meaningful to consider the ethnic differences in the value of social capital and its role in the creation of human capital within a group. Oakes and Rossi indicate;

PAGE 27

15 There are several advantages to incorporate social capital into a measure of SES. It provides an understanding of the variation in social contexts. --And social capital assists in understanding the all important micro-macro (man to structure and structure to man) transitions, and thus family and neighborhood and institutional level impacts and outcomes (2003: 777). Thus, social capital can provide a mechanism through which behavioral norms are generated and maintained, and can promise to provide a link between individuals, society, and health as a human capital. In previous studies, the influence of socioeconomic status on child mortality is significant (Gortmaker 1979, Mosley and Chen 1984, Cramer 1987, Hogue et al. 1987, Mangold and Powell-Griner 1991, Hummer 1993), and the mothers health care practices for her child such as prenatal care, breastfeeding, and immunization have a significant impact as intervening factors (Rosensweig and Schultz 1982, Goldberg et al. 1984, Huffman 1984, Maison et al. 1987, Trussell et al. 1991, Kadende 1994, Chaulagai 1993, Humphreys et al. 1998, Alan Ryan 1998, Forste 2001). Colemans social system theory can provide a theoretical framework for the linkage between health and behavioral norms, individual perceptions, and social collectivities. It is also appropriate in guiding the examination of whether or not ethnic identity overwhelms the associations between child mortality and the two clusters of predictive variables, socioeconomic and health care factors, and in defining the stratification system of TT society. Conceptual Framework for Child Survival The previous research provides evidence that child mortality is an indicator of mothers social well-being. Ultimately mother health outcomes are enmeshed in a web of causality. To comprehend the relationships between child mortality and various confounding factors, ideally, we would wish to have child specific information related to health status; illness, nutritional inputs and growth that would allow controls for health

PAGE 28

16 heterogeneity in TT. Unfortunately, census type data generally does not include such information. Therefore it is useful to have a framework conceptualizing how we understand the connections among the proximate factors, socioeconomic factors (education, household composition and housing quality), demographic factors (place of residence and marital status), and health related factors (record of immunizations for children, prenatal care, place of child born, breastfeeding) which we can obtain from the TTDHS related to child survival. The framework for the study of child survival in developing countries, which has had a major influence on the Demographic Health Survey (Boerma 1996), was first presented by Chen in 1983 and developed by Mosley and Chen in 1984. In the Mosley-Chen framework, a set of proximate or intermediate determinants, which directly link to the risk of child morbidity and mortality, are divided into five socioeconomic factors: material factors, environmental contamination with infectious agents, availability of nutrients to the fetus and infant, injuries, and personal illness control (Mosley and Chen 1984). All social and economic determinants, such as mothers education and household income, operate through the proximate determinants to affect child growth and mortality. Proximate determinant framework, however, has met criticism from researchers because it is more likely to lead to research focusing on individual level decision-making rather than on broader society processes as a result of its complex web of factors influencing behavior or analyses at other levels such as family and community (Ewbank 1994, Boerma 1996). Since the major interest of this study is to determine the relationship between ethnic background and child mortality as a reflection of mothers socioeconomic status, the

PAGE 29

17 Figure 1. Conceptual framework for the analysis of the effects of socio-economic and socio-cultural factors on child survival

PAGE 30

18 conceptual and analytical framework of this study needs some modifications with inclusion of factors rooted in their ethnic background. Figure 1 presents a broader conceptual framework. Socio-cultural factors, which are associated with ethnic background, serve as key covariates influencing or operating on a mother and childs health care and standards of living at both the individual and community levels. In the case that ethnic differences are not statistically significant in the presence of socioeconomic determinants and health care interventions of child mortality, then, as the United Nations reports, the variations in mortality across ethnic and race groups probably reflect mainly differences in such factors as socioeconomic status and accessibility of health facilities and services, rather than innate differences among the groups themselves (1985: 77). This account can be understood in the way that differences in socioeconomic status and accessibility of health facilities and quality of health services may be nested in ethnic background. If racial differences in child mortality appear, and they are statistically significant after controlling for socioeconomic factors and health care factors, then the higher child mortality ethnic group is subject to additional disadvantages beyond those associated with socioeconomic standings and quality of health care.

PAGE 31

CHAPTER 2 LITERATURE LEVIEW This chapter presents a review of the literature that is relevant to this study of the relationships of child mortality and ethnic background. The first section of the chapter provides an overview of the definitions of race and ethnicity in the study of health and child mortality. The characteristics of ethnic relations in TT are reviewed through historical transformation of the islands population. TT has established segments of ethnic divisions on which the basis of national arguments over social, economic, and political allocations are laid. The second section reviews the prior studies of child mortality with emphasis on the significant roles of socioeconomic factors influencing other factors, followed by the vital statistics including child mortality rates in TT. Trinidad and Tobago Defining Differences between Ethnicity and Race A vast literature on health, infant and child mortality, and child survivorship has used either of two terms ethnicity and race, or both terms. It is important here to clarify the terms of race and ethnicity in the study of child mortality. Race is characterized primarily by phenotypic features but has been used to imply genetic or biological bases of health behavior or outcome (King 1997, Gutman 1999). Much research has focused upon studying racial differences between whites and blacks; however, usage of race in studying health has decreased with the recognition that genetic differences are greater among individuals within a given racial group than 19

PAGE 32

20 between racial groups (Michaud et al. 2001). Cooper, an epidemiologist, conceptualized race in the study of public health and medicine; In the biologic sense, there are not such things as races. ---. The appearance of a highly consistent pattern of differential mortality between races can be ascribed only to environmental (i.e., social), not genetic, factors. The concept race itself is a social category. Whether it be [sic] Catholic in Ulster, Jew in Germany, Tamil in Sri Lanka, or blacks in the United States, the definition of a population subgroup is a result of economic and historical, not evolutionary, development. Health status of racial group should be viewed within this context (1984: 722). For Cooper, although the character of the health disadvantage particular to a racial group, e.g., higher risk of coronary heart disease for African-American, may evolve, the disadvantage itself is not likely to diminish until the intensity of racial discrimination is successfully reduced. Ethnicity, on the other hand, is a common set of practices, values, and beliefs held by a collective and transmitted from one generation to the next (Helman 1990, Bhopal 1997). Barth indicated that ethnicity is a form of social organization and a fundamental means of ordering social life; one that relies on manipulating cultural traits and ideas about origin so as to communicate difference. Ethnic definitions are based on ascription and self-ascription-manipulation of identities and their situational character (Barth 1969). For this study, the term race would be inappropriate to distinguish collectives in the contemporary TT society. The island was populated with three distinct immigrant races; Europeans, Africans, and East Indians. Europeans and Africans were differentiated in terms of color; however, no conventional color correspondence is assigned to East Indians. There is no inherent affinity between people sharing a common racial identity; rather racial identities are seen as historical products, which shape social affinities and antipathies, and thereby precipitate various social groupings and boundaries (Segal 1993).

PAGE 33

21 Notably, classification by race was practically impossible and mostly meaningless because of divergent ethnic groups who were brought by the colonial government adopting a variety of immigration schemes in order to import laborers from other Caribbean islands and other countries including Portugal, Syria, Lebanon, China, and India, including former American slaves. TT has not been a racially-stratified society, but rather has exhibited an ethnic-class social structure. Bridget Brereton (1993) shows that Trinidad 1 ethnicity cannot be racially-stratified society because of the following three reasons: First, Trinidad entered into its phase of plantation development relatively later. 2 Consequently, Trinidads experience of plantation slavery was brief (about fifty years). 3 Second, Trinidad entered the post-abolition era with an unusually large middle tier. 4 Third, large-scale immigration, which was a result of labor shortage after the abolition, transformed the three-tier model by introducing new ethnic groups. Ethnic differences within classes were important and each ethnic or class influenced the other in terms of culture and values in creation of identity in the host society (Yelvington 1993). Therefore, ethnic classification has been 1 Concerning the ethnic composition, Tobago society cannot be considered in the same sphere as Trinidad society. Tobagos population is dominated by Africans (92%). In addition, Tobago was a completely separate entity with no administrative links to Trinidad up to 1889 when Tobago forcefully was made a ward of Trinidad by Britain. Its historical experience was quite different from Trinidads. Importantly, Tobago does not have the same history of multi-ethnic immigration as Trinidad. Therefore, in the TT history, Trinidad alone is referred or each island is described respectively until 1889. 2 Trinidad became a significant producer of West Indian export crops in 1784 when Britain ceded St. Lucia and Dominica islands and French planters moved on to Trinidad. Especially it was enlarged after the disturbances in St. Vincent and Grenada in 1795, most British planters moved into Trinidad which was comparatively tranquil and contained large areas of uncultivated, and unoccupied land (Rogozinski 1994). 3 The years between 1784 and 1838 were the period of slavery in Trinidad. This compares to 200 years of the classic slave society such as Jamaica between 1655 and 1838, and Martinique 1635 and 1834. 4 In 1838, 42% of the population in Trinidad belonged to the middle tier, while the middle class in Jamaica was 12% and 32% in Martinique.

PAGE 34

22 appropriate for studying the TT society; most researchers prefer to use ethnicity as an analytical category embracing political, economic and ideological relations. As stated above, race has been recognized as a socially constructed phenomenon like ethnicity, therefore the usage of two terms are dependent upon the social context of each society. Hence, in this study, race and ethnicity is considered the same analytical division as a reflection of socioeconomic, cultural, political, behavioral, and health differences between collectives in a specific society. Ethnic Context in Trinidad and Tobago According to the 1990 census, TT poly-ethnic society includes, besides the two major ethnic groups, the East Indian and African, 19.0% Mixed heritage people and 1.7% other ethnic groups composed of Spanish, French, Portuguese, Syrian, Lebanese, Chinese, Philippines and others. African and East Indian peoples have played important roles in economic, political, and cultural development in TT. The dynamics of the relationships between the African division and the East Indian division has significantly influenced the islands transformation from a colonial society to a multi-cultural and multi-ethnic republic. The former was brought to colonial Trinidad as slaves for working on the new plantations as a result of Cedula (low in Spanish) of Population in 1783 substituting the extinguishing indigenous population. The latter came to Trinidad as indentured servants for, similar to Africans, working on the growing sugar plantation, and for replacing the emancipated Africans. The indenture system was merely a new system of slavery (Tinker 1974). The relationship between Africans and East Indians has been described distinctively by each ethnic group. Each holds persistent negative perceptions of subordination and superordination of the other. This relationship was established at the

PAGE 35

23 arrival of East Indians through two main contexts. The first is the physical and occupational isolation of the East Indians from the Africans. Many East Indians succeeded in becoming peasant proprietors and a result the economic interests of most East Indians shifted from sugar plantation labor into small-scale cultivation on their own land. Therefore, they settled into a rural way of life, which contrasted with the lives of the Africans in the urban areas who have lived mainly within ethnic enclaves. The second is a unique circumstance in which a dividing line was drawn between aristocracies of whites and former slaves of Africans versus East Indians who formed the bottom tier of the society at the arrival. Consequently, this was the foundation that helped TT realize the relative but equal distribution of economic and political power to each ethnic group. The Africans seized political power for over 30 years since TTs independence while the East Indians become economically competitive. Race first provided the basis for communal identity and resistance to colonialism. Both Africans and East Indians maintained aspects of their own cultures, distinguished themselves from their European overlords, and challenged colonialism. Each preserved many aspects of their traditional behaviors, customs, beliefs, and orientations, perhaps for the purpose of conserving their identity, in part, for protection against external pressures such as governmental policies, economic transitions, and offenses and censures from members of the other ethnic culture. Slavery to collective identity Before the arrival of immigrants from India, stereotypes based on race had already emerged. These stereotypes had antecedents in Spanish culture. The initial encounter between any people of diverse cultures and civilizations of immigrants and slaves from various places, naturally gives rise to comparison by self-examination. Biases, prejudices,

PAGE 36

24 and other sentiments emerge from such comparison (Moore 1995). The superiority of the white race became the basis for ideological justification for coloreds servitude. British and European intellectuals developed the idea of racial types as the most important method of classifying people. They thought mankind was divided into permanently different biological types. The doctrine of racial type and social Darwinism helped to create a climate of opinion which was hostile to dark-skinned peoples everywhere, especially when dealing with uncertain newcomers (Brereton 1979). Africans were regarded by the planters as being lazy and irresponsible, having a penchant for drinking and conspicuous consumption, and being prone to profligacy (Brereton 1979). Despite the acceptance by these despised Africans of many European cultural and religious practices, they successfully defended some aspects of their own culture and lifestyle in the face of determined and powerful scorn, and occasional opposition, from the ruling class. However, the Africans themselves would eventually behave somewhat like the white and white Creoles, as they developed scornful stereotypes of those who were to come after them. In 1833 the British government passed the Act of Emancipation, declaring it a law in the following year. 5 In the new society after the emancipation, the system 6 gave every incentive for the ex-slaves to leave the estates and seek independence as a small holder and a part-time wage laborer in the city (Vertovec 1992). 5 Slave-owners throughout the empire were duly compensated, while the slaves themselves were originally obliged to labor as apprentices for an additional six years. The requirement of apprenticeship was halted in 1838 finally, and over 20,000 slaves of African descent were freed in Trinidad (Williams 1962, Brereton 1981). 6 After slavery was abolished in 1834, planters tried to maintain labor in their sugarcane fields offering a rate of wages far higher than in the other British West Indies, as well as rent-free huts. These wages for field labor between 50 and 65 cents per task or per day were higher than any paid in Trinidad for a century to come in 1938, unskilled labors in sugar were earning 35 cents per day (Brereton 1981). Despite the high wages, freed Africans were reluctant to settle far in the interior at a distance from existing centers of population with their schools, churches and rudimentary social amenities and Africans themselves pursued

PAGE 37

25 When the large indentured labor population from India arrived, those ethnic groups already living in Trinidad took care to distinguish themselves from these newly arrived East Indians. Brereton points out that there is evidence that a Creole identity shared by local white and educated colored and blacks was emerging in Trinidad, a Creole solidarity in opposition both to the British representatives and to the Asiatic immigrants (Brereton 1979: 208). East Indians were especially singled out in this process of hostile ethnic stereotyping. After all, they looked different, dressed and behaved oddly, spoke different languages, ate strange foods, practiced queer customs, and worshipped weird gods. In every sense, they seemed were in striking contrast to Western ways. As the immigrants came from widely different regions of the Indian Subcontinent, the newly created migrant world in Trinidad was characterized by substantial differences in culture 7 and economy. The remarkable heterogeneity of the migrant population and their broad range of language were multiplied by distinct dialects due to smaller sphere where they lived (Vertovec 1992). Initially, the internal heterogeneity of Trinidadian society was not restricted to African/East Indian differences. There were strong internal differences within the East Indian group itself. Some of these differences were religions, distinguishing Hindus from non-Hindus. But even among the Hindus themselves, who were 85% of the total immigrants from India, there were regional and cultural differences such variables as languages and caste systems. These varied backgrounds contributed to the demise of a significant portion of caste foundation and caste-based ideology. a legal or nominal freedom. A large number of Africans entered the skilled trades and moved to Port of Spain, San Fernando, and larger villages. These are now considered African dominated areas. 7 Including language and dialect, dress, cuisine, caste composition and structure, architecture and village layout (Vertovec 1996).

PAGE 38

26 Therefore East Indians had not simply conserved pre-immigration cultural forms but have created a series of syncretic or other modified cultural forms. Regarding the East Indians position within Trinidads social, economic, and political structure, new types of relationships among East Indians and with other ethnic groups have periodically worked to produce changes in the conservation of East Indian culture. As the strong internal differences among immigrants from India lessened, simultaneously, the strength of the African/Indian differences became more evident. These subjective ethnic stereotypes had a solid grounding in an objective emergent division of labor: whites as plantation owners; Chinese and Portuguese in trading occupations; Africans and coloreds moving into the professions and skilled manual occupations; and East-Indians were almost completely in agricultural occupations. Because of these occupational differences, the two largest groups, Africans and East Indians, were separated geographically as well as culturally. Many Africans have been urban-based in two cities; Port-of-Spain and San Fernando, while East-Indians have lived in the rural central and southern parts of the island with a strong core found in the plains of the sugar belt. Therefore, a pervasive and fundamental physical, geographical separation characterizes Trinidadian society, as Premdas describes as the Creole-cum-colored portion versus the Indian portion (1993: 100). Hence, social confrontation has involved the indigenization process of several migrant groups, divided first by race then by ethnicity, language and religion, plantations, small-holdings, villages and growing towns. It has been characterized by the infighting of these groups, both within each group and against other groups. The ethnic diversity may have encouraged East Indians to seek common ground. Interestingly, this situation has

PAGE 39

27 not been influencing only the East Indian community but also the African community. This situation is vividly remarked by an Afro-Trinidadian friend of mine. If we wouldnt have Indo-Trini, we wouldnt be the Afro-Trini. If we were not here, they were not East Indian, we may be complementary to each other Ethnicity and class consciousness Trinidads situation as a colony under foreign control changed when partial self-government was instituted in 1925. The first political organizations in TT developed in the 1930s, when a worldwide economic depression spurred the formation of labor movements. Full adult suffrage was introduced in 1946. In 1956 the Peoples National Movement (PNM) was formed by Dr. Eric Williams, who became the first Prime Minister in 1958, drawing on the support of mainly African elements of the population. The opposition 8 drew support from the East Indians. 9 The PNM continued to win elections. By the 1970s, the islands industrial structure had shifted from an agriculture-based economy to an oil industry-based economy 10 that produced revenues for the government. It has been said that revenues from oil exports were used to assist African population subgroups who were identified as underprivileged. Members of the East Indian group perceived that most of the funds went to poor Africans in urban areas. East 8 The Peoples Democratic Party (PDP) was established in 1953 by Bhadase Maraj supported by the rural Hindus. 9 In 1960, the composition of the population was African 43.3% and East Indian 36.5%. An East Indian majority was first noted in the 1990 Census. 10 The chief domestic beneficiary of oil income is the central government, which receives oil revenue through taxes, royalties, and ownership. (In Trinidad and Tobago 28% of the industry was in the governments hands in 1996). The resulting expansion of the public sector crowds out the private sector. Then, many of the public sector commitments made during the boom were difficult to reverse and so caused delays in adjustment when the boom ended. In Trinidad, the public sector accounted for 30% of GDP, (compared to the other resource-supplying nations such as Chile 8%, and Argentina 18%, the public sector is unfavorably large), 30% of total employment, and over 50% of salaried employees. (World Bank 1996).

PAGE 40

28 Indians further concluded that urban Africans benefited more from job opportunities provided by the government. 11 In this way, government policies contributed to inter-ethnic tensions. The ethnic competitions, which began with the importation of forced and indentured labor early in Trinidads history, were kept in force when Williams provided special attention and assistance for Africans, who he considered to have suffered past discrimination. This process, termed a symmetrical political patronage, led to feelings of alienation on the part of other ethnic groups (Center for Ethnic Studies 1993). The traditional view differences in occupations persist today, based as it is in certain objective facts. Rural-based East Indians have the lowest income. People of African and mixed heritage have reached the mid-point of the income distribution while Whites and Off-Whites 12 have the highest income (Harewood and Henry 1985). However, the actual situation is more complex than these general statistics suggest. For example, government workers mainly earn more than those in private enterprise, 13 a fact that favors Africans. On the other hand a majority of millionaires are Syrian or East Indian. In addition, the Center for Ethnic Studies reported that there is no sufficient 11 The PNM maintained a patronage network targeted at Africans, especially urban ones. One method was the establishment of the government's Development and Environment Works Division (DEWD), which employed workers for road construction and maintenance. Almost every DEWD project was aimed at African areas and hired African workers (Yelvington 1995). Percy C. Hintzen (1989) and Steven Vertovec (1992) stated that patronage was accomplished most effectively through the state sector, and the PNM's industrial strategy was aimed at urban Africans, at the expense of agriculture, the livelihood of many rural East Indians. With the end of the oil boom, the oil money ran dry and the subsidies were removed. Ironically, this sudden recession made Africans suffer severely because the majority work for the public sector. 12 Off-White is applied to immigrant groups, essentially, who are perceived as very close to White in skin color, but are seen as less powerful politically. This category includes Portuguese, Syrian/Lebanese. Chinese are treated as such by other groups. 13 An average monthly wage in the public sector is higher (TT$ 2,300) than that of in the formal private sector (TT$1,500) (World Bank 1995).

PAGE 41

29 evidence for the African divisions heavy dominance and underrepresentation of East Indians in the public sector (1993). Race was identified as a factor influencing promotion in some of the public companies surveyed however, racial discrimination was in fact a tendency towards speculation that glided easily into charges and counter-charges of discrimination (Center for Ethnic Studies 1993). Given such disparate facts, it is not easy to determine which group is actually better off. Objective complexity notwithstanding, the subjective perceptions remain among the people. The East Indians believe the Africans benefit most from the government, while the Africans think the East Indians discriminate against them in the formal private sector. Selwyn Ryan indicates that perceptions of economic status among the ethnic groups tend to be viewed from an individualistic point of view (1991). He explains, all groups (with the possible exception of the Syrians) believe that they are economically dispossessed. The latter however also believe that they are dispossessed in the sense that they have not been given the social recognition they deserved and that they are still the butts of ridicule (Selwyn Ryan 1991: 78). Both groups are concerned about the economic gains made by the other group, creating in effect a zero-sum game them. Conflicts between the two groups over education, which is one of the most important avenues of upward mobility in a developing ethnically heterogeneous society is an example of the compound product of the cultural confrontations. Religion, especially Christian missionary, which represented the western culture, played a significant role in establishing educational facilities in TT. Religious conversion was practiced among some East Indians in order to gain material benefits, namely, education. When East Indians entered Trinidadian society, they were considered by the host society a lower class and

PAGE 42

30 minority population. This stereotyping by religious denominations led to the creation of separate school facilities and to the subsequent PNM policy dealing with the location of schools, the language of instruction, religious orientation, the admission of students, awarding of scholarships, and treatment of teachers that led to the establishment of the East Indians denominational schools. These dynamics injected an element of ethnic exclusiveness into the educational sector (Gosine 1986). 14 The feeling of East Indians that East Indians had of being educationally disadvantage continued under the PNM administration. Many East Indians believed that they would not receive a fair share of educational benefits especially in terms of the awarding of scholarships and the hiring and promotion of teachers. Considering the geographical advantage of Africans, namely, their presence in urban areas and consequently their greater proximity to the majority of schools, the educational disadvantages of the East Indians might not derive from any deliberate action on the part of the government. Did the PNM intentionally favor its own ethnic group and slight the interests of the East Indians within the educational sector? This question is hard to answer empirically and objectively. However, the subjective perceptions of intentional educational discrimination against East Indians is obvious and simple common sense to the people of Trinidad, especially to members of the East Indian group. The empirical evidence indicates that by 1980s, Africans and East Indians had leveled in terms of group income (Yelvington 1995). Since 1996, political power has 14 East Indians could not receive a fair share of educational benefits under the PNM administration. The educational institutions, by and large, are located in such areas where they best suit the convenience of black students in areas inhabited predominantly by that race. The participation of East Indian students in higher education causes African students to regard them as socio-political threats. Similarly, East Indian students feel threatened by the gains of the African students whom they see as the recipients of government support (Gosine 1986). On the other hand, some recognize that the PNMs education strategy contributes to both ethnic groups (Selwyn Ryan 1991).

PAGE 43

31 oscillated between PMN and the United National Congress, supported by East Indians. It may be true that this closeness and their juxtaposition could cause them to be competitive with each other over allocation of resources, and through a class-consciousness that is, each group thinks it is superior. Properties of Child Mortality Difference between Infant Mortality and Child Mortality In analyzing mortality in the postnatal period, the age at which mortality is measured calls for an important consideration because of the social and biological factors that affect mortality vary by the age of child (Wood and Lovell 1990). Practically, analyses concerned with the causal pathways of postnatal deaths are hitherto divided into two groups; infant mortality (under 1 year) and child mortality (1-5 years). This dichotomization is mainly distinguished by the terms of cause of deaths, and two sets of causes are designed as endogenous and exogenous. Generally, the former class of death is presumed to arise from the genetic makeup of the infant, the circumstances of prenatal life, and the conditions of labor, which are difficult to prevent or treat in the present state of knowledge. The latter class is presumed to arise from purely environmental or external causes, i.e., it is related to the contact of the infant with the external world. Exogenous mortality primarily includes infections and postnatal accidents, which are relatively preventable or treatable (Shryock et al. 1976). Considering the statistical facts that the highest risk for infants is under one month of age over 95 percent of infant deaths (Shryock et al. 1976), causes of infant deaths are mainly considered endogenous, or at least the proportion of the endogenous causes is larger than that of exogenous causes for the infant deaths. On the other hand, child mortality is determined by the combined effect of both endogenous and exogenous factors, and it is

PAGE 44

32 also more sensitive to a broad range of environmental conditions (Wood and Lovell 1990). Related to the age issues stated above, there is another dichotomization in a wide range of demographic research. Prior research has indicated that the nature of child mortality in regard to emphasizing the importance of sociological research in illustrating how various factors causing child mortality shape the risk of child death. Therefore, research concerned using the causal pathways of child mortality can be dichotomized. One view focuses on the direct impact of social and economic environments of the risk of child mortality and the other view focuses on the impact of variations in health services. The relationship between socioeconomic environment and child mortality is captured in a remark by Wagner, a neonatologist and perinatal epidemiologist, quoted by Gortmaker and Wise (1997: 156); Infant mortality is not a health problem. Infant mortality is a social problem with health consequences. It is analogues to traffic accident mortality in children: the first priority for improving traffic accident mortality in children is not to build more and better medical facilities, but rather to change traffic laws and better educated drivers and children. In other words, the solution is not primarily medical, but environmental, social and educational. The same is true for infant mortality: the first priority is not more obstetricians or pediatricians or hospitals, nor even more prenatal clinics or well-baby clinics, but rather to provide more social, financial and educational support to families with pregnant women and infants (1997: 156). The powerful role of socioeconomic forces in the prediction of disparities in infant mortality is stressed by this perspective. The other view focuses on health services. Control of childhood diseases is typical of health services, which can be provided equally to nations to improve health care technology. But the relations between child mortality and quality of health services cannot inclusively be studied without the elevation of social pathways. Especially, health care technology or westernizing health care may widen the disparities between haves

PAGE 45

33 and have-nots. This account is expressed by Gortmaker and Wise indicating that the trends of decline of child mortality are mainly due to innovations in health services; such technological change also creates new opportunities for socioeconomic differentiation as life-saving therapies or preventive interventions potentially are made available only to the economically advantaged (1997: 148). Combining the two categorizations suggested above, the research spheres for infant mortality and child mortality seem to have been closer. As researchers have become more interested in the linkage between social inequality and mortality outcomes, they begin to emphasize the diverse mechanisms through which this relationship is manifested, as well as how various mortality influences vary in their causal priority and proximity to the biological event of death (Eberstein 1989). Postnatal mortality, including both infant and child mortality, represents the cumulative effects of all factors characterized by both endogenous and exogenous factors, and by pathways affecting postnatal mortality, socioeconomic circumstances, and health care accessibility. Moreover, child mortality may be a more intricate composite of a number of component rates, each with its own set of relationships with social factors, health services and individual mothers health orientations. This is why demographers and sociologists have been interested in using child mortality to explain the social inequality and social stratification within a society, and the primary reason for this study, which uses child mortality as a general term including both infant mortality and child mortality. Determinants of Child Mortality Demographers and public health scientists have developed and shared the common perception of the contribution of socioeconomic development and the medical and primary health services offered by public health programs in the reduction of mortality

PAGE 46

34 (Preston 1975, Caldwell 1979). Differentials in child mortality between population groups have been a constant topic to social scientists. The most common variables used in the study of child mortality differences between sub-groups within a nation are socioeconomic standings as the main effect or as surrogates for other variables about which information was not directly available (Hobcraft et al. 1984). The primary reason is because the extreme dependence of children especially under 5 years old makes child mortality a sensitive measure of the quality of life. As Gortmaker (1979: 281) explained, infants exercise no responsibility for their environment and health status, and thus an infants own motivations and actions have little impact upon its chances for survival; most influences should come from its parents and the surrounding environment. Hence, child mortality has been considered as a mirror of the quality of given circumstances for each child, and the child mortality differences among the people have a significant role in measuring the disparities between population groups within a nation. Socioeconomic and Demographic Variables on Child Mortality Income Income is an influential factor in consideration of individual socioeconomic status and an important determinant of child mortality (Gortmaker 1979, Mosley and Chen 1984, Hummer 1993). Higher incidence of poor pregnancy outcome and child mortality were found among women from disadvantaged socioeconomic backgrounds. The mere existence of differentials by household income however, does not constitute a fully satisfactory explanation of the disparities of child mortality in income levels as well as educational levels. There are a variety of proximate and intervening factors related to infant mortality that are also associated with income and education and thus, need to be

PAGE 47

35 controlled. Such factors fall within a context that rationalizes tests of relationships between socioeconomic status and child survivorship. Maternal education Mothers socioeconomic standings are represented by several factors such as education, household income, work situation, occupation, and quality of housing. Maternal education as a socioeconomic indicator appears most frequently in studies of child mortality, because other measures that might be preferable, such as family income, are not available in the vital statistics records that constitute the basis of most research in this area (Hummer et al. 1999: 1087). Also, maternal education has drawn a wide range of research interest because it is the most relevant and intuitive from the standpoint of child health policy relating to education-conducted use of health services. Although the results analyzing the connection between education and child mortality vary in strength of impact on child mortality outcome (Benyoussef and Wessen 1974, Caldwell et al. 1983, Hobcraft et al. 1984, Ce Chen and Williams 1997), the most valuable nature of education as well as household income are that they appear to be the most common variables tapping into not only the cohort of socioeconomic factors such as work situation, occupation, and quality of housing but also into almost all explanatory variables of child mortality. If the magnitude of education as a socioeconomic variable on child mortality varies in each society or each cohort within a society, the covariates could shape and state the important variance explaining each societys characteristics. Marital status and residential characteristics The influence of demographic characteristics such as residential (urban-rural) and marital union characteristics on child mortality are often examined with socioeconomic status. Mothers place of residence (urban-rural distinction) is used as a proxy measure

PAGE 48

36 for living conditions to illustrate both public and medical health provisions. This is due to the lack of infrastructure such as electricity (especially for refrigerator), drinking water, non-drinking water (for flush toilet), and sewage, and access to basic health care facilities which may be life threatening to the children in rural areas (Suwal 2001). Urban residents were found to have better conditions compared to rural residents. This is especially likely given the confounding of many other socioeconomic attributes with place of residence. However, in some cases, the residential differences appear to show internal differences. Compared with the homogeneity of experience for urban residents, mortality in the traditional rural areas varies widely even between sub-groups with similar attributes (Hobcraft et al. 1984). The association between marital status and child mortality is examined by using various controls such as education (Keller 1978), race (Cramer 1987, Eberstein 1989), race and age (Gee et al. 1976), and race and intervening factors (Eberstein et al. 1990, Hummer et al. 1999). The magnitude of marital status on child mortality differs, depending on the covariates, but these studies found that significant interactions, between marital status and education, and race and unmarried status, are associated with higher child mortality. However, Cramer indicates that; marital status, similar to residential and age, may not be an independent risk factor. ---. In general, it is not known which social factor or combination of factors is causally responsible for the observed group differences (1987: 299). For example, children born to unmarried women may be at higher risk for mortality as a result of inadequate familial resources rather than marital status per se (Eberstein et al 1990). Therefore, marital status is considered to be substitutive to the level of quality of life of mothers. We may, however, have to be

PAGE 49

37 careful that marital context varies in each society. The demographic factors as such mentioned above are generally attributive covariates. Intervening Health Care Variables in Relation to Socioeconomic Variables The association between child mortality and maternal education has been the most common finding in the child mortality literature as mentioned above. Several lines of inquiry on educations direct and indirect function on child survivorship have been determined. There are two major directions of study in the association between education and child mortality; the first linking to economic status, and the second linking to skills and health care practices. The former is more likely conditional and influenced by her familial construction. The economic status of married women is largely determined by her husbands educational achievement and his occupation and employment status. Paternal education captures variation in household wealth or disposable income, and the relation between child mortality and education is accounted for by the various adopted indices of household economic condition (Hobcraft et al. 1984). Fathers education also influences attitudes, preference, and choice of consumption goods, including childcare services. Therefore, in many cases correlations between health effects and educational level of fathers (or other non-childbearing, economically productive adult members in a household) largely occur because of operations on the proximate determinants through the income effects (Mosley and Chen 1984: 34). While paternal education has a role in limiting or facilitating the orientation and selection of mothers way of life and mothers characteristics, maternal educational level is more likely linked to skills in health care practices and health orientation directly. Because of biological links between the mother and infant during pregnancy and lactation, mothers health and nutritional status influence the health and survival of the

PAGE 50

38 child (Mosley and Chen 1984). Studies by Benyoussef and Wassen (1974) and Boerma (1990) show that better educated mothers more commonly use maternal and child health services than less educated mothers. Furthermore, Streatfield, et al. (1990) indicate that educated women have greater awareness of correct immunization schedules. This can be another dimension of formal education, which Caldwell has argued, that the significance of maternal educations role is to change traditional patterns of familial influences so that women may improve their understanding of the importance of using modern medical services (Caldwell 1979, Caldwell et al. 1983). Mothers educational level can affect child survival by influencing her choices and increasing her skills in health care practices, such as nutrition, hygiene, preventive health care, and disease treatment. Prenatal care Prenatal care has generally been considered to contribute to good birth outcomes and also as a predictive variable of child health and child mortality. The adequacy of prenatal care as evidenced by prenatal checkups, the presence of trained health professionals, whether doctors, midwives, or traditional birth attendances during delivery, tetanus toxoid immunization and nutritional supplementation, etc., has a direct impact on maternal morbidity and mortality (United Nations Population Fund 2000). Prenatal care is considered as a package of necessary services (Shiono and Behrman 1995). Therefore a number of benefits accrue; such that prenatal visits play an additional role: to enable women to obtain general information on infant and child health care as well as specific medical attention. Several recent studies have examined the relationship between race/ethnicity and child health and mortality while controlling for limited sets of confounding factors, such as the risk of infant birth weight, with household income and education (Kleinman and Kessel 1987, Collins and David 1990). These

PAGE 51

39 studies indicate that the interrelationships among prenatal care, birth weight, and child mortality document the importance of socioeconomic and other social variables on the probabilities of low birth weight and the risk of child death. The relationships between prenatal health care and maternal education have been found to be highly correlated with each other (Cramer 1987, Hogue et al. 1987, Mangold and Powell-Griner 1991). Socioeconomic variables, household income and education, serve as indicators of knowledge, amount of medical services and level of household income. These variables exercise an important effect on the ability to obtain medical provisions. Echeverarra and Frisbie point out the potential of prenatal care practice: an increased utilization of a wider array of postnatal health care services was found among mothers who practice an adequate level of prenatal care (2001). Preventive health care Represented by breastfeeding and timely immunization, postnatal care includes feeding children nutritious solid food and sanitation, and a hygienic way of living is vital in preventing possible postnatal child deaths. Child immunization is one of the health systems principal interventions aimed at lowering child mortality. Preventive health care has been improved dramatically since the Second World War such that relatively rapid child mortality reduction in mid-20th century was primarily attributed to this health technological improvement (Suwal 2001). The evidence, which supports the relationships between accessibility to health services and child mortality, has been provided by Maison and Sekeito (1987) and Chaulagai (1993). However, according to Cleland and van Ginneken (1988), the nature of the interaction between accessibility and utilization of health service influencing the reduction of child mortality is context-sensitive. Its nature depends mainly on the level of development of the health infrastructure. Education and

PAGE 52

40 geographical accessibility are substitutive, but maternal education has more potential to compensate for the disadvantage of mothers lack of accessibility to health services (Rosensweig and Schultz 1982, Kadende 1994, Bicego and Boerma 1996). Education equips mothers with knowledge of healthy living and encourages them to practice proper health care. Research examining the relation between ethnic groups and receipt of preventive services, which are usually in the form of care such as pap smears, breast exams, mammography, and cholesterol screening, found general differences in female preventive health care use among racial and ethnic groups. Furthermore, each ethnic group has a tendency to receive a particular preventive service (Corbie-Smith et al. 2002). For narrowing the differences by race in preventive health care practices, it is necessary to address racial differences in disease outcomes. Simultaneously, we may consider the influence of a certain ethnic groups common tendency to receive health care as one of their child care practices. The more mothers receive preventive health care within an ethnic group, the more they become aware of the importance of having their children receive immunizations appropriately. Breastfeeding Breastfeeding has been emphasized because of its significant influence on the well-being of children. Childrens health advantages are conferred by breastfeeding and, conversely, there are detrimental effects of failure to breastfeed on the child deaths. A large body of evidence indicates that children who are bottle-fed from birth run a higher risk of health and developmental problems than do breast-fed children (Goldberg et al. 1984). The relationship between breastfeeding and child mortality has been examined mainly with mothers socioeconomic characters, education and income which are

PAGE 53

41 considered as having strong predictive power, whether or not a mother breastfed. Mothers with a higher formal education and a higher monthly income are more likely to breastfeed (Goldberg et al. 1984, Huffman 1984, Trussell et al. 1992, Humphreys et al. 1998, Alan Ryan 1998, Forste 2001). Marital status is also strongly associated with breastfeeding; mothers being married are more likely to breastfeed than mothers having other union status (Hirschman 1981, Forste 2001) which indicates that the support of the childs father is important in the breastfeeding decision. Residential differences are also predictive of breastfeeding. Usually breastfeeding is considered to affect child mortality most strongly in the cities (Trussell et al. 1992). Some findings indicate that urban settings are negatively associated with mothers breastfeeding decisions, due to general perceptions toward bottle-feeding, which is considered as a modern, adequate, and convenient method. Additionally, for mothers living in urban area, it is rare to have role models of breastfeeding practices (Huffman 1984) suggesting urban-rural cultural differences. Provided that living conditions in urban areas have been improved such as access to safe drinking water, medical facility, and health information, the impact of breastfeeding on child mortality has been reduced in cities. Instead, Goldberg, et al. found that accessibility in rural areas have been left as it is or is still behind the urban areas, therefore, association between breastfeeding and child mortality becomes stronger and excess urban areas; a higher mortality risk is experienced by the non-breastfeeding division in rural areas (1984). Not only is the breastfeeding variable considered an indicator of quality of childcare, but also it may explain individual mothers perception of time, which is strongly influenced by their daily life orientation and its tempo and general health

PAGE 54

42 attitudes and beliefs of the mother and those of her social network (St Clair PA et al. 1989, Butz et al. 1993). Mosley and Chen (1984) mention the importance of a mothers time caring for her children and maintaining their living circumstances such as prenatal visits, breastfeeding, food preparation, sanitation, and sickness care. Childcare time often competes with time needed for income-generating work. They note that, A mothers time may also be required for other economically productive activities that may or may not be related to child health (Mosley and Chen 1984: 35). The physical accessibility to (modern) health services is also largely determined by mothers socioeconomic standing; thus, issues of accessibility to health institution and services extend to the variability within education. Greater physical access to health services improves survival to a greater extent among the children of less educated women than for children of more educated women (Katende 1994, Bicego and Boerma 1996). Type of place where the child is born Place of delivery; whether or not children were born in a medical facility or private home that may be unsafe and unhygienic, have been shown to influence child survivorship. Whether or not the baby was delivered in a hospital serves as one extreme indicator of lack of medical care (Gortmaker 1979). For mothers, the high risk of neonatal tetanus deaths was found to be associated with home delivery (Foster 1983), and the births taken place at a medical center and assisted by doctor, nurse, or midwife are found to have a significantly lower risk of child mortality (Suwal 2001). Although traditional birth attendants still deliver a considerable proportion of newborns in developing countries and rural areas including middle developed countries, the proportion has shown a tendency to decline (United Nations Population Fund 2000).

PAGE 55

43 Having provided the persistent interrelations among socioeconomic background, namely, income level and educational achievement, and demographic and health related factors, it is evident that these factors do not independently determine child survival chances. Eberstein et al. state that, there are reasonable a priori theoretical grounds to expect interaction among them that the effects of some of the variables may vary depending on levels of the others (1990: 414). The effective variables and their strength may also vary according to the social system of a given society. For a better understanding of the association between socioeconomic factors and child mortality, recent child mortality literature includes health related factors such as prenatal health care, preventive health care, and breastfeeding which can indicate mothers childcare orientations and cultural influence derived from their race and ethnic background. On the other hand, recent studies in the United States indicate that race is a modern social stratifying agent coinciding with the emergence of slave trade. Hummer stated that such destructive exploitation and reasoning served as fertile ground for continued inequalities in resources, status, power, and health between socially defined group (1993: 533). The phenomena remain at present and there is a general consensus that race and ethnicity, as social stratifying agents, continue to affect child mortality (National Research Council 1989, Hummer 1993, Mullings et al. 2001). As seen in the ethnic context in the TT society, which embodies a number of dimensions that are indicative of socioeconomic status, women who are from disadvantaged socioeconomic background and from an ethnic group which experienced continued lower socioeconomic status in TT should be considered higher child mortality risk group.

PAGE 56

44 Infant and Child Mortality in Trinidad and Tobago Infant and child mortality rates from 1960 to 2000 are presented in Figure 2. In 1990, 21 out of every 1,000 babies died before reaching the first birthday while 24 per 1,000 died before the fifth birthday (UNICEF 2003a). The decline of the infant mortality rate from 61 to 21 deaths per 1,000 births between 1960 and 1990 represents a 65.6% drop. An even greater decline of 72.6% is seen for under five mortality, which decreased from 73 to 20. UNICEF provides the latest under five mortality rate of 20 and infant mortality rate 17. 15 These figures represent a very low level of mortality, approaching that 493521161710286624336307357402418201531238454433661196019701980199019952000 Infant Mortality TT Under 5 Mortality TT Infant Mortality LA&Caribbean Under 5 Mortality LA&Caribbean Source: UNICEF Statistics. UNICEF End Decade Database Child Mortality. http://www. childinfo.org/cmr/revis/db1.htm, http://www.childinfo.org/cmr/revis/db2.htm. Figure 2. Infant and under 5 mortality in Trinidad and Tobago and Latin American and Caribbean, 1960 to 2000 15 UNICEF calculated infant and under five mortality rates based on an indirect estimation technique, the Brass Method. The data used in the estimation are the mean number of children ever born to five year age groups of women aged 15-49, and the proportion of these children who are dead, also for five year age group of women. Hence, the infant mortality rate indicates the probability of dying before the first birthday. The under five mortality rate is the probability of dying before the fifth birthday.

PAGE 57

45 of developed countries. By comparison, the infant and child mortality rates in TT are relatively lower compared with the other Caribbean and Latin American countries as a whole. A previous study by Heath et al. (1988) using TTDHS indicates the socioeconomic characteristics of child mortality. The results show that both infant and child mortality are lower in rural than in urban areas. This somewhat unexpected finding may reflect the homogeneity of the socioeconomic conditions in the society: there is difficulty in distinguishing urban from rural areas largely due to the developed transportation systems throughout TT. As expected, mortality for children aged 1-4 decreases as the mothers education increases. However, infant mortality appears highest among the best-educated women. It is most likely because the rates for the highest and lowest education groups are based on a small number of births. The ethnic differences in child mortality in TT are gathered from a fertility study in TT by Harewood (1978), a comparative study on child mortality between three Caribbean nations, TT, Guyana, and Jamaica, carried out by Ebanks (1984), and the survey on postnatal practices in Trinidad by Mahabir (1997). According to Harewood who conducted research in 1970, the fertility was higher in the East Indian division compared to African division -the gap between the two groups was radically narrowing over time and the situation is reversed in 1990 census. This is probably due to the natural relationship between total children ever born and child mortality though, at the time of Harewoods research, East Indian women were more likely to have experienced a child loss. Mahabir reports that the infant mortality was higher among East Indian women than Africans with respective proportions of 8% and 5% in 1994. Perhaps this is because of

PAGE 58

46 the higher concentration of the East Indian population in rural areas. Generally, more lower income families are found in rural areas than urban areas and Ebanks findings correspond to the disadvantaged income situation in rural areas indicating higher infant mortality among East Indians in rural areas. Considering the account of Mahabir, the higher infant mortality in the East Indian division may be situational. Relatively lower child mortality in this nation is supported by such indicators as large immunization coverage (DPT, Polio, measles), higher rate of perinatal care use, and higher rate of receipt assistance at delivery with respective proportion of 90% (1995), 97.6% (1987), and 99% (1997) (Pan American Health Organization et al. 1998). However, UNICEF (2003a) reports that only 1.8% of children under the age of four months were exclusively breastfed. This is considerably lower than expected. Even though immunization is free in all health centres (public sector) and the large immunization coverage is reported, UNICEF indicates that the proportion of children who have had all eight recommended vaccines in the first 12 months of life was estimated to be relatively small amounting to 7.4% (2003a). Mothers levels of formal education differentiate the rates of immunization coverage for their children, that is between mothers who had secondary or higher education and those who had only primary education with respective proportions of 18% and 7.1%. There are both public and private health care systems in TT. Health care is provided privately for varying, but generally high rates of payment. Free health care is provided by various state-owned and controlled public health facilities. Public facilities include general hospitals, located in the main urban areas, and health centers, located in all eight counties of the twin-island state. There are 101 health centres in total, as of 1996

PAGE 59

47 (Mahabir 1997), under the control of the Ministry of Health, which plays a key role in the delivery of health care to the nation. In this relatively small country, with good internal transportation, many of the same doctors at the public facilities are involved in private practice: many doctors work at public hospitals and operate their own clinics. Health care delivery at state-administered institutions has been a contentious issue among members of the public and among the providers themselves. The general population has vociferously expressed numerous complaints condemning the quality of service offered at public health facilities (Mustapha and Singh 2000). There is a perception that the services of medical practitioners in a private practice are more efficiently delivered than those offered at public health facilities which are characterized by the usual bureaucracy and inefficiency that accompany state enterprises (Rathwell and Phillips 1986). Hence one may be tempted to believe that the quality of service received is linked to patients ability to pay. Hypotheses The analysis of TTDHS is organized in order to empirically test hypotheses that are derived from theoretical perspectives of Colemans social system and based on the literature review. Social system theory concerns the balance and dynamics of interests and control over scarce resources between groups contributing to constructing the social structure. Colemans theory is informative because it gives people a way of viewing what has occurred over socioeconomic and political allocations between the two major ethnic groups in TT. Although his social system theory recognizes the interaction among an individuals purposive actions, social networks, and social capital, ethnicity or race is not considered as a pervasive criterion in a society. In the ethnic context in TT, ethnic identity can be considered as a major motive and value to seek the common social and

PAGE 60

48 economic benefits through which we can locate the status of individuals and their groups in the social structure. The following hypotheses are based upon social system theory: H1: Socioeconomic factors have the strongest and most persistent association with child mortality among all explanatory variables. H2: Ethnic identity will show an association with child mortality, but after controlling for socioeconomic, demographic, and health care factors, ethnicity cannot maintain its influence on child mortality and its statistical significance. H3: Individuals economic levels affect childrens and mothers quality of health; therefore, better health status and favorable maternity care reduce the risk of child mortality.

PAGE 61

CHAPTER 3 RESEARCH DESIGN AND METHODS The previous two chapters presented the significance of the study of the relationships between child mortality and ethnicity in TT. Many researchers have found that there are economic, political, and educational inequalities in TT society. However not only are the discourses conflictive but frequently perceptions held by each ethnic group toward others cause conflictive claims about inequality in allocation over social resources. Race/ethnicity, socioeconomic factors, and prenatal/postnatal care including breastfeeding decision have been found to have a significant impact on child survivorship, but they do not independently affect child mortality. In addition, the effects of variables may vary depending on the other variables and on the social system of a given society. The purpose of this chapter is to describe the TTDHS that is used in the next chapter to see if ethnic background plays a significant role in determining the probability of child mortality in TT society and to understand the level of influence of factors on child mortality to describe the characteristics of the social stratification system in this nation. Data and Sample Size Analyses The TTDHS, a national-level sample survey, was conducted by the Family Planning Association of Trinidad and Tobago in 1987. The sampling frame for the TTDHS was based on the 1980 Population and Housing Census, one of the Continuous Sample Surveys of Population used by the Central Statistical Office of the Republic of 49

PAGE 62

50 Table 3-1. Distribution of women 15 to 49 by ethnic and type of place of residence in 1987 and 1990, Trinidad and Tobago Characteristics 1990 Census 1987 Census Ethnicity African 39.6% 35.3% East Indian 42.9% 47.0% Mixed 15.4% 17.1% Others 2.1% 0.7% Type of Place of Residence Urban 48.7% 44.4% Rural 51.3% 55.6% Source: Population and Housing Census 1990, Central Statistical Office Demographic and Health Survey, Trinidad and Tobago, 1987 (Individual dataset). Sample population = 3,807 Trinidad and Tobago. The TTDHS is primarily concerned with family planning, maternal and child health and child survival. Information on household composition and housing quality is also included. Additionally, there are data about ethnicity, religion, and education, as well as economic indicators such as the presence or absence of consumer goods. The TTDHS has three separate sets of data based on household information, individual information, and child information. This study utilizes only one set, the TTDHS individual dataset that contains a total of 3,807 cases (individual women). The data is composed of women between the ages of 15 and 49. Table 3-1 shows the characteristics summary of the TTDHS along with corresponding figures from the 1990 Census. The original data is composed of 1,342 African women (35.3%), 1,787 East Indian women (47.0%), 649 Mixed women (17.1%), and 27 women (0.5%) claimed some other ethnicity. Comparing to the census figures, African women make up a smaller portion of the sample and East Indian women represent a larger portion of the sample. The report of TTDHS presented by Family Planning Association of Trinidad and Tobago

PAGE 63

51 partly explains the reason for the difference in the ethnic composition; it could be due to an unintended over-sampling in areas where the East Indian population is heavily concentrated and a higher response rate among this group. Since the majority of East Indians have resided in the rural areas, a less urbanized sample than the actual Census population may be a reflection of those differences in ethnic distribution of the TTDHS from the actual Census numbers. This study investigates three main questions: (1) Whether or not ethnic background influences child survivorship, (2) whether or not quality of life and education influence child survivorship, and (3) Whether or not child health care practices impacts on child survivorship. Of particular concern in this study is how the ethnicity influences the relationship between child mortality and its predictors. The data is first restricted to only women who belong to the two ethnic groups, African descent and East Indian descent. The second restriction includes only those women who had at least one child born. To assess the overall well-being of women in TT, it is pertinent to restrict the sample population to a narrow age gap as possible to avoid economic gaps over time; time lags between the time of childbirths and the time of survey, and faltering ethnic differences among the sample population. Hence the third restriction includes only women who had child(ren) within 10 years prior to the survey being conducted. The analysis of child mortality differences between the two major ethnic groups deals with various factors, such as demographic standings, quality of life created by variables of household composition and housing quality, and mothers child health care. Lastly, women who did not answer any questions used in the analysis were excluded from the analysis. These procedures reduce the sample size to 1,082 from 3,807. Consequently, the sample

PAGE 64

52 population used for the analysis in this study is composed of 584 East Indian women (54.3%) and 492 African women (45.7%). This represents 34.4% of all the respondents who belong to the two ethnic groups in the TTDHS. Measures There are three clusters of predictors of child mortality: socioeconomic, demographic, and health related. The first cluster includes demographic factors such as ethnicity, place of residence, and marital status. The second cluster includes socioeconomic factors such as maternal educational attainment and quality of life. No data was collected in the TTDHS (it is common to all other DHS-I surveys) on household income per se, however, data intended to capture variations in household wealth and disposable income were collected and they are useful for creating an indicator of economic status which can be called quality of life. The third cluster includes health related factors such as prenatal care, type of place child born, quality of preventive health care, and breastfeeding. These variables are useful indicators of whether or not the respondent has access to a higher quality of health services and whether or not the respondent provides appropriate child health care to her children. Differences in health orientation and preference of medical facility, medical doctor, or medicine between the two major ethnic groups have been observed (Mahabir 1997, Mustapha and Singh 2000, Chijiwa 2001). Selections of the type of medical services serve as not only an indicator of the quality of health of women but also a proxy for their cultural and ethnic vestiges. Before proceeding to the descriptions of predictors mentioned above, the procedures of creating the respondent variable; child mortality, are presented for first then variable definitions and constructions are provided.

PAGE 65

53 Child Mortality The analysis for this study includes women who had at least one child born within 10 years prior to the survey being conducted. Therefore, child mortality of a woman is determined in regards to her children who were born between 1978 and 1987. Information on the child survival status of each woman at the time of the survey is drawn from a variable whether or not the child is alive for computing child mortality in TT. Each woman was asked about all her children she had ever had from the youngest to the oldest. They were asked whether or not the youngest child is alive, whether or not the second youngest child is alive, whether or not the third youngest child is alive, and so forth. Additionally, women who had at least one child within ten years provided answers on whether or not their children were alive up to the 11th child. In constructing the variable of child mortality, children who had died prior to the survey are coded as 1, and children who are alive when the survey was conducted are coded as 0, and then each womans answers of all her children are combined to create the variable; how many children the woman lost within last ten years. The analysis reveals that 1,008 women out of 1,082 have never lost a child, 65 women have lost one child, five women have lost two children, and four women declared having lost from three to six children. Since child mortality distribution is heavily right-censored, the multivariate analysis uses logistic regression techniques. The newly created variable of child mortality, which is the respondent variable in the logistic regression analysis, is dichotomized: 1 = women (74 women) who have lost at least one child and 0 = women (1,008 women) who have never experienced a loss of their child.

PAGE 66

Table 3-2. Variable descriptions Proportion/Mean(Frequency)Childhood Mortalit y Categorical0Have never had a loss of child93.2% (1,008)1Have lost at least one child6.8% (74)Demographic FactorsEthnicit y Categorical0 (reference)East Indian54.5% (590)1 (dummy)African45.5% (492)Type of Place of ResidenceCategorical0 (reference)Urban40.8% (441)1 (dummy)Rural59.2% (641)Marital StatusCategorical0 (reference)Married 58.4% (632)1 (dummy)Separated / Divorced / Widowed6.4% (69)1 (dummy)Cohabiting / Visiting Relations35.2% (381)Socio-Economic FactorsYears of Education Discrete0-16-7.70Quality of LifeScale0.00-6.65-1.8656Health Related FactorsQuality of Preventive Heath Care Histor y Scale0-1.00-0.7316Prenatal Care Histor y Categorical0 (reference)Adequate Prenatal Care83.6% (905)1 (dummy)Inadequate Prenatal Care16.4% (177)Privatized Health Care Histor y Categorical0 (reference)Privatized Health Care Onl y 8.9% (96)1 (dummy)Public Hospital or Both91.1% (986)Breastfeeding Histor y Categorical0 (reference)Adequate Breastfeeding80.7% (873)1 (dummy)Inadequate Breastfeeding19.3% (209)Source: Demographic and Health Survey, Trinidad and Tobago, 1987 (Individual dataset)Sample Population = 1,082TypeValue Value LabelVariable 54

PAGE 67

55 Demographic Factors Ethnicity The original TTDHS contains four categories for ethnic background: African, Indian, Mixed, and Other as we already have seen in the previous section. For the purposes of this study, the ethnicity variable has been restricted to only women who belong to the two ethnic groups of East Indian and African with respective proportions of 54.5% (590) and 45.5% (492). In the multivariate analysis, East Indian coded as 0 (reference) and African coded as 1 (dummy). Proportions for each ethnic group before introducing the restrictions for determining the sample population in this study (excluding women who are mixed and other ethnicity) were 57.1%, East Indian, and 42.9%, African. Type of place of residence The respondents were categorized into two types of residence: urban and rural. Within the sample population, woman who lived in urban areas at the time of the TTDHS survey are 441 (40.8%), and those who lived in rural areas are 641 (59.2 %). Women who lived in urban area are in the reference category coded 0, and those who lived in rural are in the dummy variable coded 1. The successive data from census in TT indicate that urban residents have been better educated than rural residents. These data also tell us the discrepancy in levels of educational achievement between urban areas and rural areas have gradually been narrowed reflecting the improvements in educational systems, which was on going when the TTDHS was conducted. In relations to the role of maternal education in child survivorship and the significant residential differences between East Indians and Africans, urban-rural distinction may influence in determining probability of child mortality in the society of Trinidad and Tobago.

PAGE 68

56 Marital status The TTDHS has five categories for classifying the participants marital status. The actual questions concerning the respondents marital status are stratified: have you ever been married? (yes or no), are you married now? (yes or no), are you living with a common-law partner now? (yes or no), are you having a visiting relationship now? (yes or not). The respondents answers to four questions were accumulated and the participants were categorized into five categories: never married, married, living together, visiting relation, and widowed/divorced/separated. Definitions of living together and visiting relation are vague. In the case of living together, the participant may have a common-law relation with her partner, or the participants partner might have another relationship outside their living place. In the case of visiting relationship, the participant may marry the partner in the future, or she might be the second wife or partner. Since men who are in the prime of life tend to go abroad for a job, the population proportion of women to men is said to be 7 to 1 or more disproportional. These common-law type relationships instead of the legal marriage are commonly accepted and more generalized in TT. Therefore, marital status depends on their self-definition and it varies in how single women perceive their situation. These two categories exist in close relation to each other and it is difficult to distinguish between them clearly. Frequency distribution indicates that there is only one single woman in the sample population. To simplify the analysis the single woman is excluded from the analysis. Consequently, four categories of marital status were collapsed into three categories, married, cohabiting/visiting relationship, and widowed/divorced/separated with respective proportions of 58.4%, 35.2%, and 6.4%. Women who are married are

PAGE 69

57 coded as 0 (reference), and other forms of marital status are dummy variables coded as 1 respectively in the multivariate analysis. Previous investigation indicated that differences in the dominant form of marital union reflect significant cultural distinctions between the two ethnic groups. The considerably low marriage rate among African women compared to East Indian women reflects the predominance of other forms of unions, cohabiting relationship and visiting relationship, within African population. Even when we controlled for the variable of religion, ethnic identity has a very clear association with the dominant type of marital union (Harewood 1978, Chijiwa 2001). Therefore it is meaningful to investigate whether married women have lower child mortality; one of the two major ethnic groups, which has a lower probability of being married is subject to the additional disadvantages of higher incidence of child mortality. Socioeconomic Factors Maternal educational attainment The TTDHS asked questions in terms of respondents educational attainment as to what level of education they ever attended (highest educational level) and, within that level (educational level for TT), how many years (education in single years). This study uses maternal education in a single year, which has a range between 0 and 16. The mean years of schooling is 7.70 for the sample population. Quality of life The information about household composition and housing quality was drawn from questions in the TTDHS to construct an index of quality of life. The index summarizes a household component such as consumer durables as well as access water and electricity; therefore, it serves as a proxy measure of level of modernization and level of hygiene

PAGE 70

58 within household. Because, higher maintenance of sanitation may reduce a risk of infectious disease, thus reducing mortality (Wood and Carvalho 1988, Perz 1997), higher level of modernization indicates better quality of life, which may suggest income level. Quality of life is a broad concept that includes variables such as the quality of drinking water, toilet facilities, flooring materials, accessibility to electricity, and ownerships of such things as a VCR, television, automobile, and refrigerator. However, some of these variables may be more valid operational definitions of quality of life than others. To uncover the latent structure (dimension) of a set of variables, to select the factors that are considered adequate to explain the relationships among the observed variables, and to search for a plausibly appropriate value for each variable, the eight economic indicators were inspected using factor analysis. Table 3-3 presents the results of the factor analysis. The eight variables were loaded on two factors that suggest a conclusion: Quality of life (at least as measured by the variables in this study) is not a unitary concept but rather, it is a concept that consists of two different underlying concepts. The numbers presented in the two columns of Table 3-3 represent the size of the correlation between the particular variable and the underlying factor. Hence, the correlation between the variable toilet facility and Factor 1 is .744. This correlation is higher compared to the factor loading for drinking water (.690). Because the factor loadings represent the degree of correlation between the variable and the underlying factor, the loadings can be used to weight each of the variables, and then they are combined into a composite index. The variables toilet facility, drinking water, floor material, VCR, and Car are highly correlated with factor 1. One might interpret this factor as the principal measure of quality of housing (in the sample population) and to

PAGE 71

59 Table 3-3. Rotated component matrix for 8 variables of household composition Loadings Communality Variable Factor 1 Factor 2 (extraction) Toilet Facility .744 .242 .611 Drinking Water .690 .309 .572 Floor Material .619 -.002 .384 VCR .616 .168 .408 Car .587 .115 .358 Refrigerator .222 .838 .751 Electricity .138 .825 .699 TV .143 .778 .626 Variance (%)* 39.61 15.50 Source: Demography and Health Survey, Trinidad and Tobago, 1987 (child dataset) Note: Extraction Method: Principal Component Analysis Rotation Method: Varimax with Kaiser Normalization KMO and Bartllett's Test: sampling adequacy = .801 Rotation sums of squared loadings construct a composite index for, namely, luxury items. Factor 2, which is highly correlated with refrigerator, electricity, and TV, is interpreted as a measure such as consumer durables. In this study, however, the result of factor analysis is regarded as a means of determining which variables can be adequate for creating an index of quality of life. 16 Considering the moderate communality (the proportion of variance explained by factors) for each variable, I decided to leave all eight variables in the analysis. The weight in Factor 1 for each variable has a positive direction indicating better quality of household composition and housing quality; in addition, Factor 1 accounts for 39.61% of the 16 First of all, factor analysis was conducted with 11 variables; in addition to 8 variables, has stove, has radio, any family member has house/apartment, were included, and this analysis extracted three factors. Since these three variables negatively correlated with other variables in the analysis, they were excluded from the following analysis.

PAGE 72

60 variability of the original eight variables, which has greater variability than Factor 2 (15.50% of the variability). Hence, the weights in Factor 1 for all eight variables are combined into one index quality of life instead of combining them into each index 17 The new variable has a range between 0 and 6.65, an interval of 92. The mean score of quality of life for the sample population is 1.8656. Health Related Factors Each woman provided health related information on prenatal care, type of place child born, use of immunizations, and breastfeeding about her children from the youngest to the 5th youngest (e.g., whether or not the respondent had prenatal care when she was pregnant the youngest child, whether or not had prenatal care when she was pregnant the second youngest child, and so on.) Hence, all respondents have five variables for each question. Each variable are dichotomized into 1 and 0. Since the number of children is different among mothers, five dichotomous variables of a certain question are combined and averaged out for creating a variable. For instance, in the case of prenatal care, if a woman had two children and had prenatal care when she was pregnant the youngest child (coded 1), but she did not receive prenatal care for her second youngest child (coded 0), and then her prenatal care history is 0.5. These newly created variables are considered as child health care histories of mothers. 17 I once constructed two variables for each factor; Factor 1 = luxury items and Factor 2 = consumer durables. However, the means test for luxury life by ethnicity indicated that East Indians are higher in quality of life on luxury items than Africans (sig.=.031), and the means test for consumer durables by ethnicity showed that Africans are slightly higher than East Indians with no statistical significance. To simplify the analyses and to enhance the magnitude of quality of life in addition to the reason which all variables have the same positive directions in Factor 1, I decided to combine all variables loadings for creating the index of Quality of Life.

PAGE 73

61 Prenatal care The actual question for this variable was when you were pregnant did you see anyone for a check on this pregnancy? and answer categories were no one, doctor, trained nurse, trained midwife, traditional birth attendant, and other. The frequency distribution revealed no use of a trained midwife, traditional birth attendant, or other. I conferred with some doctors whom I met during my research in TT. They confirmed that generally there is no difference in medical care for pregnant women between being taken care of by doctors and by trained nurses in the case of normal pregnancy. However, in the case of unusual pregnancy or emergency, the probability of the incidence of miscarriage (Fetal deaths are not included in this study) and the risk of postnatal health and child health could be higher with no doctor in attendance. Hence three values for prenatal care are collapsed into two categories: received prenatal care from medical doctors (1) and did not receive prenatal care from medical doctors (0). After combining the variables of five children, 134 women have never received prenatal care from doctors (12.4%), 905 women received prenatal care from only doctors whenever they where pregnant their children (83.6%), and 43 women have received prenatal care from doctors or trained nurses (4.0%). The newly constructed variable, which indicates history of adequacy for prenatal care is collapsed into two categories. Women who received prenatal care from only doctors are coded as 0 (reference), namely adequate prenatal care, and those who have received prenatal care from doctors but not always are combined and coded as 1 (dummy), namely inadequate prenatal care. Type of place where the child is born The variable of the type of place where the respondents children were born is used to measure the respondents preference for privatized health care. The respondents were

PAGE 74

62 asked, in what type of place was your child born? for which respondents answer fell into four categories: government hospital, private hospital, private home, and other. In TT, mothers are able to receive public medical services that are cheaper than medical facilities, including free child delivery at public hospitals. The ability of a woman to afford a private hospital or private doctor may be considerably influenced by her financial status. In addition to this purely economic sense, if there is a greater utilization of private medical facilities by certain ethnic group, and if the variable of privatized health care is statistically significant after controlling for other factors, then one can predict that there may be a cultural factor influencing the probability of child mortality through their own cultural preferences and tendencies. Hence, the variable for type of place child born is dichotomized into two categories; private doctor or facility (0=reference) versus public hospital (1=dummy). In the same manner as the creation of the variable of prenatal care history, variables of in what type of place was your child born for all children (up to the fifth youngest) of a mother were combined and averaged out. Of the sample population of 1,082, 956 women use only public hospitals (88.4%), 96 women use only private doctor or facility (8.9%), and other 30 women use both (2.7%). Further, the variable is collapsed into two categories for the final analysis; one is public hospital only or both coded as 1 (dummy), and the other is privatized health care only coded as 0 (reference). Quality of preventive health care for child Infants and children are extremely vulnerable to epidemic diseases and infectious illness, hence, inadequate child preventive health care directly risks child survival and a mothers choices in health care practices influence the health and survival of the child. To assess mothers health practices related to child preventive health care, a new variable of

PAGE 75

63 quality of preventive health care history was created using a number of questions about vaccination use for children in TTDHS. Children were assigned to either use or non-use category based on vaccination information from the child health cards, such as polio1, polio2, polio3, DPT1, DPT2, DPT3, yellow fever, measles, and if mothers have the child health and had tetanus. In the case mothers did not have the child health card, they were asked to recall whether or not the child had received a specific vaccination. The variable whether or not a health card for hospital care is possessed by mothers serves as a rough indicator of the accessibility for hospital care and the mothers attention to her childrens health. UNICEF reported that for children without child health cards, the proportion of vaccinations given is smaller than for children with child health card (2003a). Overall, these questions intended to capture mothers care for their children and relations between actual child health and frequencies of receiving the health services including vaccination. Questions on ten vaccinations and the child health card are asked mothers about her children from the youngest to the fifth youngest. All variables are dichotomous; yes (1) or no (0), which are combined into one index, quality of preventive health care history. The newly created variable has the range between 0 and 1. Of the sample population, 103 mothers had their all children receive all vaccinations and had child health cards of all their children (9.5%), 40 mothers had never had any of their children receive vaccination and had no child health card (3.7%), and the rest, 939 mothers were inconsistent to provide their children preventive health care (86.8%). The index of preventive health care history is used as scale variable, and the mean score is 0.7316 for the sample population. Mahabir reported that East Indians tend to have eastern medical treatment or folk/traditional medical treatment (1997). Although education contributes to improved

PAGE 76

64 awareness of importance in receiving vaccination appropriately, one may presume that a womans skills in health care practices and health orientation are have been nurtured influenced by her family orientation and may be eventually rooted in their cultural and ethnic origin. Hence, the level of quality of preventive health care is considered as important measures for both influence of cultural orientation on childs health and accessibility to modern medicine. Breastfeeding Breastfeeding have been found to be an important factor in infant survival, even after controlling for other variables that affect child mortality; infants who are bottled-fed from birth run a higher risk of health and development problems than do breast-fed children (McCann et al. 1981, Goldberg et al. 1984). Forste et al. (2001) reported that after controlling for socio-economic background and birth characteristics, race remained a strong predictor of breastfeeding. If differences in breastfeeding practices between African women and East Indian women would be observed in the sample, and if women, who did not breastfeed, have experienced more loss of their children than their counter part, then we can assume that one of the two major groups, which practice less adequate breastfeeding than the other racial group is subject to the additional disadvantage of higher child mortality. The variable of months of breastfeeding is used for creating the variable, breastfeeding history. It is a discrete variable including two special codings for inconsistent and never breastfeed. Similar to the three health related variables explained above, mothers provided answers how long they breastfeed their children from the youngest to the fifth youngest child, and never breastfeed is recoded as 0 and one or more months are recoded as 1. After combining the variables of five children and calculating

PAGE 77

65 the average months of breastfeeding, 209 mothers have never breastfed (9.9%), 873 mothers breastfed all their children (80.7%), and 67 mothers are inconsistent in breastfeeding (9.4%). This variable, namely breastfeeding history, is dichotomized into two categories; mothers who breastfed all their children is coded as 0 (reference), namely adequate breastfeeding, and mothers who never breastfed or were inconsistent in breastfeeding is coded as 1 (dummy variable), namely inadequate breastfeeding. Procedures of Data Analysis The first stage of the analysis is to use descriptive statistics stratified by ethnicity to summarize the basic characteristics of the sample. This step is meaningful for assessing the basic relationships between ethnicity and other factors such as demographic, socioeconomic, and health-related factors, and for simply assessing how well ethnicity can be a significant factor and predictor in the analysis of this study. The second step will be bivariate analyses to examine the relationships between all predictors in the later logistic regression analyses and the variable of child mortality. First, the proportion of women who have lost at least one child for all explanatory variables are presented, followed by odds of woman who experienced a child loss for all variables in order to check individual predictive power, and correlation coefficients between ethnicity and other predictors in order to demonstrate the influence of ethnic background on each factor. The last stage in the analysis will consist of an examination of multivariate logistic regression models to determine whether or not influences of mothers ethnic background on child mortality can be observed after controlling for all other socioeconomic and health care factors. The Generalized Linear Model used will be the logistic regression model where the random component is (X) the probability that a mother, who has at

PAGE 78

66 least one child in the past 10 years, has lost a child, and X is the vector of explanatory variables, namely ethnicity = X 1 place of residence = X 2 unmarried (a dummy variable of marital status) = X 3 cohabitating/visiting relations (another dummy variable of marital status) = X 4 years of educational attainment = X 5 quality of life = X 6 quality of preventive health care history = X 7, inadequate prenatal care (dummy variable of adequacy of prenatal health care history) = X 8, public hospital use only (dummy variable for privatized health care history) = X 9 inadequate breastfeeding (dummy variable of breastfeeding history) = X 10 The full logistic regression model has the form; Logit() = + 1 X 1 + 2 X 2 + 3 X 3 + 4 X 4 + 5 X 5 + 6 X 6 + 7 X 7 + 8 X 8 + 9 X 9 + 10 X 10 The first logistic regression model of the first phase will include only mothers ethnic background (Model 1). The second regression model introduces variables for demographic factors, place of residence and marital status, in the equation (Model 2), followed by the inclusion of socioeconomic factors, years of education and quality of life (Model 3). The next is the full model introducing the four health related indicators (Model 4). The multiple logistic regression analysis attempts to provide the combined effects of all variables on the likelihood of mother who have experienced a child loss, and to indicate the potential variables playing a significant role in determining the probability of child mortality in the society of TT. A series of nested logistic regression models systematically provide ethnic background variation for estimating direct or indirect effect on the child mortality through other factors determining child survival.

PAGE 79

CHAPTER 4 DATA ANALYSIS Ethnic Differentials in Child Mortality and Its Determinants The previous research have found that race and ethnicity as well as urban-rural differences, marital status, education, income (in this study, quality of life), and health related variables are important predictors of child survival chance and child mortality. Since a major concern of this study is whether mothers ethnic background matters to the incident of child mortality after controlling for other predictors, sample population distribution by all other variables are presented first. Descriptive examinations of the dependent and independent variables by ethnicity are presented in Table 4-1, which show how ethnicity connects with other factors predicting child mortality in this study. Reading across each row, we can compare the proportions of African and East Indian within each category of dichotomous independent variables and the means years of education, quality of life, and quality of preventive health care for African and East Indian. Demographic Characteristics Ethnic differences in child mortality presented in the first two rows of Table 4-1 indicate that approximately 7% of the mothers have experienced a loss of their child. The proportion of mothers with children born within 10 years who have lost at least one child is slightly higher among African mothers (8.3%) compared with East Indian mothers (5.6%). The differences between the two ethnic groups are moderately significant at 0.09 statistically. 67

PAGE 80

68 Table 4-1. Characteristics of mothers with children born in last 10 years by ethnicity [1] Childhood Mortality(*)Never Lost Child93.2%91.7%94.4%Have Lost at Least One Child6.8%8.3%5.6%[2] Type of Place of Residence**Urban40.8%55.5%28.5%Rural59.2%44.5%71.5%[3] Marital Status**Married58.4%34.1%78.6%Separated / Divorced / Widowed6.4%8.7%4.4%Cohabiting / Visiting Relations35.2%57.1%16.9%[4] Years of Education (0~16)**Mean(Standard deviation)[5] Quality of Life (0~6.65)**Mean(Standard deviation)[6] Preventive Health Care History (0~1.00)(*)Mean(Standard deviation)[7] Prenatal Care HistoryAdequate Prenatal Care83.6%83.7%83.6%Inadequate Prenatal Care16.4%16.3%16.4%[8] Privatized Health Care History**Privatized Health Care Only8.9%4.9%12.2%Public Hospital or Both91.1%95.1%87.8%[9] Breastfeeding History**Adequate Breastfeeding80.7%85.6%76.6%Inadequate Breastfeeding 19.3%14.4%23.4%Source: Demographic and Health Survey, Trinidad and Tobago, 1987 (Individual dataset)Sample Population = 1,082Statistical significance for the association between ethnicity and variables: (*) 0.10
PAGE 81

69 Ethnic differences in demographic background are noted in the next two panels, [2] type of place of residence and [3] marital status. Both variables are strongly associated with ethnic background as indicated by the statistical significances of p<0.01. The TT population is historically considered to be divided into two groups; roughly, Africans inhabiting urban areas and East Indians inhabiting rural areas. The finding from the TTDHS represents this socially constructed residential difference. The majority of the East Indian division live in rural areas (71.5%) while a greater number of the African division live in urban area (55.5%). The distinct ethnic difference is also observed in the marital patterns. The marriage rate of East Indian mothers (78.6%) is more than double the marriage rate of African mothers (34.1%). The comparatively smaller marriage proportion among African mothers reflects the predominance of cohabiting/visiting relations within this population division; the proportion of African mothers who have cohabiting/visiting relations is substantially higher than the comparable figures for East Indian mothers with respective proportions of 57.1% and 16.9%. The conspicuous ethnic differences in marital status are found controlling for religious affiliations in the separate analysis [data not shown]. Although the influence of religion on the marital pattern are observed, there are still clear association with the dominant type of marital status cohabiting or visiting relationship for African mothers, married for East Indian mothers. This shows that ethnic identity on marital status permeates religious influence and it could manifest a distinctive cultural notion for each ethnic group. Urban-rural residence is regarded as a crude proxy measure for physical access to modern health services and thus, the potential for variation in the effect of education on child mortality risk across accessibility is expected. In considering this point and while

PAGE 82

70 remembering that the majority of East Indians reside in rural areas, the East Indian division could be considered the higher risk group in child mortality in terms of accessibility to modern health facilities and maternal education. Socioeconomic Characteristics The next two variables presented in Table 4-1 are maternal socioeconomic background, years of education and a measure of quality of life. The mean number of years of education among African women is 8.00 and East Indian womens educational attainment is lower than African women by 0.55 years. This difference is highly statistically significant with p-value less than 0.01. Referring to the growth ratio of educational attainment comparing younger generation (15-34) and older generation (35-48) in the separate analysis, younger East Indian women have increased their educational accomplishments by 31.8% and the comparable ratio among younger African women is 5.7% [data not shown]. The two numbers suggest that years of schooling had increased for both ethnic groups, and East Indian women had experienced much greater relative improvement. Consequently, the gap in the average years of education between the two major ethnic groups decreased from 1.79 years to 0.37 year, but still East Indian women were slightly disadvantaged in educational attainment as of the time the TTDHS was conducted. Quality of life is another indicator that enables us to observe mothers socioeconomic status of the mothers as an alternative of income in this study. The higher mean of quality of life observed for East Indians (1.952) compared with Africans (1.762). The difference is highly significant less than 0.01. Contradictory arguments made by each ethnic group in TT over which ethnic group is economically advantageous may be nested in the contradictions of two important factors representing socioeconomic status; educational achievement and economic

PAGE 83

71 advantage. It reflects a unique influence of ethnic backgrounds on probability of child mortality in TT. The probability of child mortality that a mother would lose at least one child also can be used as a measure to examine which of the two factors is a more plausible reality expressing well-being in TT. Child Health Care Practices This study is also interested in whether or not mothers ethnic background influence health care practices. If there is an association between ethnicity and health care factors, then interest is furthered as to what degree and how mothers ethnic/cultural orientation involves their health care practices related to their childrens survivorship. Preventive health care history measures mothers practices on whether they have their children receive vaccinations against infectious diseases and illness based on the records of their children born within 10 years prior to the survey. The mean of preventive health care for African women is higher than for East Indian women with respective means of 0.745 and 0.720. The difference is statistically significant moderately at 0.093. The next indicator related to health care is prenatal care history whether or not women had adequate prenatal care history based on the records of their children born within 10 years. The proportions of women who had adequate prenatal care history for African mothers and East Indian mothers are almost same with respective proportions of 16.4% and 16.3% and there is no statistical significance. Accessibility to appropriate and necessary medical care is another important dimension of a populations level of well-being. Mothers can receive public medical services in TT that are cheaper than private medical services, particularly, child delivery at public hospital is free. Therefore, one could assume that a person who can afford medical care at private hospitals should experience a higher quality of health care. But it

PAGE 84

72 may not only be influenced by their socioeconomic level. For example, their selection as to whether to receive vaccination or western medicine, should be taken into consideration by paying careful attention to the cultural context peculiar to the society of TT. The sample population is cross-classified by privatized health care only versus public hospital or both for each ethnic group (Table 4-1 [8]). Percentage for East Indian mothers who tend to practice privatized health care within ethnic group is 12.2%, which appears to be roughly 2.5 times larger than African mothers (4.9%). Referring to the growth ratio of proportion of privatized health care for younger generation (15-34) compared to older generation (35-40) in the separate analysis [data not shown], the proportions of women for privatized health care are lower among younger age; 3.8% for African and 9.7% for East Indian, compared to among older age; 6.7% for African and 16.7% for East Indian. The gap between the two ethnic groups slightly narrowed over time (-4.1 point); however, a greater use of private clinics among East Indian mothers is observed in both age groups. This situation implies that although delivery is free in public hospitals for all women in TT, East Indian women have tended to use private clinics. Last two panels present breastfeeding selection indicating that about 19.3% of the mothers adequately breastfed their children. The percentage is considerably higher for African mothers (85.6%) compared with East Indian mothers (76.6%). The difference is highly statistically significant with p-value less than 0.01. African mothers are more likely to have breastfed adequately compared to East Indian mothers. Based prior literature and examination of associations between ethnicity and maternal characteristics, East Indian mothers are more likely to have characteristics associated with higher incidence of child mortality living in rural areas, lower

PAGE 85

73 educational attainment, lower quality of preventive health care history, lower level of breastfeeding history than are their African counterparts. However, observed in this study, African mothers seem also disadvantageous in terms of marital status, quality of life, and privatized health care history. Not only can we not satisfactorily determine which of the factors is most influential on child mortality but also, whether privatized health care is mainly influenced by purely economic condition or whether it is more likely a the matter of cultural preference. From the findings, the socioeconomic section however, we observed an unusual inversion of the standard relationship between two indicators of human capital; educational and economic achievement. Both have been considered important determinants of child survivorship. This unusual situation may facilitate TT society to conceive an unfeasible sphere in which factors interact uniquely. For furthering our understanding of nature of explanatory variables on child mortality in TT, the individual effects of these variables on child mortality are examined next. Influence of Maternal Characteristics on Child Mortality The first column in Table 4-2 shows the proportions of mothers who have lost at least one child within 10 years prior to the TTDHS, and mean years of education, quality of life, and preventive health care for those mothers. The second column reports the odds ratio of mothers who have experienced a child loss for each of the factors. The third column presents the correlation coefficients indicating the associations and their directions between ethnicity and all independent variables. Influence of Demographic and Socioeconomic Factors The proportions of mothers who have experienced a child loss within categories of each of the demographic variables appear in panels [1], [2], and [3] in the first column of Table 4-2. The percentage is higher for rural areas (7.0%) compared with urban areas

PAGE 86

74 (6.6%). More women who have cohabiting or visiting relationships have experienced a child loss (7.9%) than separated/divorced/widowed women (4.3%) and married women (6.5%). These results are somewhat unexpected and contradictory to previous research, since the support of the childrens father is important for child survival and health economically, practically, and emotionally, the category of married women is normally expected to be the lowest disadvantaged group in child mortality. In addition, the group of separated/divorced/widowed women who are normally considered insufficient in maternity assistance has a relatively small proportion of mothers experiencing child loss. However, mothers in the category of separated/divorced/widowed mothers do not seem to associate with higher child mortality. This may be caused by the considerably smaller number of separated/divorced/widowed mothers; 69 cases, 3.1% of the sample population. The individual effect of each demographic factor on child mortality is reported in the second column of Table 4-2. With the exception of ethnicity, each of demographic factors does not significantly influence the likelihood of being women with a child loss. Ethnicity correlates with urban-rural differences and marital standings as shown in the third column. 18 Education level is an important achieved human capital variable because it covariates with economic life chances, and therefore with child health and survivorship. The mean years of education for mothers with an experience of child loss is slightly shorter (7.59years) compared with their counterparts (the mean for all alive =7.70 years), 18 The variable of marital status is dichotomous in the examination of correlation with ethnic identity. Married is 0 and other two categories, separated/divorced/widowed and cohabiting/visiting relationship, are collapsed into one category as 1.

PAGE 87

75 Table 4-2. Proportions and odds for mothers who have lost at least one child and correlations between ethnicity and maternal characteristics [1] Ethnicity-East Indian (0)5.6% (*)African (1)8.3% 1.53(*)[2] Type of Place of Residence-0.274**Urban (0)6.6%Rural (1)7.0%1.07 ** Married (0)6.5%Separated / Divorced / Widowed (1)4.3%0.66Cohabiting / Visiting Relations (1)7.9%1.23[4] Years of Education (0~16)0.990.450**Mean(Mean for All Alive)[5] Quality of Life (0~6.65)0.79*-0.083**Mean(Mean for All Alive)[6] Preventive Health Care History (0~1.00)0.35**0.051(*)Mean(Mean for All Alive)[7] Prenatal Care History0.049(*)Adequate Prenatal Care (0)6.2% *Inadequate Prenatal Care (1)10.2%2.99(*)[8] Privatized Health Care History0.128**Privatatized Health Care Only (0)2.1% *Public Hospital or Both (1)7.3%3.70(*)[9] Breastfeeding History-0.113**Adequate Breastfeeding (0)4.2% **Inadequate Breastfeeding (1)17.7%4.86**Source: Demographic and Health Survey, Trinidad and Tobago, 1987 (Individual dataset)Sample Population = 1,082Statistical significance: (*) 0.10
PAGE 88

76 but the two groups are not statistically independent. The odds of mothers with a child loss indicates that the higher the educational attainment, the lower the odds of being a mother who has a child loss. But the difference is too small and there is also no statistical evidence. Another measure of socioeconomic status, mean score of quality of life, indicates that level of quality of life for mothers with a child loss is also lower (1.580) than those with no child loss (1.887). A t-test for independence showed that the mean score of quality of life for mothers with a child loss and those with no child loss differs significantly with p-value less than 0.01. The odds of having experienced a loss of child for quality of life is 0.79 (b-coefficient = -.239) means that the higher the level of quality of life, the lower the likelihood of child mortality. Moreover, the quality of life influences the likelihood of child loss significantly at the p-value of 0.028. These results support the previous studies that quality of life, which is considered a factor having an abstract concept substituting for income and housing quality in this study and having a significant association with child mortality. As we observed in the previous section, there is an unusual inversion in educational attainment and level of quality of life between the African division and the East Indian division. The unexpected non-significant influence of education on child mortality may be a reflection of this unique relationship. In terms of likelihood of child mortality, we may expect that the influence of quality of life is stronger than that of education. Health-related Proximate Factors Lastly, this study is interested in the important inquiry involving possible modification in the child mortality pattern by the particular role of health-related

PAGE 89

77 orientation. The level of accessibility to modern health services is expected to narrow the differences in the higher and the lower child mortality group. The mean score of quality of preventive health care for women who have experienced a child loss (0.664) is lower compared with that for women who have no experience of child loss (0.737) that is exhibited in the panel [6] in the first column of Table 4-2. A t-test for independence demonstrates that differences between the mean preventive health care score between the two groups are statistically significant less than 0.01. Inadequate prenatal care represents that poorer prenatal care associates with higher probability of child mortality. The group of mothers who are inadequate for prenatal care has a larger proportion of mothers with a child loss (10.2%) compared to their counterparts (6.2%). This association is moderately significant at 0.071. The odds of having a child loss for mothers having received inadequate prenatal care is 2.99 times greater than that for mothers having received adequate prenatal care, and is moderate statistically significant at 0.09. The public hospital category composes 7.3% of mothers with a child loss, and the odds of using public hospital only for mothers with a loss of child is 3.70 times larger than that of their counterpart with a moderate statistical significance at 0.071. Breastfeeding history, which showed a strong association with ethnic identity in the previous section, also presented a significant association on the likelihood of child mortality. The proportion of mothers with a child loss within the group of inadequate breastfeeding is 17.7% that is larger that of adequate breastfeeding (4.2%). As clearly shown in the second column, inadequate breastfeeding is 4.86 times higher probability of child mortality than adequate breastfeeding. The association is

PAGE 90

78 highly statistically significant less than 0.01. Hence, mothers in higher quality of preventive health care, adequate prenatal care, privatized health care, and adequate breastfeeding appear to have characteristics of lower child mortality. These results indicate that generally favorable treatments for the pregnancy and delivery correspond with a greater chance of child survival. While ethnicity, quality of life, and health related factors demonstrate significant influences on child mortality, type of place of residence, marital status, and years of education seem to have no influence on child mortality. A review of previous studies provides substantial evidence of causal connections in the level of strength among child survival and geographical settings concerning access to basic health resources, services and maternal education. Particularly, education has been stressed in playing a pivotal role in decreasing child mortality. However, overall, the level of educational attainment itself could not explain it conclusively. Instead, specific knowledge and awareness of appropriate maternity and child health care, rather than formal education, may be more proximate factors of child survival. The degree of strength in causal connections would also depend on the geographical location of the nation and the extent of the public transportation system within the nation. Living on small and relatively wealthy islands in the Caribbean, TT society has been able to develop an extensive cheap private transportation system. If people have fairly equal access to the modern health services, the differences between African and East Indian in use of preventive health care, prenatal care, privatized health care, and decision to breastfeed can be considered a purely ethnic conventional preference influenced by their community and their culture.

PAGE 91

79 Multivariate Logistic Regression Models I presented bivariate analyses for understanding the individual relationship between child mortality and each of independent variables in multivariate analyses. Some findings in terms of rural-urban differences, marital status, and educational attainment did not support previous research. Simultaneously, there are paradoxes when the associations between ethnic background and other mothers characteristics were employed. To examine the net of these various factors influence on the probability of child mortality, the relationships between child mortality and indicators by means of the multivariate logistic regression analysis, which allows us to obtain more definite comparisons between the two ethnic groups in terms of child mortality is explored next. Ethnic Influence on Child Mortality Table 4-3 presents five nested logistic regression analyses. Model 1 includes only ethnicity. The coefficient for ethnicity is .428. By taking the antilog of the coefficient of ethnicity, the probability that African mothers have lost at least one child is 1.534 times more than East Indian mothers. Demographic factors and socioeconomic factors are introduced into Model 2 and Model 3 respectively. The coefficients for ethnicity in the two models indicate that the East Indian division consistently has a lower probability of having experienced a child loss than the African division after controlling for demographic factors in Model 2, and after controlling for socioeconomic factors in Model 3. Model 4 includes all explanatory variables. The odds ratio of Africans increases to 2.177. The non-significant factors are removed from the model to produce the best fit with the fewest variables, and this final model is presented in the last column. Differences between the chi-squares reported for Model 4 and Model 5 are not significant ( 2 =2.727

PAGE 92

80 with df=3, p>0.250) indicating that the simple model fits the data as well as the full model. Not only can ethnicity be considered as a powerful indicator of the probability of child mortality in this society indicated as p-value = 0.014, but also its Exp(b) for Africans keep its value as 2.026 accounting the probability having experienced a child loss in this model. The crucial question for this study if ethnic background persists in child mortality controlling for socioeconomic, demographic, and health-related factors in the TTDHS is answered; ethnic background is statistically significant. The East Indian division appears to have a lower probability in child mortality than the African division after controlling for maternal characteristics. As variables are included in order, the differences between the two divisions seem to widen and strengthen. Urban-Rural Setting, Marital Status, and Scio-economic Influences Demographic factors, place of residence and marital status are introduced in Model 2. After controlling for ethnicity and marital status, mothers living in rural areas are 21.1% more likely to have experienced a child loss compared with their counterparts. After controlling for ethnicity and place of residence, separated/divorced/widowed mothers are 76.3% (e .567 = 1.763) more likely and mothers who have cohabiting/visiting relationships are 0.1% more likely having had a child loss compared with married women. But the influence of disadvantage for rural settings and forms of union (married and separated/divorced/widowed) in child mortality is no statistically significant. Model 3 introduces socioeconomic variables of years of education and quality of life. Mothers who have a higher educational attainment are less likely to have experienced a child loss, but this is not statistically significant. Instead, the second socioeconomic variable, quality of life statistically significant, therefore, mothers who have a

PAGE 93

Table 4-3. Probability of having lost at least one child controlling for demographic, socioeconomic, and health care factors (Logistic Regression) 81 Constan t -2.612**-2.89**-2.393**-1.521*-1.079*EthnicityEast Indian (ref.)African .428 ( ) 1.534.499 ( ) 1.647.484 ( ) 1.623.778 ( ) 2.177.709*2.026Place of ResidenceUrban (ref.)Rural.1911.211.0781.082.2301.259Marital StatusMarried (ref.)Separated / Divorced / Widowe d -.567.567-.692.501-1.357 ( ) .257-1.367 ( ) .255Cohabiting / Visiting Relations.0011.001-.115.891-.140.869-.106.899Years of Education-.029.733-.016.984Quality of Life-.232*.793-.217 ( ) .805-.266*.766Preventive Health Care History-.983*.374-.977*.377Prenatal Care HistoryAdequete Prenatal Care (ref.)Inadequate Prenatal Care1.477*4.3781.491*4.444Privatized Health Care HistoryPrivatized Health Care Only (ref.)Public Hospital or Both.9222.514Breastfeeding HistoryAdequete Breastfeeding (ref.)Inadequete Breastfeeding1.718**5.5741.713**5.548-2Log likelihood2 (df)Model p-valueSource: Demographic and Health Survey, Trinidad and Tobago, 1987 (Individual dataset)Sample Population = 1,082Statistical significance: (*) 0.10< P <0.05, 0.05< P <0.01, ** P <0.01<.001.076.307.164<.001479.4103.146 (1)4.810 (4)9.176 (6)63.147 (10)60.420 (7)536.684535.019530.654476.683 b coef.Exp(b) b coef.Exp(b) b coef.Exp(b) b coef.Exp(b)Variables/ValuesModel 1Model 2Model 3Model 4Model 5 b coef.Exp(b)

PAGE 94

82 lower level of quality of life potentially have a higher risk of child loss. After controlling for ethnicity, place of residence, marital status, years of education, when quality of life decreases by one point, the probability to have lost a child is multiplied by 1.26 (e 0.232 ). Model 3 is not significantly stronger than Model 2 (4.366 with df=2, 0.25>p>0.1). However, quality of life seems to be a valuable determinant for child mortality directly/indirectly, combining the findings in the bivariate analysis and the multivariate analysis. Since the level of quality of life appears to be higher for the East Indian division, both ethnic background and economic situation work towards East Indian mothers preferably. Influence of Health-related Factors on Child Mortality Model 4 assesses the effect of ethnic background on child mortality by adding health-related factors. In this model all factors are included. Two findings stand out in the model. Ethnicity achieves its statistical significance at 0.01, and odds ratio for African mothers increases from 1.623 in Model 3 to 2.177 in Model 4. This implies that the effect of ethnic differences with socio-demographic factors only and the combined effect including health-related factors account for approximately 55% of the child mortality difference between Africans and East Indians. The inclusion of four health related factors adds a significant strength in the inclusive model. Differences between the chi-squares reported for Model 3 and Model 4 are highly significant ( 2 =51.147 with df=4, p<0.001). Effects of the health-related indicators on the probability of having a child loss follow the findings in the bivariate analysis. The odds for dummy variables for prenatal care and privatized health care show that mothers who have received adequate prenatal care, who have practiced privatized health care, and who have breastfed adequately are less likely to have experienced a loss

PAGE 95

83 of their child with respective odds of 4.378, 2.514, and 5.574. Prenatal care and breastfeeding are statistically significant less than 0.05 but privatized health care has no statistical significance. The probability having experienced a loss of child decreases with a increase of quality of preventive health care score; for women who have a higher score of quality of preventive health care are less likely to have had a child loss; when preventive health care increases one point, child mortality decreases by 2.67 (e .983 ). This is statistically significant less than 0.01. With the exception of years of education, the inclusion of health-related factors increases the effects of ethnicity, urban-rural settings, marital status, and quality of life, indicate that they are not working through health-related factors. On the other hand, educational difference is weakened by the inclusion of health-related factors suggesting that education potentially works through health-related factors to influence child mortality. Notably, ethnic background continues to have an independent effect on child mortality. To examine the effect of ethnicity further, the interaction terms between ethnicity and the variables in the simplified model (Model 5) was conducted separately, however, none of the interaction terms were significant suggesting that the factors influencing the probability of child mortality in reported in Model 5 do not vary by ethnic background.

PAGE 96

CHAPTER 5 CONCLUSION The purpose of this thesis has been to explore the economic and cultural differences between African women and East Indian women in TT by means of examining ethnic differentials in child mortality. There are a variety of factors relating to child mortality and to each other as well. All the factors are considered to be affected by mothers socio-economic standings hence, the issue of child mortality has an aspect of being socially determined similar to the context of ethnic issues. Discourse concerning inequality in socio-economic standings is split along ethnic lines in TT. Differences in the historical experiences of each population of Africans and East Indians and the development of ethnic identity through discriminatory relationships between old-timers and new-comers are found in the contemporary representation of political identity, socio-economic position, and residential isolation. As such, an unusual association between educational achievement and economic advantage is considered due to the unique distribution of political power and economic power sharing between the African division and the East Indian division. The juxtaposition of strikingly different ethnic group identities has resulted in a construction of crossing perceptions about the inequalities in general socio-economic standings toward others. Ethnicity in the Analysis of Child Mortality In order to explore ethnic inequalities in the health context, as a vital human capital, this thesis made use of the variable of child mortality constructed from the TTDHS data. Quantitative analysis in this study included three clusters of child mortality factors: 84

PAGE 97

85 demographic, socio-economic, and health-related factors. The findings are summarized as follows. First, Africans and East Indians were distributed quite differently in terms of residential and marital patterns. The ethnic difference in place of residence followed the historical evidence that the majority of African women reside in urban areas while the majority of East Indian women appeared to be rural residents. Also African women markedly differ from East Indians in terms of marital status. African women are more likely to have cohabiting or visiting relationships, while most East Indian women secure matrimonial relationships. The proportion of women who are married in the East Indian division is as twice as much as that of the African division. Second, an important contrast emerged in the socio-economic variables. The findings demonstrate manifest relationships between the level of educational attainment and the economic standings peculiar to the historical context of the TT society. African women had a higher level of educational attainment however, they had a lower score on quality of life compared to East Indian women. The differing livelihoods secured by the women in the African division compared to East Indian division women reflected the historical background held by each ethnic group. Third, in the cluster of health service use, strong differences were also found in terms of privatized health care history and breastfeeding history. East Indian women tended to use privatized health care compared to African women of whom less than 5% women use only private facility for delivery. In breastfeeding selection, African women are more likely to have breastfed compared to East Indian women. They also differed in preventive health care use. Proportionally, African women are more likely to have their children vaccinated compared to East Indian women, however, the difference is not so

PAGE 98

86 striking. Lastly, there was no evidence that these women are different in terms of prenatal care use. This result corresponds to the high attainment of prenatal care in TT. Unfortunately, there is no previous record to provide the information about the ethnic differences between Africans and East Indians, The results illustrated the heterogeneity between Africans and East Indians. Though the findings did not always indicate specific ethnic group advantages to maternal and child health, ethnic differences in residential and marital union patterns manifest that ethnic identity is historically and culturally constructed in the livelihood of Trinibagonian mothers. The unusual situation in which the wealthier East Indians are less educated compared the economically disadvantaged Africans may contribute to level off the socio-economic distinction between the two ethnic groups. At the same time, it may also cause an unclear relationship between health care factors and ethnic background. The second section of analysis presented the independent effect of each variable. How each of the mothers characteristics proportionally differentiates the child mortality is summarized as follows. First, the association between ethnic background and child mortality was observed. Africans have the larger proportion of mothers having experienced a child loss compared to East Indians. Second, the variations in demographic factors influence the slight difference in child mortality. But the associations are not always in the expected direction. Rural settings disfavor mothers in child mortality, which supports previous research. Though the maternal union patterns did not follow the common consensus of researchers, the married status did not seem to be the favored marital status, instead, the groups of mothers who have been separated, divorced, or widowed appear to compose the smallest proportion of which mothers have experienced

PAGE 99

87 a child loss. Third, years of school also did not appear to be a significant variable in determining child mortality, which is dissimilar to the previous findings in child mortality studies. On the other hand, the quality of life indicator showed its strength in influencing the child mortality. In TT, urban-rural distinction, marital status, and level of educational achievement do not seem to differentiate the child mortality. Fourth, health related factors indicated that higher standards of maternal and child health care show potential for reducing child mortality. The presence or absence of use in preventive health care, prenatal health care, and breastfeeding, as well as the delivery at private facility evidently yield the disparity in child mortality. An examination of the independent effect of each variable on child mortality showed that mothers who are of East Indian descent have higher scores of quality of life and have had a child exclusively at a private facility, both of which indicate affordability to access a better quality of health care. On the other hand, African women have characteristics of women who have no child loss in terms of better preventive health care history and adequate breastfed. In sum, referring prior research on determinants of child mortality, African women are more likely to have a lower risk of child mortality in terms of urban settings, higher education, higher score of preventive health care history, adequate prenatal care, and breastfeeding history, while East Indian women have been shown to have a lower risk of child loss in terms of married status, higher score of quality of life, and privatized health care history. The economically advantaged ethnic group, East Indian, is found to be lower usage of appropriate health care. The final analysis presents logistic regression models that elaborate the findings in the previous two analysis sections. The findings continuously showed that East Indian

PAGE 100

88 women had experienced lower levels of child mortality than African women after controlling for demographic, socio-economic, and health related factors. There is strong evidence that ethnic background differentiates the child mortality as the ultimate outcome of Trinibagonian mothers livelihood. Quality of life played a significant role throughout all models. The roles of health-related factors, with the exception of privatized health care history, were strongly significant. Specifically, breastfeeding and prenatal care use are very important factors to reduce the risk of child death. In the meantime, ethnic background conspicuously appeared to be a unique and significant factor related to child mortality. The findings indicate that ethnic identity pervades each of the variables. Characteristics of each of the factors relating to ethnic identity accumulated as clusters were added to the equation. This finding supports the previous research regarding the steadfast causal pathway of household socio-economic standings impacting child survival chances. On the other hand, ethnic background somehow plays a significant role to determine child survival chances. Interestingly, here ethnicity seems to behaving somewhat independently; although analyses do control for all other factors. The effect of ethnic background on child mortality gains its strength every time that the equation includes other variables with the exception of Model 3 in which socio-economic factors were included. Comparing Model 2 and Model 3, the effect of ethnic background on child mortality was slightly reduced from b-coefficient of 0.499 to that of 0.488, or reduced the association between ethnic background and child mortality by 3.7% (the odd ratio increased from 1.647 to 1.623). This result implies that the socio-economic factors contribute to reducing the risk of child mortality; i.e., the ethnic difference in child mortality is dependent upon the distribution of socio-economic factors. However,

PAGE 101

89 recalling the association between quality of life and ethnicity, the economic factor also works towards the East Indian sub-division. Thus, although socio-economic factors seem to compensate the ethnic gap in child mortality, the African mothers situation is even worse if they do not have an additional benefit of advantageous socio-economic factors. The evidence of Africans disadvantage in child mortality is further clarified in the comparison between Model 3 and Model 4. Inclusion of the effects of health care factors increased the association between ethnicity and child mortality by 88.9% (the odd ratio increased from 1.623 to 2.177). Thus, the ethnic difference in child mortality is less likely dependent upon the distribution of health care factors. While in the comparisons of health care factors between the two ethnic groups, Africans seem more advantaged in reducing the risk of child loss in terms of preventive health care and breastfeeding, which both appear to be significantly important determinants of child mortality in TT, the coefficient of the dummy variable for ethnicity increased from 0.484 (Model 3) to 0.778 (Model 4). This indicates that if an African woman does not have an additional benefit of health care, the risk of child loss is even worse. These empirical evidence from the TTDHS demonstrates that the level of educational attainment may not be measures for lowering the risk of child loss in TT, but economic status can be an important factor in reducing child mortality among African mothers. For mitigating the effect of economic inequality in child mortality, health care services in terms of providing further universal preventive health care services and further diffusion of knowledge of breastfeeding as well as distribution of accessibility to maternal care can substitute for the effects of economic disadvantage on child mortality.

PAGE 102

90 Locating Ethnic Context in Trinidad and Tobago In sum, the analysis implies an important reality in this nation that the ethnic identity exists in the context of child mortality in TT, while quality of life and three health care factors have significant powers in predicting the probability of child loss. In line with the posited hypotheses, I would like to begin with the second hypothesis, which was posited paradoxically in explaining ethnic context in TT society. Within the framework of social system theory: after introducing all variables in the equation, ethnic identity does not have a predictive power ethnic identity can be a means for forming groups to pursue common goal, but consequently, this collective identity can be absorbed into a social system. From the analysis of the TTDHS, however, distinct ethnic differences were observed in the aggregate and continue when the examinations were carried out. This implies that ethnicity has a direct and an indirect effect on child survival. Notably, the variables such as place of residence, education, privatized health care, and marital status, which are correlated with ethnic identity, are not significant in predicting child mortality. Ethnicity uniquely incorporates the roles of these variables features. Ethnic values within each factor may be compatible with some factors while the ethnic values can be bargained with other factors. This account coincides with the concept of the indirect influence of race or ethnicity on child survival; Hummer noted that, while some socio-demographic and proximate factors work uniformly across racial groups to affect infant mortality, others may operate uniquely within groups to promote or reduce the chances of survival (1993: 534). In his study, child mortality of African American women is approximately twice as high as that of Anglos for both exogenous and endogenous causes of death, which are nearly identical for the finding in this study. However, after controlling for demographic,

PAGE 103

91 socio-economic, health care factors, the effect of race on child mortality decreases every time the analysis includes factors; furthermore, racial influence is eliminated after controlling for child health (gestation and birth weight). We need further research exploring and including such factors as child health in comparisons between the ethnic groups in child mortality for capturing present misspecifications; however, in the analysis of the TTDHS, the ethnic child mortality differential cannot be attributable to health care factors in TT, although they do universally contribute to reducing the risk of child loss. Thus, we may conclude that in the TT society, ethnicity uniquely determines child survivorship, and there is a strong influence of ethnic identity functioning and organizing the social system where the social inequality results from differences in interests and control over scarce resources, and power depends on the relationship of the two groups. For the first hypothesis, I posited that socio-economic status is the most significant variable to predict the probability of child mortality. Indeed, quality of life remained significant after all other variables were included; however, it did not seem to be appropriate to say that socio-economic status is the strongest determinant for child mortality in TT society. In the meantime, the education effect was unexpectedly insignificant. Generally, a low level of education significantly elevates the risk of child mortality, but the relationship between formal education and child mortality has been ambiguous in previous studies. The findings support educations role, which is considered to assist women to overcome disadvantages of physical accessibility, i.e., proximity in terms of distance and travel time. On relatively small islands, people can have moderately equal access to modern health services. It is inferred from these results

PAGE 104

92 that urban-rural distinctions do not have the power to predict the probability of child mortality. However, we cannot assume that education as well as residential differences as indices of proximity may not matter in TT society. Although the African sub-population is generally considered to have characteristics of the lower child mortality group in terms of education and accessibility, it appeared to be the higher risk division in child mortality. Often, child mortality indicates that the education advantage in child survival is more pronounced in urban areas because of the more complicated social activities and the less reliance on familial mechanisms (Bicego and Boerma 1996). On the other hand, women in rural areas are still under various constraints, such as maintenance of traditional practices and beliefs. In these respects, it can be beneficial to investigate pregnancy outcomes within each community so that we may have more tangible understandings about the relations between education and child mortality. Caldwell indicates that maternal educations role is to change traditional patterns of family influences so that women may improve their understanding of the importance of using modern medical services and may overcome their skeptical perception of modern technology (Caldwell 1979, Caldwell et al. 1983). At the same time, investigation of the womens situation in the community of their ethnic counterpart may be informative in terms of comparisons between formal education and maternal education. Implicitly, privatized health care was used in this study as an indicator of socio-economic status as well as an indicator of cultural differences. It was expected to appear as a significant variable to strengthen the differences in child mortality between the two groups. However, probably because of correlation with

PAGE 105

93 ethnicity, it appeared to be offset one-sidedly. Nevertheless, privatized health care orientation is an important dimension considering the differences between Africans and East Indians in the TT society. In research examining the perceptions of the Health Centres, ethnic differences are manifest. Africans are more likely to be dissatisfied with doctors services while East Indians seem to be dissatisfied with nursing services, pharmacy services, and the management of the health centres. Mustapha and Singh (2000) conclude that the health centre users perceptions are probably biased in favor of members of their own ethnic group because the majority of doctors are of East Indian descent and the majority of other health services personnel, particularly nurses, are of African descent. Mahabir (1997) also reports that many women in the relatively traditional division, East Indian, relocate to urban areas to be among fellow East Indians, which makes it practical for them to congregate and share common cultural activities. Creolization may occur thorough modernization which can be seen by East Indian as pressures by the overwhelming Afro-Creole notion to modify East Indians behavior or to conform to a creolized society and culture. Thus, medical services provided by the public sector should seek a culture-sensitive health care scheme while, avoiding falling into the situation where the officially sanctioned medical system is based only on western science and technology. The beneficial impacts of the three health care factors were found as posited in the third hypothesis. No interaction between ethnicity and the three health care factors is significant, implying that each health care factor works independently from ethnic identity and thus, better health care associates with decrease of child mortality universally for all women in TT. Therefore, this result corresponds to the implication of previous

PAGE 106

94 research, that general efforts to lower infant mortality, including provision of more adequate perinatal care and preventive health care, may have the preferable impact for all groups (Hummer 1999). Recalling the increase of ethnic influence when health care factors are included, there should be unmeasured and latent factors in the context of the multi-cultural society beyond these strong health care factors in determining child mortality. This account can provide an answer to the question; does ethnicity matter in child mortality in TT? However, the answer itself conceives another question; why does ethnicity matter? What does ethnicity mean in this society? Unfortunately, we could not examine this question with the TTDHS. The event of child loss may be only a tip of the pile of accumulated thousands of causalities. We may, therefore, first wish to distinguish child deaths in age at death and in cause of death (as well as causes designed as endogenous and exogenous), and second wish to include misspecified factors in the framework of child survivorship in terms of; selectivity such that if the child loss occurred spontaneously, or because of abortion, miscarriage, or pregnancy complication, or because of fertility orientation. Maternal health and child health are also expected to be predictive measures such that if the baby is provided adequate nutrition, if habitus influences birth outcomes, and if the stressful circumstances affect the maternal conditions as well as physical information such as birth weight and genetic factors such as endogenous factors. Perception toward medical care, such as preference and choice of consumption goods, including childcare services, different resource of treatments, is expected to explain ethnic differences in health orientation. Specifically, we may argue that there is interaction between factors relating to individual behavior and orientation derived from their ethnic/cultural notion and socio

PAGE 107

95 economic status. Both may be highly controversial topics because each is socially constructed -health status and birth outcome are affected by the terms of ethnicity and socio-economic factors. Hence, there is no easy way to measure ethnicity and socio-economic status. Longitudinal research is requisite and investigations into the components and mechanisms of socio-economic status as they relate to health outcomes are crucial. Studying the interrelations between material capital, human capital, and social capital may bring out the mechanisms of the social determinants of health (Oakes and Rossi 2003). This way of studying may further shape our understanding of the ethnically stratified societys system, while we continue to refine our measures and collect data to satisfy the model of child survivorship. We can attach importance to socio-cultural factors, which operate on their health behaviors and attitudes in order for us to increase our understanding of the associations between ethnicity, socio-economic status, and child mortality. For accomplishing these aims, community level investigations and comparisons of ethnic differences in each community in terms of formal education, maternal education, and level and quality of associations and proximity of members within a community can be meaningful and may provide us profound insight into the relationships between ethnicity, health, and social systems. This study outlined ethnic identity, which developed and transformed into the competitive divisions within historical, political, and economic contexts. In this study, child mortality was employed as an indicator of socio-economic assessment and an outcome of mothers livelihood which is shaped by and influencing everyday economic activities and cultural activities. Child mortality itself cannot explain the ethnic

PAGE 108

96 differences between the two major ethnic groups directly; however, the quality of life, which is composed of multi-layered causations from politics to culture, may influence child well-being in various ways. The quality of health is composed of intricate combinations of perceptional, behavioral, and cultural characteristics. Over the long run, results of more in-depth research would be important to the formulation of public policy related to health programs. In many communities establishing timely and effective intervention of governmental health programs is imperative. A more complete socio-economic analysis of poly-ethnic communities in TT, emphasizing the cultural and customary differences in each community is needed in order to provide beneficial information capable of filling gaps in the understanding of Trinibagonian plural society and the appropriate administrative intervention in health care. In the light of the irreplaceable peculiarity of this nation we should regard cultural conservation, economic opportunity, and allocational equality in scarce resources including health care variations, separately and equally. These competitions between ethnic groups enliven citizens of TT, which, I believe, are the culture and the preciousness of this nation.

PAGE 109

APPENDIX LIST OF VARIABLES AND VALUES OF THE DEMOGRAPHIC AND HEALTH SURVEY IN TRINIDAD AND TOBAGO 1987 DATA Demographic ETHNIC: Ethnicity Value Label 1 African 2 Indian 3 Mixed 4 Other Missing Values: 9 AGE: Current age respondent Scale URBRURAL: Type of place of residence Missing Values: 9 Value Label 1 Urban 2 Rural MARISTA: Current marital status Value Label 0 Never married 1 Married 2 Living together 6 Visiting relation 7 Widowed/Divorced/Separated Socioeconomic EDSINGLE: Education in single years Scale Missing Values: 99 DWATER: Source of drinking water FLOOR: Main floor material Value Label 1 Piped into residence 2 Piped into yard/plot 3 Public tap 4 Well with handpump 5 Well w/o handpump 6 River, spring, surface 7 Tanker truck, vendor 8 Rainwater 9 Other Missing Values: 99 TOILET: Type of toilet facility Value Label 0 No facilities 1 Flush 2 Bucket 3 Pit 4 Other Missing Values: 9 ELECTRI: Has electricity Value Label 0 No 1 Yes TV: Has television Value Label 0 No 1 Yes Missing Values: 9 REFRIGE: Has refrigerator Value Label 1 No 2 Yes Missing Values: 9 CAR: Has car Value Label 0 No 1 Yes Missing Values: 9 Value Label 0 Other 1 Wood planks 2 Cement 3 Dirt 4 Terrazzo 5 Parquet/polished wd 6 Carpet 7 Linoleum, vinyl 8 Ceramic tile Missing Values: 9 97

PAGE 110

98 VIDEO: House has video Value Label 1 Yes 2 No Child Birth Record (child number 1 to 16) [Child-1] BIDX$01: Birth column number 1 Scale BORD$01: Birth order number 1 BIDX$05: Birth column number 5 Scale B2$01: Year of birth 1 Scale CALIVE01: Child is alive 1 B2$05: Year of birth 5 Value Label 1 No 2 Yes [Child-2] BIDX$02: Birth column number 2 Scale BORD$02: Birth order number 2 BIDX$06: Birth column number 6 Scale B2$02: Year of birth 2 Scale CALIVE02: Child is alive 2 B2$06: Year of birth 6 Value Label 1 No 2 Yes [Child 3] BIDX$03: Birth column number 3 Scale BORD$03: Birth order number 3 BIDX$07: Birth column number 7 Scale B2$03: Year of birth 3 Scale CALIVE03: Child is alive 3 B2$07: Year of birth 7 Value Label 1 No 2 Yes [Child-4] BIDX$04: Birth column number 4 Scale BORD$04: Birth order number 4 Scale B2$04: Year of birth 4 Scale CALIVE04: Child is alive 4 Value Label 4 No 2 Yes [Child-5] Scale BORD$05: Birth order number 5 Scale Scale CALIVE05: Child is alive 5 Value Label 1 No 2 Yes [Child-6] Scale BORD$06: Birth order number 6 Scale Scale CALIVE06: Child is alive 6 Value Label 1 No 2 Yes [Child-7] Scale BORD$07: Birth order number 7 Scale Scale CALIVE07: Child is alive 7 Value Label 1 No 2 Yes

PAGE 111

99 [Child-8] BIDX$08: Birth column number 8 Scale BORD$08: Birth order number 8 [Child-12] Scale B2$08: Year of birth 8 Scale CALIVE08: Child is alive 8 Scale Value Label 1 No 2 Yes [Child-9] BIDX$09: Birth column number 9 Scale BORD$09: Birth order number 9 Scale B2$09: Year of birth 9 Scale CALIVE09: Child is alive 9 Value Label 1 No 2 Yes [Child-10] BIDX$10: Birth column number 10 Scale BORD$10: Birth order number 10 Scale B2$10: Year of birth 10 Scale CALIVE10: Child is alive 10 Value Label 1 No 2 Yes [Child-11] BIDX$11: Birth column number 11 Scale BORD$11: Birth order number 11 Scale B2$11: Year of birth 11 Scale CALIVE11: Child is alive 11 Value Label 1 No 2 Yes BIDX$12: Birth column number 12 Scale BORD$12: Birth order number 12 B2$12: Year of birth 12 Scale CALIVE12: Child is alive 12 Value Label 1 No 2 Yes [Child-13] BIDX$13: Birth column number 13 Scale BORD$13: Birth order number 13 Scale B2$13: Year of birth 13 Scale CALIVE13: Child is alive 13 Value Label 1 No 2 Yes [Child-14] BIDX$14: Birth column number 14 Scale BORD$14: Birth order number 14 Scale B2$14: Year of birth 14 Scale CALIVE14: Child is alive 14 Value Label 1 No 2 Yes [Child-15] BIDX$15: Birth column number 15 Scale BORD$15: Birth order number 15 Scale

PAGE 112

100 B2$15: Year of birth 15 Scale CALIVE15: Child is alive 15 Value Label 1 No 2 Yes [Child-16] BIDX$16: Birth column number 16 Scale BORD$16: Birth order number 16 Scale B2$16: Year of birth 16 Scale CALIVE16: Child is alive 16 Value Label 1 No 2 Yes Maternal and Child Health Care Information (child number 1 to 6) [Child-1] PRENATA1: Prenatal care before birth-1 Value Label 0 No one 1 Doctor 2 Trained nurse 3 Trained midwife 4 Birth attendant 5 Other Missing Values: 9 MBREAST1: Months of breastfeeding-1 Value Label Scale 94 Never breastfed 95 Inconsistent Missing Values: 99 HCARD_1: Has health card-1 Value Label 0 No card 1 Yes, seen 2 Yes, not seen Missing Values: 9 BCG_1: Received BCG-1 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT1_1: Received DPT 1-1 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO1_1: Received POLIO 1-1 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT2_1: Received DPT 2-1 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO2_1: Received POLIO 2-1 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT3_1: Received DPT 3-1 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO3_1: Received POLIO 3-1 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 MEASLE_1: Received MEASLES-1 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9

PAGE 113

101 CHIBORN1: Type of place child born-1 Value Label 1 Government hospital 2 Private hospital 3 Private home 4 Other Missing Values: 9 YFEVER1: Received Yellow Fever-1 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 [Child-2] PRENATA2: Prenatal care before birth-2 Value Label 0 No one 1 Doctor 2 Trained nurse 3 Trained midwife 4 Birth attendant 5 Other Missing Values: 9 MBREAST2: Months of breastfeeding-2 Value Label Scale 94 Never breastfed 95 Inconsistent Missing Values: 99 HCARD_2: Has health card-2 Value Label 0 No card 1 Yes, seen 2 Yes, not seen Missing Values: 9 BCG_2: Received BCG-2 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT2_2: Received DPT 1-2 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO1_2: Received POLIO 1-2 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT2_2: Received DPT 2-2 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO2_2: Received POLIO 2-2 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT3_2: Received DPT 3-2 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO3_2: Received POLIO 3-2 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 MEASLE_2: Received MEASLES-2 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 CHIBORN2: Type of place child born-2 Value Label 1 Government hospital 2 Private hospital 3 Private home 4 Other Missing Values: 9

PAGE 114

102 YFEVER2: Received Yellow Fever -2 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 [Child-3] PRENATA3: Prenatal care before birth-3 Value Label 0 No one 1 Doctor 2 Trained nurse 3 Trained midwife 4 Birth attendant 5 Other Missing Values: 9 MBREAST3: Months of breastfeeding-3 Value Label Scale 94 Never breastfed 95 Inconsistent Missing Values: 99 HCARD_3: Has health card-3 Value Label 0 No card 1 Yes, seen 2 Yes, not seen Missing Values: 9 BCG_3: Received BCG-3 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT1_3: Received DPT 1-3 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO1_3: Received POLIO 1-3 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT2_3: Received DPT 2-3 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO2_3: Received POLIO 2-3 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT3_3: Received DPT 3-3 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO3_3: Received POLIO 3-3 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 MEASLE_3: Received MEASLES-3 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 CHIBORN3: Type of place child born-3 Value Label 1 Government hospital 2 Private hospital 3 Private home 4 Other Missing Values: 9 YFEVER3: Received Yellow Fever-3 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9

PAGE 115

103 [Child-4] PRENATA4: Prenatal care before birth-4 Value Label 0 No one 1 Doctor 2 Trained nurse 3 Trained midwife 4 Birth attendant 5 Other Missing Values: 9 MBREAST4: Months of breastfeeding-4 Value Label Scale 94 Never breastfed 95 Inconsistent Missing Values: 99 HCARD_4: Has health card-4 Value Label 0 No card 1 Yes, seen 2 Yes, not seen Missing Values: 9 BCG_4: Received BCG-4 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT1_4: Received DPT 1-4 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO1_4: Received POLIO 1-4 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT2_4: Received DPT 2-4 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO2_4: Received POLIO 2-4 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT3_4: Received DPT 3-4 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO3_4: Received POLIO 3-4 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 MEASLE_4: Received MEASLES-4 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 CHIBORN4: Type of place child born-4 Value Label 1 Government hospital 2 Private hospital 3 Private home 4 Other Missing Values: 9 YFEVER4: Received Yellow Fever-4 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 [Child-5] PRENATA5: Prenatal care before birth-5 Value Label 0 No one 1 Doctor 2 Trained nurse 3 Trained midwife 4 Birth attendant 5 Other Missing Values: 9

PAGE 116

104 MBREAST5: Months of breastfeeding-5 Value Label Scale 94 Never breastfed 95 Inconsistent Missing Values: 99 HCARD_5: Has health card-5 Value Label 0 No card 1 Yes, seen 2 Yes, not seen Missing Values: 9 BCG_5: Received BCG-5 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT1_5: Received DPT 1-5 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO1_5: Received POLIO 1-5 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT2_5: Received DPT 2-5 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO2_5: Received POLIO 2-5 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT3_5: Received DPT 3-5 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO3_5: Received POLIO 3-5 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 MEASLE_5: Received MEASLES-5 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 CHIBORN5: Type of place child born-5 Value Label 1 Government hospital 2 Private hospital 3 Private home 4 Other Missing Values: 9 YFEVER5: Received Yellow Fever-5 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 [Child-6] PRENATA6: Prenatal care before birth-6 Value Label 0 No one 1 Doctor 2 Trained nurse 3 Trained midwife 4 Birth attendant 5 Other Missing Values: 9 MBREAST6: Months of breastfeeding-6 Value Label Scale 94 Never breastfed 95 Inconsistent Missing Values: 99 HCARD_6: Has health card-6 Value Label 0 No card 1 Yes, seen 2 Yes, not seen Missing Values: 9

PAGE 117

105 BCG_6: Received BCG-6 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT1_6: Received DPT 1-6 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO1_6: Received POLIO 1-6 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT2_6: Received DPT 2-6 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO2_6: Received POLIO 2-6 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 DPT3_6: Received DPT 3-6 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 POLIO3_6: Received POLIO 3-6 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 MEASLE_6: Received MEASLES-6 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9 CHIBORN6: Type of place child born-6 Value Label 1 Government hospital 2 Private hospital 3 Private home 4 Other Missing Values: 9 YFEVER6: Received Yellow Fever-6 Value Label 0 No 1 Yes 2 Mother reported 8 Dont know Missing Values: 9

PAGE 118

LIST OF REFERENCES Barth, Fredrik. 1969. Ethnic Groups and Boundaries: Social Organization of Culture Difference. Prospect Heights, Illinois: Waveland Press, Inc. Benyoussef, Amor and Albert F. Wessen. 1974. Utilization of Health Services in Developing Countries Tunisia. Social Science and Medicine, vol.8, pp.287-304. Bhopal, Raj. 1997. Is Research into Ethnicity and Health Racist, Unsound, or Important Science? British Medical Journal, vol.134, pp.1751-1756. Bicego, George T. and J. Ties Boerma. 1996. Maternal Education and Child Survival: A Comparative Study of Survey Data from 17 Countries. J. Ties Boerma (ed.), Child Survival in Developing Countries. The Netherlands: Royal Tropical Institute. Birdsall, Nancy. 1980. Population and Poverty in the Developing World. World Bank Staff Working Paper, no. 404. Washington D.C.: World Bank. Boerma, J. Ties (ed.). 1996. Child Survival in Developing Countries. Amsterdam, The Netherlands: Royal Tropical Institute. Boerma, J. Ties, Sommerfelt A.E., Rutstein S.O., and Rojas G. 1990. Immunization: Levels, Trends, and Differentials. DHS Comparative Studies No. 1. Colombia, Maryland: Institute for Resource Development. Braithwaite, Lloyd. 1960. Social Stratification and Pluralism. Vera Rubin (ed.), Annals of the New York Academy of Sciences, vol. 83, art. 5, pp.816-836. Brereton, Bridget.1979. Race Relations in Colonial Trinidad 1870-1900. Cambridge, United Kingdom: Cambridge University Press. -----------. 1981. A History of Modern Trinidad 1783-1962. London, United Kingdom: Heinemann Educational Books Inc. -----------. 1993. Social Organization and Class, Racial and Cultural Conflict in Nineteenth Century in Trinidad. Kevin A. Yelvington (ed.), Trinidad Ethnicity. Knoxville, Tennessee: University of Tennessee Press. 106

PAGE 119

107 Butz, Arlene M., Ann Funkhouser, Leila Celeb, and Beryl J. Rosenstein. 1993. Infant Health Care Utilization Predicted by Pattern of Prenatal Care. Pediatrics, vol.92, no.1, pp.50-54. Caldwell, John C. 1979. Education as a Factor in Mortality Decline: An Examination of Nigerian Data. Population Studies, vol.33, no.3, pp.395-413. Caldwell John C., Reddy P. and Caldwell P. 1983. The Social Component of Mortality Decline: An Investigation in South India Employing Alternative Methodologies. Population Studies, vol.37, no.2, pp.185-205. Caribbean Epidemiology Center. 2001. Special Programme on Sexually Transmitted Infections. URL: http://www.carec.org/programmes/std.html. August 16, 2002. Center for Ethnic Studies. 1993. Employment practices in the Public and Private Sectors in Trinidad and Tobago. Vol.1 The Public Sector. St. Augustine, Trinidad: Center for Ethnic Studies, University of the West Indies. Chaulagai, C.N. 1993. Urban Community Health Volunteers. World Health Forum, vol.14, no.1, pp.16-19. Chen, Lincoln C. 1983. Child Survival: Levels, Trends, and Determinants. Rodolfo A. Bulatao and Ronald D. Lee (eds.), Determinants of Fertility in Developing Countries, Volume 1: Supply and Demand for Children. London, United Kingdom: Academic Press. Cleland, John G. and Jerome K. van Ginneken. 1988. Maternal Education and Child Survival in Developing Countries: The Research for Pathways of Influence. Social Science and Medicine, vol.27, issue.12, pp.1357-1368. Coleman, James S. 1988. Social Capital in the Creation of Human Capital. American Journal of Sociology, vol.94, supplement, pp.S95-S120. -----------. 1990. The Foundation of Social Theory. Cambridge, United Kingdom: Belkmap. Collins, James W. Jr. and Richard J. David. 1990. The Different Effect of Traditional Risk Factors on Infant Birthweight among Blacks and Whites in Chicago. American Journal of Public Health, vol.80, pp.679-681. Cooper, Richard. 1984. A Note on the Biologic Concept of Race and Its Application in Epidemiologic Research. American Heart Journal, 108, pp.715-723. Corbie-Smith, Gisell, Elaine W. Flagg, Joyce P. Doyle, and Megan A. OBrien. 2002. Influence of Usual Sources of Care Differences by Race/Ethnicity in Receipt of Preventive Services. Journal of General Internal Medicine, vol.17, pp. 458-464.

PAGE 120

108 Cramer, James C. 1987. Social Factors and Infant Mortality: Identifying High-Risk Groups and Proximate Causes. Demography, vol.24, no.3, pp.299-321. Cuff, E.C., W.W. Sharrock, and D.W. Francis. 1998. Perspective in Sociology. Forth edition. New York, New York: Routledge. Davis, Kingsley and Wilbert E. Moore. 1945. Some Principle of Stratification. American Sociological Review, vol.10, pp.242-249. Ebanks E. Edwards. 1984. Infant and Child Mortality and Fertility: Trinidad and Tobago, Guyana and Jamaica. London, United Kingdom: World Fertility Survey. Eberstein, Isaac W. 1989. Demographic Research on Infant Mortality. Sociological Forum, vol.4, no3, pp.409-422. Eberstein, Isaac W., Charles B. Nam, and Robert A. Hummer. 1990. Infant Mortality by Cause of Death: Main and Interaction Effects. Demography, vol.27, issue 3, pp. 413-430. Echevarra, Samuel and W. Parker Frisbie. 2001. Race/Ethnic-Specific Variation in Adequacy of Prenatal Care. Social Forces, vol. 80, pp.633-654. Ewbank, Douglas C. 1994. Maternal Education and Theories of Health Behavior: A Cautionary note. Health Transition Review, vol.4, no.2, pp.215-223. Forste, Renata, Jessica Weiss, and Emily Lippincott. 2001. The Decision to Breastfeed in the United States: Does Race Matter? Pediatrics, vol.108, no.1, pp.291-296. Foster, Stanley O. 1983. Immunizable and Respiratory Diseases and Child Mortality. Population and Development Review, vol.10 (supplement), pp.119-140. Fuligni, Allison Sidle and Jeanne Brooks-Gunn. 2000. The Healthy Development of Young Children: SES Disparities, Prevention Strategies, and Policy Opportunities. Brian D. Smedley and S. Leonard Syme (eds.), Promoting Health: Intervention Strategies from Social and Behavioral Research. Washington D.C.: National Academy Press. Gee, S.C., E.S. Lee, and R.N. Forthofer. 1976. Ethnic Differentials in Neonatal and Postnatal Mortality: A Birth Cohort Analysis by a Binary Variable Multiple Regression Method. Social Biology, vol. 23, pp.317-325. Goldberg, H.I., W. Rodrigues, A.M.T. Thome, Barbara Janowitz, and L. Morris. 1984. Infant Mortality and Breast-feeding in Northern-Eastern Brazil. Population Studies, vol.38, pp.105-115. Gortmaker, Steven L. 1979. Poverty and Infant Mortality in the United States. American Sociological Review, vol.44, pp.280-297.

PAGE 121

109 Gortmaker, Steven L. and Paul H. Wise. 1997. The First Injustice: Socioeconomic Disparities, Health Services Technology, and Infant Mortality. Annual Review of Sociology, vol.23, pp.147-170. Gosine, Mahin. 1986. East Indians and Black Power in the Caribbean: The Case of Trinidad. New York, New York: Africana Research Publications. Gutman, M.C. 1999. Ethnicity, Alcohol and Acculturation. Social Science and Medicine, vol.48, pp.173-184. Harewood, Jack. 1978. Female Fertility and Family Planning in Trinidad and Tobago. Mona, Jamaica: Institute of Social and Economic Research, University of the West Indies. Harewood, Jack and Ralph M. Henry. 1985. Inequality in a Post-Colonial Society: Trinidad and Tobago 1956-1981. St. Augustine, Trinidad: University of The West Indies. Heath, Kenneth, Dana da Costa-Martinez, and Amy R. Sheon. 1988. Trinidad and Tobago Demographic and Heath Survey 1987. Port of Spain, Trinidad: Family Planning Association of Trinidad and Tobago. Helman, Cecil G. 1990. Culture, Health, and Illness (4th ed). Oxford, United Kingdom: Butterworth-Heinemann. Henry, Ralph M. 1988. The State and Income Distribution in an Independent Trinidad and Tobago. Selwyn Ryan (ed.), Trinidad and Tobago: The Independence Experience 1962-1987. St. Augustine, Trinidad: Institute of Social and Economic Research, University of West Indies. Hintzen, J.M.M. 1989. The Cost of Regime Survival. Cambridge, United Kingdom: Cambridge University Press. Hirschman, Charles and Marilyn Bulter. 1981. Trends and Differentials in Brest Feeding: An Update. Demography, vol.18, no.1, pp.39-54. Hobcraft, J.N., J.W. McDonald, and S.O. Rutstein. 1984. Socio-Economic Factors in Infant and Child Mortality: A Cross-National Comparison. Population Studies, vol.38, issue 2, pp.193-223. Hogue, Carol, James Buehler, Lilo Strauss, and Jack Smith. 1987. Overview of the National Infant Mortality Surveillance (NIMS) Project Design, Methods, Results. Public Health Reports, vol.102, pp.126-138. Huffman, Sandra L. 1984. Determinants of Breastfeeding in Developing Countries: Overview and Policy Implications. Studies in Family Planning, vol.15, no.4, pp.170-183.

PAGE 122

110 Hummer, Robert A. 1993. Racial Differentials in Infant Mortality in the U.S.: An Examination of Social and Health Determinants. Social Forces, vol.72, issue 2, pp.529-554. Hummer, Robert A., Monique Biegler, Peter B. De Turk, Douglas Forbes, W. Parker Frisbie, Ying Hong, and Starling G. Pullum. 1999. Racial Differentials in Infant Mortality in the United States. Social Forces, vol.77, issue 3, pp.1083-1118. Humphreys, Amy S., Nancy J. Thompson, and Kathleen R. Miner. 1998. Intention to Breastfeed in Low-Income Pregnant Women: The Role of Social Support and Previous Experience. Birth, vol.23, no.3, pp.169-174. Katende, Charles. 1994. Impact of Access to Health Services on Infant and Child Mortality in Rural Uganda. URL: http://www.uaps.org/journal/9/j9_4.htm. November 25, 2002. King, Gary. 1997. The Race Concept in Smoking: A Review of the Research on African Americans. Social Science and Medicine, vol.45, pp.1075-1078. Kleinman, Joel C. and Samuel S. Kessel. 1987. Racial Differences in Low Birth Weight: Trends and Risk Factors. New England Journal of Medicine, 317, pp.749-753. LeClere, Felicia B., Richard G. Rogers, and Kimberly D. Peters. 1997. Ethnicity and Mortality in the United States: Individual and Community. Social Forces, vol.76, issue 1, pp.169-198. Loustaunau, Martha O. and Elisa J. Sobo. 1997. The Cultural Context of Health, Illness, and Medicine. Westport, Connecticut: Bergin & Garvey. Mahabir, Noor Kumar. 1997. Traditional Health Beliefs and Practices of Postnatal Women in Trinidad. Ph.D. dissertation, University of Florida. Malison M.D., P.L. Sekeito, P.L. Henderson, R.V. Hawkins, S.I. Okware, and T.S. Jones. 1987. Estimating Health Service Utilization, Immunization Coverage and Childhood Mortality: A New Approach in Uganda. Bulletin of the World Health Organization, vol. 65, no.3, pp.325-330. Mangold, William D., and Eve Powell-Griner. 1991. Race and Parents and Infant Birthweight in the United States. Social Biology, 38, pp.13-27. Marshal, Ronald and Dhanayshar Mahabir. 2000. The Socio Economics of Health Care and Health Reform in Trinidad and Tobago. Ronald Marshal and Dhanayshar Mahabir (eds.), Readings in the Socio-Economics of Health Care and Health Reform in Trinidad and Tobago, Volume 1. Implications for the Wider Caribbean. St. Augustine, Trinidad: University of West Indies.

PAGE 123

111 McCann, Margaret F., Laurie S. Liskin, Phyllis T. Piotrow, Ward Rinehart, and Gordon Fox. 1981. Breastfeeding, Fertility and Family Planning. Population Reports, Series J, Number 24, p.525. Michaud, Pierre-Andr, Robert W. Blum, and Gail B. Slap. 2001. Cross-Cultural Surveys of Adolescent Health and Behavior: Progress and Problems. Social Science and Medicine, vol.53, pp.1237-1246. Moore, Dennison. 1995. Origins & Development of Racial Ideology in Trinidad: The Black View of the East Indian. Tunapuna, Trinidad: Chakra Publishing House. Mosley, W. Henry and Lincoln C. Chen. 1984. An Analytical Framework for the Study of Child Survival in Developing Countries. Population and Development Review, vol.10, Supplement, pp.25-45. Mullings, Leith, Alaka Wali, Diane McLean, Janet Mitchell, Sabiyha Prince, Deborah Thomas, and Patricia Tovar. 2001. Quantitative Methodologies and Community Participation in Examining Reproductive Experiences. Maternal and Child Health Journal, vol.5, no.2, pp.85-93. Mustapha, Nasser and Harry Singh. 2000. Socio-Demographics Variables Related to Perceptions of Health Centers in Trinidad and Tobago. Ronald Marshal and Dhanayshar Mahabir (eds.), Readings in the Socio-Economics of Health Care and Health Reform in Trinidad and Tobago, Volume 1. Implications for the Wider Caribbean. St. Augustine, Trinidad: University of West Indies. National Research Council. 1989. A Common Destiny: Blacks and American Society. Gerald David Jaynes and Robin M. Williams, Jr. (eds.). Washington, D.C.: National Academy Press. Nock, S.L. and Rossi Peter H. 1979. Household Types and Social Standing. Social Forces, vol.57, pp.1325-1345. Oakes, J. Michael and Peter H. Rossi. 2003. The Measurement of SES in Health Research: Current Practice and Steps toward a New Approach. Social Science and Medicine, vol.56, pp.769-784. Pan American Health Organization. 1998. Health in the Americas, 1998 Edition, Volume II. URL: http://www.paho.org/english/HIA1998/Trinidad.pdf. January 18, 2003. Persons, Talcott. 1940. An Analytical Approach to the Theory of Social Stratification, American Journal of Sociology, vol.45, no.6, pp.841-862. Perz, Stephen G. 1997. The Environment as a Determinant of Child Mortality Among Migrants in Frontier Areas of Par Rondnia, Brazil, 1980. Population and Environment, vol.18, no.3, pp.301-324.

PAGE 124

112 Portes, Alejandor. 1998. Social Capital: Its Origins and Applications in Modern Sociology. Annual Review of Sociology, vol.24, pp.1-24. Premdas, Ralph R. 1993. "Race, Politics, and Succession in Trinidad and Guyana." Anthony Payne and Paul Sutton (eds.), Modern Caribbean Politics. Kingston, Jamaica: Ian Randle Publishers. Preston, Samuel H. 1975. The Changing Relation between Mortality and Level of Economic Development. Population Studies, vol.29, no.2, pp.231-248. Rathwell, Thomas and David Phillips. 1986. Health, Race, and Ethnicity. Dover, New Hampshire: Croom Helm. Rogozinski, Jan. 1992. A Brief History of the Caribbean from the Arawak and the Carib to the Present. New York, New York: Penguin Books USA Inc. Rosensweig, Mark R. and T. Paul Schultz. 1982. Child Mortality and Fertility in Colombia: Individual and Community Effects. Health Policy and Education, vol.2, pp.305-348. Ryan, Alan S. 1998. The Resurgence of Breastfeeding in the United States. Pediatrics, vol.99, no.4. URL: http://www.pediatrics.org/cgi/content/full/99/4/e12. January 2, 2003. Ryan, Selwyn (ed.) 1991. Social and Occupational Stratification in Contemporary Trinidad and Tobago. St. Augustine, Trinidad: Institute of Social and Economic Research, University of The West Indies. -----------. 1999. The Jhandi & the Cross: The Clash of Cultures in Post-Creole Trinidad and Tobago. St. Augustine, Trinidad: University of The West Indies. Shen, Ce and John B. Williamson. 1997. Child Mortality, Womans Status, Economic Dependency, and State Strength: A Cross-National Study of Less Developed Countries. Social Forces, vol.76, issue 2, pp.667-700. Shiono, Patricia H. and Richard E. Behrman. 1995. Low Birth Weight: Analysis and Recommendations. The Future of Children, vol.5, pp.4-18. Shryock, Henry S., Jacob S. Siegel, and associates. 1976. The Methods and Material of Demography. Condensed edition by Edward G. Stockwell. San Diego, California: Academic Press, Inc. Singer, Burton H., and Carol D. Ryff. 2001. New Horizons in Health: An Integrative Approach. Washington, D.C.: National Academy Press. St Clair P.A., V.L Smeriglio, C.S. Alexander, and D.D. Celentano.1989. Social Network Structure and Prenatal Care Utilization. Medical Care, vol. 27, pp.823-832.

PAGE 125

113 Streatfield, Kim, Masri Singarimbun, and Ian Diamond. 1990. Demography, vol. 27, no.3, pp.447-455. Swall, Juhee V. 2001. The Main Determinants of Infant Mortality in Nepal. Social Science and Medicine, vol.53, pp.1667-1681. Tinker, Huge. 1974. A New System of Slavery: The Export of Indian Labour Overseas, 1830-1920. London, United Kingdom: Oxford University Press. Trusell, James, Laurence Grummer-Strawn, German Rodriguez, and Mark Vanlandingham. 1992. Trends and Differentials in Breastfeeding Behavior: Evidence from the WFS and DHS. Population Studies, vol.46, pp.285-307. UNICEF. 2000a. Multiple Indicator Cluster Survey Trinidad and Tobago: Full Report. URL: http://www.childinfo.org/MICS2/newreports/trinidad/trinidadtobago.PDF. February 9, 2003. -----------. 2000b. Child Survival and Health Child Nutrition, Maternal Health Water and Sanitation, Education. End Decade Database. URL:http://www childinfo.org/eddb/index.htm. February 9, 2003. United Nations. 1985. Socio-Economic Differentials in Child Mortality in Developing Countries. New York, New York: Department of International Economic and Social Affairs. United Nations Population Fund. 2000. The State of the Philippine Population Report 2000. URL: http://www.popcom.gov.ph/sppr/index.htm. January 12, 2003. Vertovec, Steven. 1992. Hindu Trinidad: Religion, Ethnicity and Socio-Economic Change. London, United Kingdom: Macmillan Education Ltd. Williams, Eric. 1962. History of the People of Trinidad and Tobago. Port of Spain, Trinidad: PNM Publishing Co., Ltd. Wood, Charles H. and Peggy A. Lovell. 1990. Indirect Measures of Child Mortality: Overview and Application to Brazil, 1980. Social Indicators Research, vol.23, pp.247-267. -----------. 1992. Racial inequality and child mortality in Brazil. Social Forces, vol.70, issue 3, pp.703-724. World Bank. 1996. Trinidad and Tobago Macroeconomic Assessment and Review of Public Sector Reform and Expenditures: The Changing Role of the State. Report No. 15187-TR. Author. Yelvington, Kevin (ed.) 1993. Trinidad Ethnicity. Knoxville, Tennessee: University of Tennessee Press.

PAGE 126

114 Yelvington, Kevin. 1995. Producing Power. Philadelphia, Pennsylvania: Temple University Press.

PAGE 127

BIOGRAPHICAL SKETCH Kuniko Chijiwa grew up in Fukuoka, Japan. She earned her first degree, a B.A. in industrial design at Tama Art University, Tokyo, Japan. She then worked for television advertising companies as a production manager. During these years she had several opportunities to work in developing countries, including the Caribbean region. In 1993, she returned to school to obtain a B.A. in social sciences at Waseda University. Her B.A. thesis was titled The Economic Development in Caribbean States: The Future of Regional Economic Co-operation and the Re-Integration of Cuba (written in Japanese). Kuniko came to the University of Florida where she earned an M.A. in Latin American Studies. In the Fall of 2001 she entered the masters program in sociology at the University of Florida. In May 2003, she earned her second M.A. Her areas of academic interest are ethnic relations and social inequality; paying particular interest to how cultural variations incorporate the social status and the quality of health in multiethnic nations. 114


Permanent Link: http://ufdc.ufl.edu/UFE0000661/00001

Material Information

Title: Locating Ethnic Context: Mother's Characteristics and Child Mortality in Trinidad and Tobago
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000661:00001

Permanent Link: http://ufdc.ufl.edu/UFE0000661/00001

Material Information

Title: Locating Ethnic Context: Mother's Characteristics and Child Mortality in Trinidad and Tobago
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000661:00001


This item has the following downloads:


Full Text












LOCATING ETHNIC CONTEXT:
MOTHER'S CHARACTERISTICS AND CHILD MORTALITY
IN TRINIDAD AND TOBAGO
















By

KUNIKO CHIJIWA


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS

UNIVERSITY OF FLORIDA


2003

































Copyright 2003

by

Kuniko Chijiwa


































To my brother and my sister















ACKNOWLEDGMENTS

My deepest appreciation first goes to my committee members, Dr. Barbara

Zsembik and Dr. Charles Wood. Thanks go to my supervisor Dr. Zsembik for being

instrumental in the development of all aspects of this study. She opened my eyes to vital

issues in studying ethnicity in relation to health and quality of life. Especially, she led me

to the world of data philosophy. Dr. Wood originally stimulated my interests in data

analysis and its application to racial and ethnic studies. He also facilitated the basis of this

study. I especially owe him for his patience in teaching me the enjoyment of data

analysis.

I also owe a debt to Trinidadians following my initial research in Trinidad for my

thesis in Latin American Studies. This time, I especially owe Dr. Beni N. Balkaran of the

Mt. Hope Hospital, Dr. Robert Lee of the Caribbean Epidemiology Center, and Dr.

Victor Coombs for providing information on children's and mothers' health and child

mortality; and Ms. Elizabeth Welsh of the Ministry of Health and Ms. Raynette Pierre of

the Central Statistical Office for providing vital statistics. My special thanks go to Rajnie

Ramlakhan, Esther Langoo and her family members for their sincere and everlasting

friendship, which nourished me richly while I was in Trinidad.

I would like to express my special gratitude to Macro International Inc. for

allowing me to use the dataset from the Demographic and Health Survey, Trinidad and

Tobago 1987. This survey established the framework of this thesis.









Finally, I am very grateful to my family who supported me financially and

emotionally, understood me, and allowed me to be selfish. I thank my best friend for

providing constant moral support and encouragement; and my cats who reminded me to

relax from time to time.
















TABLE OF CONTENTS

Page

A C K N O W L E D G M E N T S ................................................................................................. iv

T A B L E O F C O N T E N T S ................................................. ............................................ vi

LIST OF TABLES ........................ ......................... .. .. ............... .... ix

LIST OF FIGURES ............................... ... ...... ... ................. .x

ABSTRACT .............. .......................................... xi

CHAPTER

1 IN TR OD U CTION ............................................... .. ......................... ..

B ack g rou n d .................. .............................................................................. 2
Significance of the Study of Child M ortality ......................................................2
Ethnic Relations, Social Allocation, and Study of Child Mortality in TT ............4
Practical Significance ................. ......... ....... .... ..... ............... 6
Theoretical Significance ................. .................. ............ ............... .............. 11
Conceptual Framework for Child Survival............... .............................................. 15

2 L ITE R A TU R E L E V IE W ............................................ ......................................... 19

Trinidad and Tobago....................... ........ .. .................. .............. .. ......19
Defining Differences between Ethnicity and Race ..........................................19
Ethnic Context in Trinidad and Tobago ................................... .................22
Slavery to collective identity............................................... .................. 23
Ethnicity and class consciousness .... .......... .......................................27
Properties of Child M ortality................................................ ... ...... .......... 31
Difference between Infant Mortality and Child Mortality ................................31
Determ inants of Child M ortality ......................................................... 33
Socioeconomic and Demographic Variables on Child Mortality......................34
Incom e ............................................................... .. ... ......... 34
M maternal education ............... .... ................ ...................................... .. ..35
Marital status and residential characteristics............... .....................35
Intervening Health Care Variables in Relation to Socioeconomic Variables .....37
P ren atal c are ................................................................ .. 3 8
Preventive health care ............................................................................ 39









B reastfeed in g .................................................................. 4 0
Type of place where the child is born .....................................................42
Infant and Child Mortality in Trinidad and Tobago................ ... ............ 44
H y p oth eses............................. ........................................................... ............... 4 7

3 RESEARCH DESIGN AND METHODS ..................... ....................................49

D ata and Sam ple Size A nalyses ........................................ ........................... 49
M e a su re s ............................................................................................................... 5 2
C h ild M mortality ..............................................................53
D em graphic Factors............ ............................................. .......... ............. 55
E ethnicity ....................... ...................... .. .. ... ........ ....... .. 55
Type of place of residence .................................. ............... ....................55
M arital statu s ................................................................................. 56
Socioeconom ic Factors ......................................................... ............... 57
M aternal educational attainment .............. ............................................. 57
Q quality of life ....................................................................................... .. 57
H health R elated Factors ............................................... ............................. 60
P ren atal c are ................................................... ................ 6 1
Type of place where the child is born ................. ............ 61
Quality of preventive health care for child.............................. ............... 62
B reastfeeding ...................................................................... ............... 64
P procedures of D ata A nalysis.......................................................................... .... 65

4 D A T A A N A L Y SIS .......................................................................... ....................67

Ethnic Differentials in Child Mortality and Its Determinants ....................................67
Demographic Characteristics.................... ....... ........................... 67
Socioeconomic Characteristics.................... ....... .......................... 70
Child Health Care Practices................................. ........................... 71
Influence of Maternal Characteristics on Child Mortality............... ...................73
Influence of Demographic and Socioeconomic Factors.............................. 73
Health-related Proximate Factors ..... .................... ...............76
Multivariate Logistic Regression Models .............. ..........................................79
Ethnic Influence on Child Mortality............... ...... ...............79
Urban-Rural Setting, Marital Status, and Scio-economic Influences..................80
Influence of Health-related Factors on Child Mortality .............................. 82

5 C O N C L U S IO N ............................................................................................. .. 84

Ethnicity in the Analysis of Child M ortality ................................... .................84
Locating Ethnic Context in Trinidad and Tobago........... .....................................90

LIST OF VARIABLES AND VALUES OF THE DEMOGRAPHIC AND HEALTH
SURVEY IN TRINIDAD AND TOBAGO 1987 DATA .......................................97









L IST O F R E FE R E N C E S ......................................................................... ................... 106

BIOGRAPHICAL SKETCH ............................................................. ..................114
















LIST OF TABLES


Table page

1-1 Cause of infant m ortality, 1988 .... ......................... ............... ....... ...............

1-2 Cause of child m ortality, 1988 ............................................................................7

1-3 Causes of infant m ortality, 1997 ...................... .............................. ............... 8

1-4 Cause of child m ortality, 1997 .......................................................................

3-1 Distribution of women 15 to 49 by ethnic and type of place of residence in 1987
and 1990, Trinidad and Tobago ........................................ .......................... 50

3-2 V variable description s........................................................................ ... ............... 54

3-3 Rotated component matrix for 8 variables of household composition...................59

4-1 Characteristics of mothers with children born in last 10 years by ethnicity ...........68

4-2 Proportions and odds for mothers who have lost at least one child and correlations
between ethnicity and maternal characteristics......................................................75

4-3 Probability of having lost at least one child controlling for demographic,
socioeconomic, and health care factors (Logistic Regression) ................................81















LIST OF FIGURES


Figure page

1 Conceptual framework for the analysis of the effects of socio-economic and socio-
cultural factors on child survival ....................................................... ................ ... 17

2 Infant and under 5 mortality in Trinidad and Tobago and Latin American and
C aribbean, 1960 to 2000 ................................................ .............................. 44















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Arts

LOCATING ETHNIC CONTEXT:
MOTHER'S CHARACTERISTICS AND CHILD MORTALITY
IN TRINIDAD AND TOBAGO

By

Kuniko Chijiwa

May 2003

Chair: Barbara A. Zsembik
Major Department: Sociology

The objective of this study is to examine ethnic differentials in child mortality in

Trinidad and Tobago. Child mortality is considered as an outcome of mother's

demographic and socio-economic characteristics and quality of health care for a child.

The discriminatory perceptions that incite ethnic conflicts over socioeconomic and

political allocations between the two major ethnic groups, African and East Indian, were

contextualized within Trinibagonian history reflecting the conditions of social inequality

experienced by each ethnic group. In this study, child mortality serves as a variable to

ascertain the socio-economic and cultural differences between African and East Indian

for capturing a unique social stratification system in this ethnically polarized society.

The Demographic and Health Survey in Trinidad and Tobago is employed to

determine whether ethnicity differentiates child mortality. The data analysis consists of

multivariate logistic regression models using nine explanatory variables that are divided

into three clusters; demographic factors, socioeconomic factors, and health care factors.









The analysis is organized in order to empirically test hypotheses that are derived

from theoretical perspectives of Coleman's social system considering the child

survivorship framework originated by Mosley and Chen. Coleman's social system theory

is concerned with the balance and dynamics of competing interests in and controls over

scarce resources between groups, which contribute to the construction of the social

structure. In Trinidad and Tobago, ethnicity is considered as a major motive to pursue the

common social and economic benefits through which we can locate the status of

individuals and groups in the social structure.

Logistic regression analysis on child mortality indicates that after controlling for all

demographic, socioeconomic, and health care factors, ethnicity is statistically significant;

and confirms that African mothers have more than twice the higher risk of child loss that

East Indian mothers have. The beneficial impacts of health care factors are found. No

interaction terms between ethnicity and the three health care factors are significant, which

means that each health care factor works independently from ethnicity. Thus, better

health care associates with decrease of child mortality universally for all women in this

nation. The inclusion of health care factors, however, widens the child mortality gap

between them, implying that the ethnic gap in child mortality cannot be attributed to

health care factors. Therefore, Africans continue to be disadvantaged in child mortality.

There must be continued efforts toward improving the socio-economic status and quality

of health care of Africans. The analysis also implies that there may be misspecifications

beyond factors associated with socioeconomic standings and quality of health care that

may be culturally influencing their health practices and behaviors.














CHAPTER 1
INTRODUCTION

The objective of this study is to examine ethnic differentials in child mortality in

the Republic of Trinidad and Tobago (henceforth called TT). Child mortality is

considered as an outcome of mother's overall quality of life and as a variable, which can

be used as an analytical category to ascertain economic and cultural differences between

the two major ethnic groups, African and East Indian. This thesis utilizes a cross-

sectional survey, the Demographic and Health Survey in Trinidad and Tobago, 1987

(henceforth called the TTDHS), to determine whether ethnic backgrounds differentiate

child mortality among mothers in TT.

In light of the social and economic context, the study focuses on specifying the

different dimensions of the concept "social inequality" and on analyzing the way in

which various factors affect child survivorship. If we control for the socioeconomic

determinants and health care factors of child mortality, the question raised is, whether the

children of women who belong to the disadvantaged ethnic group in child mortality

continue to experience higher death rates than those born to the advantaged ethnic group

in child mortality. In the case where ethnic background continues to be statistically

significant after controlling for key social economic factors and health care factors, the

results may suggest that the higher child mortality group is subject to additional

disadvantages beyond factors associated with socioeconomic standings and health care

factors; there may be cultural elements influencing their health practices and behaviors.









For the objective of this study, the following specific aims were set:

* to determine whether the ethnic background of the mother significantly influences the
probability of child mortality;

* to assess whether ethnic background differentiates the association between
socioeconomic status and child mortality;

* to assess whether ethnic background differentiates the association between health care
practices and child mortality; and

* to measure the overall magnitude of ethnic influence on child mortality in TT.

Achieving these aims may improve our understanding of the characteristics of

ethnic differences in well-being and health in TT. In the multi-ethnic communities within

a nation like TT, it is important to improve and renovate health care systems taking into

consideration cultural and behavioral characteristics in order to reduce the inequalities in

quality of health. The quality of health care for a particular ethnic group may be largely

influenced by socioeconomic status, by accessibility to facilities, and by availability of

the kind of medical care the ethnic group prefers.

Background

Significance of the Study of Child Mortality

Studies of race and ethnic differences in health have attracted the interests of

researchers in analyzing health-related needs and problems in order to provide necessary

and accessible health services for people from various ethnic backgrounds due to the

changing population composition. Racial or ethnic background is a social construct in

any kind of society and is associated with mother's absolute health status during the

perinatal period, which may directly or indirectly affects fetuses, neonates, infants, and

children. Because some ethnic groups are comparatively more likely to fall into

disadvantaged social categories, in that they are more likely to be poor, unmarried, and









less educated, they will have higher mortality which implies that observed mortality

differences are primarily compositional in nature (LeClere et al. 1997). Because of the

extreme dependency, children's health and survival chances rely on their parents.

Therefore, child mortality is considered as an important indicator of the well-being of a

population (Birdsall 1980, Eberstein 1989, Wood and Lovell 1992, Hummer et al. 1999).

Quality of child health care and maternal education as indices for socioeconomic

status have been found to be highly correlated with infant and child health and mortality

(Cramer 1987, Hogue et al. 1987, Mangold and Powell-Griner 1991). Income, which is

closely related to occupation and education, exercises an important effect on the ability to

obtain medical provisions (Hobcraft, et al. 1984). Thus income is an influential factor in

consideration of individual socioeconomic status and is an important determinant of child

mortality (Gortmaker 1979, Mosley and Chen 1984, Hummer 1993). A higher incidence

of poor pregnancy outcomes and child mortality among women from disadvantaged

socioeconomic backgrounds has been indicated in previous research on determinants of

child mortality.

Socioeconomic variables also serve as indicators of knowledge and make-up of

medical services. Therefore, it is likely that people of higher socioeconomic status groups

will be better able not only to afford drugs and other expensive care but also have access

to valuable information on pregnancy and child bearing. They are also likely to have

better housing and are more likely to be connected to water supplies and sewage systems.

Education expands the role of the socioeconomic variable by disseminating knowledge

on medical sanitary requirements. This knowledge can range from simple elements of

child health care involving cleanliness and sterilization to more complex knowledge of









what drugs and vaccinations are required; and the ability to find and use services

(Hobcraft et al. 1984). Hence, socioeconomic status, specifically income, can be

considered as a pivot of human resources, however, the mere existence of differentials by

income, does not constitute a satisfactory explanation of the association between these

income differences and child mortality (Gortmaker 1979).

Because of the situation of children, especially new born babies, depending entirely

on their parents, child morality is considered as a sensitive measure of their parents'

quality of life associated with different socioeconomic environments. Child mortality is

often regarded of an extreme case of poor child health, which mirrors the parents' life, in

that it is an aggregate of realities derived from the parent's circumstances and everyday

choices. Although child mortality has been reduced by substantial improvements in

parents' standard of living, educational attainment, and access to medical care during the

recent decades, differences persist. There are a variety of latent factors related to child

mortality. To reduce such inequality within a multi-cultural society, we need a better

understanding of the mechanisms of the relations between ethnicity and mothers'

characteristics; and of the possible reasons why specific ethnic groups continue

experiencing lower quality of life and higher mortality. This understanding will help

provide general health services to entire nations and the accessible health services to

specific ethnic groups.

Ethnic Relations, Social Allocation, and Study of Child Mortality in TT

The southernmost island formation in the Caribbean Archipelago, TT is the nation

composed of twin islands. An oil- and gas-rich republic of 1.16 million people, TT is

evenly split along ethnic lines with slightly less than half its people of East Indian descent

(40.3%) and the same number of African descent (39.6%). The distinct ethnic identity of









each has created a modern culturally bi-polarized society within the nation. The political

and economic rivalry between them has been extensively studied by researchers. Many of

them found evidence of differences between the two ethnic groups in terms of their

aspirations, income distribution, social mobility, political behavior, and occupational

placement -- each of which has been formed in its history of development (Harewood and

Henry 1985, Henry 1988, Selwyn Ryan 1991 1999, Center for Ethnic Studies 1993,

Yelvington 1995). While such differences in economic superiority and social mobility

between the two groups have attracted a great deal of scholarly attention, differences in

health care practices derived from or largely influenced by their respective cultures which

encourage or restrain their acceptance of modem medical practices, have not been well

studied.

Accessibility to health care services is concerned with allocational issues within a

nation, since it is considered both a universal right for the entire nation and a privilege for

a people who can afford a considerably higher quality of medical care for severe injury

and prolonged illness. Health care is a fundamental right when one considers that

societies have a duty to preserve life and also promote quality of life. Health care is a

combination of intertwined social structure, which reflects economic, social and political

inequalities, and the allocation of scarce resources. Health is a condition of physical,

mental and emotional well-being (Fuligni and Brooks-Gunn 2000, Singer and Ryff 2001)

that all people should enjoy, hence, health should not be a privilege for the few, but like

education, should be universal; everyone should have access to it, regardless of age,

religion, ethnicity, culture, nationality, or social class. In this respect, health care assumes

social and economic importance as various groups within society jostle for access to this









vital resource, which inevitably appears to be "scarce." However, the right to something

that is a scarce commodity in many developing countries is what creates dissatisfaction

and tension among social collectives (Marshall and Mahabir 2000).

A study of health requires investigating a complex combination of socioeconomic

and cultural characteristics; physical, practical, and behavioral. The cultural variations

can significantly predict child survivorship, which is disaggregated by social indicators,

such as place of residence, household income, and parents' level of educational

attainment. Cultural variation may characterize the socioeconomic differences between

the two major ethnic groups by examining the differences in child mortality. The results

from these efficient indicators, which have been historically established by interactions

between the two ethnic groups, provide valuable insights into an ethnically polarized

society's stratification system.

Practical Significance

Table 1 presents the causes of infant mortality and child mortality in 1988 and

1997. In both 1988 and 1997, any incidence that occurs during the perinatal period was

ranked the number 1 cause for infant mortality, followed by congenital anomaly and

pneumonia in 1988, infectious and parasitic disease in 1997. The trend of both the rates

for infant mortality and child mortality, which have continuously shown a downward

curve since 1950s, appears to be reaching a low point. Table 1-1 and 1-3 provide list of

cause of infant mortality which show that problems during the perinatal period remain the

top cause; moreover, the problem gains strength as a driving factor behind infant

mortality: 62.8% in 1988 and 69.6% in 1997.

Informal interviews with medical doctors and epidemiological researchers in

Trinidad, summer 2002, reveal current major concerns of infant mortality in TT in terms












Table 1-1. Cause of infant mortality, 1988

Causes of Death Male Female Total

Certain Conditions Originating in the Perinatal Period 127 81 208
Congenital Anomalies 24 24 48
Pneumonia 17 7 24
Infectious and Parasitic Diseases 7 4 11
Signs, Symptoms, and Ill-defined Conditions 6 4 10
Injury and Poisoning 3 7 10
Accidents and Adverse Effects 1 7 8
Diseases of the Circulatory System 3 2 5
Other Protein-Calorie Malnutrition 2 0 2
Malignant Neoplasms (Leukaemia) 0 1 1
Nutritional Marasmus 1 0 1
Anaemias 1 0 1
Influenza 0 1 1
Appendicitis 0 1 1
Total 192 139 331
Source: Deaths Report 1988. Central Statistical Office, 1990.




Table 1-2. Cause of child mortality, 1988

Causes of Death Male Female Total

Injury and Poisoning 13 9 22
Accidents and Adverse Effects 11 9 20
Congenital Anomalies 2 10 12
Infectious and Parasitic Diseases 7 3 10
Diseases of the Circulatory System 5 1 6
Pneumonia 4 2 6
Malignant Neoplasms (Leukaemia) 2 3 5
Anaemias 3 3
Bronchitis, Emphysema, and Asthma 1 2 3
Signs, Symptoms, and Ill-defined Conditions 2 1 3
Other Protein-Calorie Malnutrition 2 2
Meningitis 1 1 2
Certain Conditions Originating in the Perinatal Period 1 1
Total 51 44 95
Source: Deaths Report 1988. Central Statistical Office, 1990.












Table 1-3. Causes of infant mortality, 1997

Causes of Death Male Female Both

Certain Conditions Originating in the Perinatal Period 119 96 215
Congenital Anomalies 25 23 48
Infectious and Parasitic Diseases 4 7 11
Pneumonia 6 5 11
Injury and Poisoning 3 3 6
Accidents and Adverse Effects 3 3 6
Signs, Symptoms, and Ill-defined Conditions 3 2 5
Diseases of the Circulatory System 1 2 3
Malignant Neoplasms (Leukemia) 1 1
Diabetes Mellitus 1 1
Nutritional Marasmus 1 1
Chronic Liver Disease and Cirrhosis 1 1
Total 166 143 309
Source: Population and Vital Statistics Report 1997. Central Statistical Office, 1990.






Table 1-4. Cause of child mortality, 1997

Causes of Death Male Female Both

Injury and Poisoning 8 4 12
Congenital Anomalies 5 4 9
Accidents and Adverse Effects 6 3 9
Infectious and Parasitic Diseases 5 2 7
Malignant Neoplasms (Leukemia) 3 4 7
Diseases of the Circulatory System 1 3 4
Pneumonia 1 3 4
Signs, Symptoms, and Ill-defined Conditions 2 2
Homicide 1 1 2
Meningitis 1 1
Bronchitis, Emphysema, and Asthma 1 1
Direct Obstetric Deaths -1 1
Total 32 27 59
Source: Population and Vital Statistics Report 1997. Central Statistical Office, 1990.









of ethnic differences. One concern was the fundamental transformation of dietary habits,

which has caused serious health related problems, especially the potential for diabetes

among women during pregnancy. The other is sexual activity, which has played a major

role in increasing HIV/AIDS cases in TT. Recent research conducted by the Caribbean

Epidemiology Centre reported that approximately 3% of new born babies have been

infected with HIV/AIDS at birth and some 5% of the causes of death among children are

related to HIV/AIDS. Mother-to-child transmission of HIV/AIDS has become a major

problem in TT with up to three infants being infected everyday, via this route, assuming

an HIV prevalence of between 2 and 3 % among pregnant women. An estimated 1,806

adults and children were newly infected with HIV/AIDS during 2000 (Caribbean

Epidemiology Center 2001). These facts are not temporal social phenomena, but rather

represent the inherent socio-cultural differences between the two ethnic groups. It is

believed rapid dietary habit transitions are largely due to the Americanization of food

culture that affects especially the people of East Indian descent. Higher prevalence of

HIV/AIDS appears among the people of African descent.

Due to the relatively lower infant and child mortality rate, the study of infant and

child mortality has drawn little attention in TT. Few, if any, comparative studies on

ethnic differences in terms of infant and child mortality in TT have been conducted.

Although the TTDHS collected ethnic data, researchers have not attempted to use

ethnicity for determining differences on health issues between African and East Indian

women. The national statistical data on infant and child mortality comparing ethnic

groups does not even exist, despite the fact that ethnic issues and "allocations" in every

term are always prime interests among Trinibagonians. TT has a relatively more









prosperous economic environment compared to other Caribbean nations. Oil production

has in fact contributed to facilitating development of medical institutions and health

development programs thus, lowering the infant and child mortality rate. With the pride

of not being a developing county, the TT government no longer refers child mortality as

an indicator of the level of quality of life among the nation, instead, they aim to expand

and upgrade the public health facilities in delivering high quality health care to every

citizen. As a small and oil-rich island, TT nations shall achieve their aim in the near

future. However, infant and child mortality rates are widely regarded among researchers

as variable indicators of the physical well-being of children as much literature has

insisted (Birdsall 1980, Eberstein 1989, Wood and Lovell 1992, Hummer et al. 1999).

To understand the differences in child survival associated with ethnicity is

important for health policies and interventions. These differentials identify the highest

risk group across ethnic groups, indicating the need to overcome the socioeconomic

inequalities between ethnic groups, and the need for development health programs to take

the inequalities into consideration. Given the numerous issues associated with ethnic

strife over socioeconomic and political allocations, health care "allocation" is not

exception to this; health care services should be considered a limited package of

resources. Curiously, ethnic context has rarely appeared as one of the features of the

country profile in research on child health and mortality (Harewood 1978, Heath et al.

1988, Marshall and Mahabir 2000, UNICEF 2003) as already mentioned. Hence this

study will be a springboard for advancing our awareness of ethnic differences in terms of

health, health care, and child mortality as accumulations of and confounding effects of









socioeconomic and cultural factors in TT, and also, for more complete research on the

association between ethnicity, child mortality, health behaviors, and quality of life.

Theoretical Significance

In the extant research of health and child mortality, socioeconomic status has been

used extensively as an explanatory variable that typically measures the extent to which

socioeconomic background is related with health. It is also used as a control variable in

looking at other correlates of child health and mortality. Hence, taking into consideration

the significance of socioeconomic influence on child mortality, and for the purpose of

determining whether ethnic background overwhelms the relationships between child

mortality and socioeconomic level, social stratification theory is considered first as

providing theoretical guidance.

The concept of social structure, social class, and socioeconomic status are central to

the study of child mortality in social sciences. The theorists of the stratification school

take a structural-functional approach. While structural functionalism considers a society

composed of interdependent elements such as culture, personalities, and social systems,

the theory of social stratification gives more weight to socioeconomic activities than to

culture and personalities. Structural-functional analysis of social stratification is

concerned primarily with the roles played by such socioeconomic activities, which

maintain social structure. Heavy rewards in valued goods are given to motivate

individuals to perform important social functions, with the heaviest reward being given to

those occupying positions of functional importance in the society for which qualifications

in the society were relatively rare (Persons 1940, Davis and Moore 1944). Although

functional stratification theorists take cultural variations and personalities into

consideration in the frame of "functions," they mainly hold that the cohesive and









integrative power of socioeconomic class linkages "horizontal lines" as it were -

surpasses the divisive power of the vertical lines, which divide one ethnic group from

another (Braithwaite 1960). Structural functionalists insist that there must be a certain

minimum of common shared values if the unity of the society is to be maintained. Hence,

a structural-functional approach suggests that after controlling for social strata as

measured by education and economic status, ethnic differences will not be statistically

significant. Therefore, structural functionalism, in the same manner as traditional social

stratification theory, chiefly attempts to describe the structure of social stratification

based on the differences among people in terms of such criteria as wealth, income,

occupation, education, descent, property, and prestige, and to specify the processes by

which the social system is generated and maintained (Cuff et al. 1998). In their view,

social system can be held together by a consensus on economic norms and values in spite

of distinct cultural and ethnic diversity.

It is important that child mortality research involves investigating how levels of

inequality and variation in social context affect health outcomes. Also, in multi-ethnic

society, socioeconomic measurements may need to capture more of the social context

than the indices of income, education or occupational position can provide. Social context

is derived from such factors as community, networks, and environment that child

mortality research appears most interested. The variables of socioeconomic status in

explaining the difference of child health outcome, and child mortality may be described

more inclusively when they involve the social context influenced by cultural context, no

matter what the degree of influence is. In this sense, a structural-functional approach has

the same weakness as traditional stratification theory; both overlook cultural norms in









quality of life and quality of health. Hence, the traditional stratification theory and the

structural functional approach are not very useful in explaining the societies composed of

many ethnic groups who devise their empirical measures either by determining the

distributional characteristics of social stratification systems, or by identifying the

positions of individuals, families, or other social groups in such systems (Oakes and

Rossi 2003).

Culture affects our perceptions and experiences of health and health care in many

ways. Health care within any group can be affected by a multitude of cultural variables;

some very basic, some more complex. It is a measure of human flexibility with diverse

ways and means of meeting human needs (Loustaunau and Sobo 1997). Hence, Oakes

and Rossi suggest that it is better to start with the question, what would be an ideal

socioeconomic status measure. Such a measure is described by Nock and Rossi (1979),

cited by Oakes and Rossi (2003: 7-8) that, "socioeconomic status is that dimension of

stratification which translates the objective distribution of social resources into

meaningful perceptions of relative desirability." This concept holds that ethnic and

cultural collectivity are elements of diversifying ways by which people share and

distinguish the perceptions and meanings, and means of meeting human needs related to

health and "well-being."

In reality, measurements of socioeconomic status are almost entirely represented by

education and income, as well as occupational position, which are obtained from census

type data due to its availability. Therefore, it may be useful to consider Coleman's social

system theory defining "the value and the role of social capital in the creation of human

capital (1988, 1990)," which is also based on social stratification theory. The central idea









of Coleman's notion concerns how positions constituting social structure emerge, and

how persons are motivated to occupy such positions.

Coleman sets forth three ideas. The first is material capital, which refers to owned

materials such as household composition and income that are tangible and analyzable.

The next is human capital, which refers to inherited physical appearance and ability as

well as education, skills, abilities, and knowledge one may acquire with one's investment.

The last is social capital, which includes obligations to and from others, information

channels, norms, and reputation effects. To possess social capital, a person must be

related to others, and it is in the potential of those relationships where social capital lies

(Oakes and Rossi 2003). Portes (1998) explains that, "there is growing consensus that

social capital stands for the ability of actors to secure benefits by virtue of membership in

social networks and other social structures." Coleman emphasizes these networks and

functions as a necessary condition for the rational action paradigm:

Just as physical capital and human capital facilitate productive activity, social
capital does as well. For example, a group within which there is extensive
trustworthiness and extensive trust is able to accomplish much more than a
comparable group without that trustworthiness and trust (1988: S101).

The trustworthiness, in his term "the role of closure," refers to obligations, expectations,

and social norms, which are largely influenced by a person's cultural perception and his

experience. The existence of sufficient ties among a certain number of people guarantees

the observance of norms. The collective perception of inequality emerges from the

balance and dynamics of interests and control over scarce resources. For understanding

the interrelationship between the socioeconomic structure and ethnicity in a multi-cultural

nation, it is meaningful to consider the ethnic differences in the value of social capital and

its role in the creation of human capital within a group. Oakes and Rossi indicate;









There are several advantages to incorporate social capital into a measure of SES. It
provides an understanding of the variation in social contexts. --- And social capital
assists in understanding the all important micro-macro (man to structure and
structure to man) transitions, and thus family and neighborhood and institutional
level impacts and outcomes (2003: 777).

Thus, social capital can provide a mechanism through which behavioral norms are

generated and maintained, and can promise to provide a link between individuals,

society, and health as a human capital.

In previous studies, the influence of socioeconomic status on child mortality is

significant (Gortmaker 1979, Mosley and Chen 1984, Cramer 1987, Hogue et al. 1987,

Mangold and Powell-Griner 1991, Hummer 1993), and the mother's health care practices

for her child such as prenatal care, breastfeeding, and immunization have a significant

impact as intervening factors (Rosensweig and Schultz 1982, Goldberg et al. 1984,

Huffman 1984, Maison et al. 1987, Trussell et al. 1991, Kadende 1994, Chaulagai 1993,

Humphreys et al. 1998, Alan Ryan 1998, Forste 2001). Coleman's social system theory

can provide a theoretical framework for the linkage between health and behavioral

norms, individual perceptions, and social collectivities. It is also appropriate in guiding

the examination of whether or not ethnic identity overwhelms the associations between

child mortality and the two clusters of predictive variables, socioeconomic and health

care factors, and in defining the stratification system of TT society.

Conceptual Framework for Child Survival

The previous research provides evidence that child mortality is an indicator of

mother's social well-being. Ultimately mother health outcomes are enmeshed in a web of

causality. To comprehend the relationships between child mortality and various

confounding factors, ideally, we would wish to have child specific information related to

health status; illness, nutritional inputs and growth that would allow controls for health









heterogeneity in TT. Unfortunately, census type data generally does not include such

information. Therefore it is useful to have a framework conceptualizing how we

understand the connections among the proximate factors, socioeconomic factors

(education, household composition and housing quality), demographic factors (place of

residence and marital status), and health related factors (record of immunizations for

children, prenatal care, place of child born, breastfeeding) which we can obtain from the

TTDHS related to child survival.

The framework for the study of child survival in developing countries, which has

had a major influence on the Demographic Health Survey (Boerma 1996), was first

presented by Chen in 1983 and developed by Mosley and Chen in 1984. In the Mosley-

Chen framework, a set of proximate or intermediate determinants, which directly link to

the risk of child morbidity and mortality, are divided into five socioeconomic factors:

material factors, environmental contamination with infectious agents, availability of

nutrients to the fetus and infant, injuries, and personal illness control (Mosley and Chen

1984). All social and economic determinants, such as mother's education and household

income, operate through the proximate determinants to affect child growth and mortality.

Proximate determinant framework, however, has met criticism from researchers because

it is more likely to lead to research focusing on individual level decision-making rather

than on broader society processes as a result of its complex web of factors influencing

behavior or analyses at other levels such as family and community (Ewbank 1994,

Boerma 1996).

Since the major interest of this study is to determine the relationship between ethnic

background and child mortality as a reflection of mothers' socioeconomic status, the














SOCIOECONOMIC
FACTORS

Place of Residence
Parental Education
Household Income
Marital Status
,.............. ...... ..............................










SOCIO-CULTURAL
FACTORS

Family Tie
Community
Tradition Belief
Habit
Support


HEALTH STATUS FACTORS


I INTERMEDIATE FACTORS

Health Service Accessibility

Living Standard
-oi (Infrastructure)
Job Situation
Household Composition
Housing Quality
.......... .... ... .... ... .... ...... ... ...

Personal Illness Control
"* Hygiene
..................... .......................................................


Perinatal Health Care
Mother's Health Status
Prenatal Care
Maternal Age and Parity
Birth Intervals
Fetal
Biological Factors
Down's Syndrome / Defects
Nutritional Factors
Maternal Nutrition
Diet During Pregnancy
Infection Factors
Infection before/during Pregnancy



Infan/Child
Nutritional Factors
Breastfeeding
Supplementation MORTALITY
Food Intake
Infection Factors
Diarrhea
Respiratory Infections SURVIVAL
Preventive Health Care


Figure 1. Conceptual framework for the analysis of the effects of socio-economic and socio-cultural factors on child survival









conceptual and analytical framework of this study needs some modifications with

inclusion of factors rooted in their ethnic background. Figure 1 presents a broader

conceptual framework. Socio-cultural factors, which are associated with ethnic

background, serve as key covariates influencing or operating on a mother and child's

health care and standards of living at both the individual and community levels. In the

case that ethnic differences are not statistically significant in the presence of

socioeconomic determinants and health care interventions of child mortality, then, as the

United Nations reports, "the variations in mortality across ethnic and race groups

probably reflect mainly differences in such factors as socioeconomic status and

accessibility of health facilities and services, rather than innate differences among the

groups themselves" (1985: 77). This account can be understood in the way that

differences in socioeconomic status and accessibility of health facilities and quality of

health services may be nested in ethnic background. If racial differences in child

mortality appear, and they are statistically significant after controlling for socioeconomic

factors and health care factors, then the higher child mortality ethnic group is subject to

additional disadvantages beyond those associated with socioeconomic standings and

quality of health care.














CHAPTER 2
LITERATURE LEVIEW

This chapter presents a review of the literature that is relevant to this study of the

relationships of child mortality and ethnic background. The first section of the chapter

provides an overview of the definitions of race and ethnicity in the study of health and

child mortality. The characteristics of ethnic relations in TT are reviewed through

historical transformation of the islands population. TT has established segments of ethnic

divisions on which the basis of national arguments over social, economic, and political

allocations are laid. The second section reviews the prior studies of child mortality with

emphasis on the significant roles of socioeconomic factors influencing other factors,

followed by the vital statistics including child mortality rates in TT.

Trinidad and Tobago

Defining Differences between Ethnicity and Race

A vast literature on health, infant and child mortality, and child survivorship has

used either of two terms "ethnicity" and "race," or both terms. It is important here to

clarify the terms of race and ethnicity in the study of child mortality. Race is

characterized primarily by phenotypic features but has been used to imply genetic or

biological bases of health behavior or outcome (King 1997, Gutman 1999). Much

research has focused upon studying racial differences between whites and blacks;

however, usage of "race" in studying health has decreased with the recognition that

genetic differences are greater among individuals within a given racial group than









between racial groups (Michaud et al. 2001). Cooper, an epidemiologist, conceptualized

race in the study of public health and medicine;

In the biologic sense, there are not such things as races. ---. The appearance of a
highly consistent pattern of differential mortality between races can be ascribed
only to environmental (i.e., social), not genetic, factors. The concept race itself is a
social category. Whether it be [sic] Catholic in Ulster, Jew in Germany, Tamil in
Sri Lanka, or blacks in the United States, the definition of a population subgroup is
a result of economic and historical, not evolutionary, development. Health status of
racial group should be viewed within this context (1984: 722).

For Cooper, although the character of the health disadvantage particular to a racial group,

e.g., higher risk of coronary heart disease for African-American, may evolve, the

disadvantage itself is not likely to diminish until the intensity of racial discrimination is

successfully reduced.

Ethnicity, on the other hand, is a common set of practices, values, and beliefs held

by a collective and transmitted from one generation to the next (Helman 1990, Bhopal

1997). Barth indicated that ethnicity is a form of social organization and a fundamental

means of ordering social life; one that relies on manipulating "cultural traits" and ideas

about origin so as to communicate difference. Ethnic definitions are based on ascription

and self-ascription-manipulation of identities and their "situational" character (Barth

1969).

For this study, the term "race" would be inappropriate to distinguish collectives in

the contemporary TT society. The island was populated with three distinct immigrant

races; Europeans, Africans, and East Indians. Europeans and Africans were differentiated

in terms of color; however, no conventional color correspondence is assigned to East

Indians. There is no inherent affinity between people sharing a common racial identity;

rather racial identities are seen as historical products, which shape social affinities and

antipathies, and thereby precipitate various social groupings and boundaries (Segal 1993).










Notably, classification by race was practically impossible and mostly meaningless

because of divergent ethnic groups who were brought by the colonial government

adopting a variety of immigration schemes in order to import laborers from other

Caribbean islands and other countries including Portugal, Syria, Lebanon, China, and

India, including former American slaves. TT has not been a racially-stratified society, but

rather has exhibited an ethnic-class social structure.

Bridget Brereton (1993) shows that Trinidad' ethnicity cannot be racially-stratified

society because of the following three reasons: First, Trinidad entered into its phase of

plantation development relatively later.2 Consequently, Trinidad's experience of

plantation slavery was brief (about fifty years).3 Second, Trinidad entered the post-

abolition era with an unusually large "middle tier."4 Third, large-scale immigration,

which was a result of labor shortage after the abolition, transformed the three-tier model

by introducing new ethnic groups. Ethnic differences within classes were important and

each ethnic or class influenced the other in terms of culture and values in creation of

"identity" in the host society (Yelvington 1993). Therefore, ethnic classification has been



SConcerning the ethnic composition, Tobago society cannot be considered in the same sphere as Trinidad
society. Tobago's population is dominated by Africans (92%). In addition, Tobago was a completely
separate entity with no administrative links to Trinidad up to 1889 when Tobago forcefully was made a
ward of Trinidad by Britain. Its historical experience was quite different from Trinidad's. Importantly,
Tobago does not have the same history of multi-ethnic immigration as Trinidad. Therefore, in the TT
history, 'Trinidad' alone is referred or each island is described respectively until 1889.

2 Trinidad became a significant producer of West Indian export crops in 1784 when Britain ceded St. Lucia
and Dominica islands and French planters moved on to Trinidad. Especially it was enlarged after the
disturbances in St. Vincent and Grenada in 1795, most British planters moved into Trinidad which was
comparatively tranquil and contained large areas of uncultivated, and unoccupied land (Rogozinski 1994).

3 The years between 1784 and 1838 were the period of slavery in Trinidad. This compares to 200 years of
the classic slave society such as Jamaica between 1655 and 1838, and Martinique 1635 and 1834.

4 In 1838, 42% of the population in Trinidad belonged to the middle tier, while the middle class in Jamaica
was 12% and 32% in Martinique.









appropriate for studying the TT society; most researchers prefer to use "ethnicity" as an

analytical category embracing political, economic and ideological relations. As stated

above, race has been recognized as a socially constructed phenomenon like ethnicity,

therefore the usage of two terms are dependent upon the social context of each society.

Hence, in this study, race and ethnicity is considered the same analytical division as a

reflection of socioeconomic, cultural, political, behavioral, and health differences

between collectives in a specific society.

Ethnic Context in Trinidad and Tobago

According to the 1990 census, TT poly-ethnic society includes, besides the two

major ethnic groups, the East Indian and African, 19.0% Mixed heritage people and 1.7%

other ethnic groups composed of Spanish, French, Portuguese, Syrian, Lebanese,

Chinese, Philippines and others. African and East Indian peoples have played important

roles in economic, political, and cultural development in TT. The dynamics of the

relationships between the African division and the East Indian division has significantly

influenced the island's transformation from a colonial society to a multi-cultural and

multi-ethnic republic. The former was brought to colonial Trinidad as slaves for working

on the new plantations as a result of Cedula (low in Spanish) of Population in 1783

substituting the extinguishing indigenous population. The latter came to Trinidad as

indentured servants for, similar to Africans, working on the growing sugar plantation, and

for replacing the emancipated Africans. The indenture system was merely a "new system

of slavery" (Tinker 1974).

The relationship between Africans and East Indians has been described

distinctively by each ethnic group. Each holds persistent negative perceptions of

subordination and superordination of the other. This relationship was established at the









arrival of East Indians through two main contexts. The first is the physical and

occupational isolation of the East Indians from the Africans. Many East Indians

succeeded in becoming peasant proprietors and a result the economic interests of most

East Indians shifted from sugar plantation labor into small-scale cultivation on their own

land. Therefore, they settled into a rural way of life, which contrasted with the lives of the

Africans in the urban areas who have lived mainly within ethnic enclaves. The second is

a unique circumstance in which a dividing line was drawn between aristocracies of

whites and former slaves of Africans versus East Indians who formed the bottom tier of

the society at the arrival. Consequently, this was the foundation that helped TT realize the

relative but equal distribution of economic and political power to each ethnic group. The

Africans seized political power for over 30 years since TT's independence while the East

Indians become economically competitive.

"Race" first provided the basis for communal identity and resistance to colonialism.

Both Africans and East Indians maintained aspects of their own cultures, distinguished

themselves from their European overlords, and challenged colonialism. Each preserved

many aspects of their traditional behaviors, customs, beliefs, and orientations, perhaps for

the purpose of conserving their identity, in part, for protection against "external"

pressures such as governmental policies, economic transitions, and offenses and censures

from members of the other ethnic culture.

Slavery to collective identity

Before the arrival of immigrants from India, stereotypes based on race had already

emerged. These stereotypes had antecedents in Spanish culture. The initial encounter

between any people of diverse cultures and civilizations of immigrants and slaves from

various places, naturally gives rise to comparison by self-examination. Biases, prejudices,









and other sentiments emerge from such comparison (Moore 1995). The superiority of the

"white" race became the basis for ideological justification for coloreds' servitude. British

and European intellectuals developed the idea of racial types as the most important

method of classifying people. They thought mankind was divided into permanently

different biological types. The doctrine of racial type and social Darwinism helped to

create a climate of opinion which was hostile to dark-skinned peoples everywhere,

especially when dealing with "uncertain newcomers" (Brereton 1979). Africans were

regarded by the planters as being lazy and irresponsible, having a penchant for drinking

and conspicuous consumption, and being prone to profligacy (Brereton 1979). Despite

the acceptance by these despised Africans of many European cultural and religious

practices, they successfully defended some aspects of their own culture and lifestyle in

the face of determined and powerful scorn, and occasional opposition, from the ruling

class. However, the Africans themselves would eventually behave somewhat like the

white and white Creoles, as they developed scornful stereotypes of those who were to

come after them. In 1833 the British government passed the Act of Emancipation,

declaring it a law in the following year.5 In the new society after the emancipation, the

system6 gave every incentive for the ex-slaves to leave the estates and seek independence

as a small holder and a part-time wage laborer in the city (Vertovec 1992).



5 Slave-owners throughout the empire were duly compensated, while the slaves themselves were originally
obliged to labor as "apprentices" for an additional six years. The requirement of apprenticeship was halted
in 1838 finally, and over 20,000 slaves of African descent were freed in Trinidad (Williams 1962, Brereton
1981).

6 After slavery was abolished in 1834, planters tried to maintain labor in their sugarcane fields offering a
rate of wages far higher than in the other British West Indies, as well as rent-free huts. These wages for
field labor between 50 and 65 cents per task or per day were higher than any paid in Trinidad for a century
to come in 1938, unskilled labors in sugar were earning 35 cents per day (Brereton 1981). Despite the
high wages, freed Africans were reluctant to settle far in the interior at a distance from existing centers of
population with their schools, churches and rudimentary social amenities and Africans themselves pursued









When the large indentured labor population from India arrived, those ethnic groups

already living in Trinidad took care to distinguish themselves from these newly arrived

East Indians. Brereton points out that "there is evidence that a "Creole identity" shared by

local white and educated colored and blacks was emerging in Trinidad, a Creole

solidarity in opposition both to the British representatives and to the Asiatic immigrants"

(Brereton 1979: 208). East Indians were especially singled out in this process of hostile

ethnic stereotyping. After all, they looked different, dressed and behaved oddly, spoke

different languages, ate strange foods, practiced queer customs, and worshipped weird

gods. In every sense, they seemed were in striking contrast to "Western" ways.

As the immigrants came from widely different regions of the Indian Subcontinent,

the newly created migrant world in Trinidad was characterized by substantial differences

in culture7 and economy. The remarkable heterogeneity of the migrant population and

their broad range of language were multiplied by distinct dialects due to smaller sphere

where they lived (Vertovec 1992). Initially, the internal heterogeneity of Trinidadian

society was not restricted to African/East Indian differences. There were strong internal

differences within the East Indian group itself. Some of these differences were religions,

distinguishing Hindus from non-Hindus. But even among the Hindus themselves, who

were 85% of the total immigrants from India, there were regional and cultural differences

such variables as languages and caste systems. These varied backgrounds contributed to

the demise of a significant portion of caste foundation and caste-based ideology.



a legal or nominal freedom. A large number of Africans entered the skilled trades and moved to Port of
Spain, San Fernando, and larger villages. These are now considered African dominated areas.

7Including language and dialect, dress, cuisine, caste composition and structure, architecture and village
layout (Vertovec 1996).









Therefore East Indians had not simply conserved pre-immigration cultural forms but have

created a series of syncretic or other modified cultural forms. Regarding the East Indians'

position within Trinidad's social, economic, and political structure, new types of

relationships among East Indians and with other ethnic groups have periodically worked

to produce changes in the conservation of East Indian culture. As the strong internal

differences among immigrants from India lessened, simultaneously, the strength of the

African/Indian differences became more evident.

These subjective ethnic stereotypes had a solid grounding in an objective emergent

division of labor: whites as plantation owners; Chinese and Portuguese in trading

occupations; Africans and coloreds moving into the professions and skilled manual

occupations; and East-Indians were almost completely in agricultural occupations.

Because of these occupational differences, the two largest groups, Africans and East

Indians, were separated geographically as well as culturally. Many Africans have been

urban-based in two cities; Port-of-Spain and San Fernando, while East-Indians have lived

in the rural central and southern parts of the island with a strong core found in the plains

of the sugar belt. Therefore, a pervasive and fundamental physical, geographical

separation characterizes Trinidadian society, as Premdas describes as "the Creole-cum-

colored portion versus the Indian portion" (1993: 100).

Hence, "social confrontation" has involved the "indigenization" process of several

migrant groups, divided first by race then by ethnicity, language and religion, plantations,

small-holdings, villages and growing towns. It has been characterized by the infighting of

these groups, both within each group and against other groups. The ethnic diversity may

have encouraged East Indians to seek common ground. Interestingly, this situation has










not been influencing only the East Indian community but also the African community.

This situation is vividly remarked by an Afro-Trinidadian friend of mine. "If we wouldn't

have Indo-Trini, we wouldn't be the Afro-Trini. If we were not here, they were not East

Indian, we may be complementary to each other..."

Ethnicity and class consciousness

Trinidad's situation as a colony under foreign control changed when partial self-

government was instituted in 1925. The first political organizations in TT developed in

the 1930s, when a worldwide economic depression spurred the formation of labor

movements. Full adult suffrage was introduced in 1946. In 1956 the People's National

Movement (PNM) was formed by Dr. Eric Williams, who became the first Prime

Minister in 1958, drawing on the support of mainly African elements of the population.

The opposition8 drew support from the East Indians.9 The PNM continued to win

elections. By the 1970s, the island's industrial structure had shifted from an agriculture-

based economy to an oil industry-based economy10 that produced revenues for the

government. It has been said that revenues from oil exports were used to assist African

population subgroups who were identified as underprivileged. Members of the East

Indian group perceived that most of the funds went to poor Africans in urban areas. East

8 The Peoples Democratic Party (PDP) was established in 1953 by Bhadase Maraj supported by the rural
Hindus.

9 In 1960, the composition of the population was African 43.3% and East Indian 36.5%. An East Indian
majority was first noted in the 1990 Census.

10 The chief domestic beneficiary of oil income is the central government, which receives oil revenue
through taxes, royalties, and ownership. (In Trinidad and Tobago 28% of the industry was in the
government's hands in 1996). The resulting expansion of the public sector crowds out the private sector.
Then, many of the public sector commitments made during the boom were difficult to reverse and so
caused delays in adjustment when the boom ended. In Trinidad, the public sector accounted for 30% of
GDP, (compared to the other resource-supplying nations such as Chile 8%, and Argentina 18%, the public
sector is unfavorably large), 30% of total employment, and over 50% of salaried employees. (World Bank
1996).










Indians further concluded that urban Africans benefited more from job opportunities

provided by the government." In this way, government policies contributed to inter-

ethnic tensions. The ethnic "competitions," which began with the importation of forced

and indentured labor early in Trinidad's history, were kept in force when Williams

provided special attention and assistance for Africans, who he considered to have

suffered past discrimination. This process, termed a "symmetrical political patronage,"

led to feelings of alienation on the part of other ethnic groups (Center for Ethnic Studies

1993).

The traditional view differences in occupations persist today, based as it is in

certain objective facts. Rural-based East Indians have the lowest income. People of

African and mixed heritage have reached the mid-point of the income distribution while

Whites and Off-Whites12 have the highest income (Harewood and Henry 1985).

However, the actual situation is more complex than these general statistics suggest. For

example, government workers mainly earn more than those in private enterprise,13 a fact

that favors Africans. On the other hand a majority of millionaires are Syrian or East

Indian. In addition, the Center for Ethnic Studies reported that there is no sufficient

I The PNM maintained a patronage network targeted at Africans, especially urban ones. One method was
the establishment of the government's Development and Environment Works Division (DEWD), which
employed workers for road construction and maintenance. Almost every DEWD project was aimed at
African areas and hired African workers (Yelvington 1995). Percy C. Hintzen (1989) and Steven Vertovec
(1992) stated that patronage was accomplished most effectively through the state sector, and the PNM's
industrial strategy was aimed at urban Africans, at the expense of agriculture, the livelihood of many rural
East Indians. With the end of the oil boom, the oil money ran dry and the subsidies were removed.
Ironically, this sudden recession made Africans suffer severely because the majority work for the public
sector.

12 Off-White is applied to immigrant groups, essentially, who are perceived as very close to White in skin
color, but are seen as less powerful politically. This category includes Portuguese, Syrian/Lebanese.
Chinese are treated as such by other groups.

13 An average monthly wage in the public sector is higher (TT$ 2,300) than that of in the formal private
sector (TT$1,500) (World Bank 1995).









evidence for the African division's heavy dominance and underrepresentation of East

Indians in the public sector (1993). "Race" was identified as a factor influencing

promotion in some of the public companies surveyed however, "racial discrimination"

was in fact a tendency towards speculation that glided easily into charges and counter-

charges of discrimination (Center for Ethnic Studies 1993).

Given such disparate facts, it is not easy to determine which group is actually better

off. Objective complexity notwithstanding, the subjective perceptions remain among the

people. The East Indians believe the Africans benefit most from the government, while

the Africans think the East Indians discriminate against them in the formal private sector.

Selwyn Ryan indicates that perceptions of economic status among the ethnic groups tend

to be viewed from an individualistic point of view (1991). He explains, "all groups (with

the possible exception of the Syrians) believe that they are economically dispossessed.

The latter however also believe that they are dispossessed in the sense that they have not

been given the social recognition they deserved and that they are still the butts of

ridicule" (Selwyn Ryan 1991: 78). Both groups are concerned about the economic

"gains" made by "the other" group, creating in effect a zero-sum game them.

Conflicts between the two groups over education, which is one of the most

important avenues of upward mobility in a developing ethnically heterogeneous society is

an example of the compound product of the cultural confrontations. Religion, especially

Christian missionary, which represented the western culture, played a significant role in

establishing educational facilities in TT. Religious conversion was practiced among some

East Indians in order to gain material benefits, namely, education. When East Indians

entered Trinidadian society, they were considered by the host society a lower class and









minority population. This stereotyping by religious denominations led to the creation of

separate school facilities and to the subsequent PNM policy dealing with the location of

schools, the language of instruction, religious orientation, the admission of students,

awarding of scholarships, and treatment of teachers that led to the establishment of the

East Indians' denominational schools.

These dynamics injected an element of ethnic exclusiveness into the educational

sector (Gosine 1986).14 The feeling of East Indians that East Indians had of being

educationally disadvantage continued under the PNM administration. Many East Indians

believed that they would not receive a fair share of educational benefits especially in

terms of the awarding of scholarships and the hiring and promotion of teachers.

Considering the geographical advantage of Africans, namely, their presence in urban

areas and consequently their greater proximity to the majority of schools, the educational

disadvantages of the East Indians might not derive from any deliberate action on the part

of the government. Did the PNM intentionally favor its own ethnic group and slight the

interests of the East Indians within the educational sector? This question is hard to answer

empirically and objectively. However, the subjective perceptions of intentional

educational discrimination against East Indians is obvious and simple common sense to

the people of Trinidad, especially to members of the East Indian group.

The empirical evidence indicates that by 1980s, Africans and East Indians had

leveled in terms of group income (Yelvington 1995). Since 1996, political power has

14 East Indians could not receive a fair share of educational benefits under the PNM administration. The
educational institutions, by and large, are located in such areas where they best suit the convenience of
black students in areas inhabited predominantly by that race. The participation of East Indian students in
higher education causes African students to regard them as socio-political threats. Similarly, East Indian
students feel threatened by the gains of the African students whom they see as the recipients of government
support (Gosine 1986). On the other hand, some recognize that the PNM's education strategy contributes to
both ethnic groups (Selwyn Ryan 1991).









oscillated between PMN and the United National Congress, supported by East Indians. It

may be true that this closeness and their juxtaposition could cause them to be competitive

with each other over allocation of resources, and through a class-consciousness that is,

each group thinks it is superior.

Properties of Child Mortality

Difference between Infant Mortality and Child Mortality

In analyzing mortality in the postnatal period, the age at which mortality is

measured calls for an important consideration because of "the social and biological

factors that affect mortality vary by the age of child" (Wood and Lovell 1990).

Practically, analyses concerned with the causal pathways of postnatal deaths are hitherto

divided into two groups; infant mortality (under 1 year) and child mortality (1-5 years).

This dichotomization is mainly distinguished by the terms of cause of deaths, and two

sets of causes are designed as endogenous and exogenous. Generally, the former class of

death is presumed to arise from the genetic makeup of the infant, the circumstances of

prenatal life, and the conditions of labor, which are difficult to prevent or treat in the

present state of knowledge. The latter class is presumed to arise from purely

environmental or external causes, i.e., it is related to the contact of the infant with the

external world. Exogenous mortality primarily includes infections and postnatal

accidents, which are relatively preventable or treatable (Shryock et al. 1976). Considering

the statistical facts that the highest risk for infants is under one month of age over 95

percent of infant deaths (Shryock et al. 1976), causes of infant deaths are mainly

considered endogenous, or at least the proportion of the endogenous causes is larger than

that of exogenous causes for the infant deaths. On the other hand, child mortality is

determined by the combined effect of both endogenous and exogenous factors, and it is









also more sensitive to a broad range of environmental conditions (Wood and Lovell

1990).

Related to the age issues stated above, there is another dichotomization in a wide

range of demographic research. Prior research has indicated that the nature of child

mortality in regard to emphasizing the importance of sociological research in illustrating

how various factors causing child mortality shape the risk of child death. Therefore,

research concerned using the causal pathways of child mortality can be dichotomized.

One view focuses on the direct impact of social and economic environments of the risk of

child mortality and the other view focuses on the impact of variations in health services.

The relationship between socioeconomic environment and child mortality is captured in a

remark by Wagner, a neonatologist and perinatal epidemiologist, quoted by Gortmaker

and Wise (1997: 156);

Infant mortality is not a health problem. Infant mortality is a social problem with
health consequences. It is analogues to traffic accident mortality in children: the
first priority for improving traffic accident mortality in children is not to build more
and better medical facilities, but rather to change traffic laws and better educated
drivers and children. In other words, the solution is not primarily medical, but
environmental, social and educational. The same is true for infant mortality: the
first priority is not more obstetricians or pediatricians or hospitals, nor even more
prenatal clinics or well-baby clinics, but rather to provide more social, financial and
educational support to families with pregnant women and infants (1997: 156).

The powerful role of socioeconomic forces in the prediction of disparities in infant

mortality is stressed by this perspective.

The other view focuses on health services. Control of childhood diseases is typical

of health services, which can be provided equally to nations to improve health care

technology. But the relations between child mortality and quality of health services

cannot inclusively be studied without the elevation of social pathways. Especially, health

care technology or westernizing health care may widen the disparities between "haves"









and "have-nots." This account is expressed by Gortmaker and Wise indicating that the

trends of decline of child mortality are mainly due to innovations in health services;

"such technological change also creates new opportunities for socioeconomic

differentiation as life-saving therapies or preventive interventions potentially are made

available only to the economically advantaged" (1997: 148).

Combining the two categorizations suggested above, the research spheres for infant

mortality and child mortality seem to have been closer. As researchers have become more

interested in the linkage between social inequality and mortality outcomes, they begin to

emphasize the diverse mechanisms through which this relationship is manifested, as well

as how various mortality influences vary in their causal priority and proximity to the

biological event of death (Eberstein 1989). Postnatal mortality, including both infant and

child mortality, represents the cumulative effects of all factors characterized by both

endogenous and exogenous factors, and by pathways affecting postnatal mortality,

socioeconomic circumstances, and health care accessibility. Moreover, child mortality

may be a more intricate composite of a number of component rates, each with its own set

of relationships with social factors, health services and individual mothers' health

orientations. This is why demographers and sociologists have been interested in using

child mortality to explain the social inequality and social stratification within a society,

and the primary reason for this study, which uses child mortality as a general term

including both infant mortality and child mortality.

Determinants of Child Mortality

Demographers and public health scientists have developed and shared the common

perception of the contribution of socioeconomic development and the medical and

primary health services offered by public health programs in the reduction of mortality









(Preston 1975, Caldwell 1979). Differentials in child mortality between population

groups have been a constant topic to social scientists. The most common variables used

in the study of child mortality differences between sub-groups within a nation are

socioeconomic standings as the main effect or as surrogates for other variables about

which information was not directly available (Hobcraft et al. 1984). The primary reason

is because the extreme dependence of children especially under 5 years old makes child

mortality a sensitive measure of the quality of life. As Gortmaker (1979: 281) explained,

"infants exercise no responsibility for their environment and health status, and thus an

infant's own motivations and actions have little impact upon its chances for survival;

most influences should come from its parents and the surrounding environment." Hence,

child mortality has been considered as a mirror of the quality of given circumstances for

each child, and the child mortality differences among the people have a significant role in

measuring the disparities between population groups within a nation.

Socioeconomic and Demographic Variables on Child Mortality

Income

Income is an influential factor in consideration of individual socioeconomic status

and an important determinant of child mortality (Gortmaker 1979, Mosley and Chen

1984, Hummer 1993). Higher incidence of poor pregnancy outcome and child mortality

were found among women from disadvantaged socioeconomic backgrounds. The mere

existence of differentials by household income however, does not constitute a fully

satisfactory explanation of the disparities of child mortality in income levels as well as

educational levels. There are a variety of proximate and intervening factors related to

infant mortality that are also associated with income and education and thus, need to be









controlled. Such factors fall within a context that rationalizes tests of relationships

between socioeconomic status and child survivorship.

Maternal education

Mother's socioeconomic standings are represented by several factors such as

education, household income, work situation, occupation, and quality of housing.

Maternal education as a socioeconomic indicator appears most frequently in studies of

child mortality, "because other measures that might be preferable, such as family income,

are not available in the vital statistics records that constitute the basis of most research in

this area" (Hummer et al. 1999: 1087). Also, maternal education has drawn a wide range

of research interest because it is the most relevant and intuitive from the standpoint of

child health policy relating to education-conducted use of health services. Although the

results analyzing the connection between education and child mortality vary in strength

of impact on child mortality outcome (Benyoussef and Wessen 1974, Caldwell et al.

1983, Hobcraft et al. 1984, Ce Chen and Williams 1997), the most valuable nature of

education as well as household income are that they appear to be the most common

variables tapping into not only the cohort of socioeconomic factors such as work

situation, occupation, and quality of housing but also into almost all explanatory variables

of child mortality. If the magnitude of education as a socioeconomic variable on child

mortality varies in each society or each cohort within a society, the covariates could

shape and state the important variance explaining each society's characteristics.

Marital status and residential characteristics

The influence of demographic characteristics such as residential (urban-rural) and

marital union characteristics on child mortality are often examined with socioeconomic

status. Mothers' place of residence (urban-rural distinction) is used as a proxy measure









for living conditions to illustrate both public and medical health provisions. This is due to

the lack of infrastructure such as electricity (especially for refrigerator), drinking water,

non-drinking water (for flush toilet), and sewage, and access to basic health care facilities

which may be life threatening to the children in rural areas (Suwal 2001). Urban residents

were found to have better conditions compared to rural residents. This is especially likely

given the confounding of many other socioeconomic attributes with place of residence.

However, in some cases, the residential differences appear to show internal differences.

Compared with the homogeneity of experience for urban residents, mortality in the

traditional rural areas varies widely even between sub-groups with similar attributes

(Hobcraft et al. 1984).

The association between marital status and child mortality is examined by using

various controls such as education (Keller 1978), race (Cramer 1987, Eberstein 1989),

race and age (Gee et al. 1976), and race and intervening factors (Eberstein et al. 1990,

Hummer et al. 1999). The magnitude of marital status on child mortality differs,

depending on the covariates, but these studies found that significant interactions, between

marital status and education, and race and unmarried status, are associated with higher

child mortality. However, Cramer indicates that; "marital status," similar to residential

and age, "may not be an independent risk factor. ---. In general, it is not known which

social factor or combination of factors is causally responsible for the observed group

differences" (1987: 299). For example, children born to unmarried women may be at

higher risk for mortality as a result of inadequate familial resources rather than marital

status per se (Eberstein et al 1990). Therefore, "marital status" is considered to be

substitutive to the level of "quality of life" of mothers. We may, however, have to be









careful that marital context varies in each society. The demographic factors as such

mentioned above are generally attributive covariates.

Intervening Health Care Variables in Relation to Socioeconomic Variables

The association between child mortality and maternal education has been the most

common finding in the child mortality literature as mentioned above. Several lines of

inquiry on education's direct and indirect function on child survivorship have been

determined. There are two major directions of study in the association between education

and child mortality; the first linking to economic status, and the second linking to skills

and health care practices. The former is more likely conditional and influenced by her

familial construction. The economic status of married women is largely determined by

her husband's educational achievement and his occupation and employment status.

Paternal education captures variation in household wealth or disposable income, and the

relation between child mortality and education is accounted for by the various adopted

indices of household economic condition (Hobcraft et al. 1984). Father's education also

influences attitudes, preference, and choice of consumption goods, including childcare

services. Therefore, "in many cases correlations between health effects and educational

level of fathers (or other non-childbearing, economically productive adult members in a

household) largely occur because of operations on the proximate determinants through

the income effects" (Mosley and Chen 1984: 34).

While paternal education has a role in limiting or facilitating the orientation and

selection of mothers' way of life and mothers' characteristics, maternal educational level

is more likely linked to skills in health care practices and health orientation directly.

Because of biological links between the mother and infant during pregnancy and

lactation, mother's health and nutritional status influence the health and survival of the









child (Mosley and Chen 1984). Studies by Benyoussef and Wassen (1974) and Boerma

(1990) show that better educated mothers more commonly use maternal and child health

services than less educated mothers. Furthermore, Streatfield, et al. (1990) indicate that

educated women have greater awareness of correct immunization schedules. This can be

another dimension of formal education, which Caldwell has argued, that the significance

of maternal education's role is to change traditional patterns of familial influences so that

women may improve their understanding of the importance of using modern medical

services (Caldwell 1979, Caldwell et al. 1983). Mother's educational level can affect

child survival by influencing her choices and increasing her skills in health care practices,

such as nutrition, hygiene, preventive health care, and disease treatment.

Prenatal care

Prenatal care has generally been considered to contribute to good birth outcomes

and also as a predictive variable of child health and child mortality. The adequacy of

prenatal care as evidenced by prenatal checkups, the presence of trained health

professionals, whether doctors, midwives, or traditional birth attendances during delivery,

tetanus toxoid immunization and nutritional supplementation, etc., has a direct impact

on maternal morbidity and mortality (United Nations Population Fund 2000).

Prenatal care is considered as "a package of necessary services" (Shiono and

Behrman 1995). Therefore a number of benefits accrue; such that prenatal visits play an

additional role: to enable women to obtain general information on infant and child health

care as well as specific medical attention. Several recent studies have examined the

relationship between race/ethnicity and child health and mortality while controlling for

limited sets of confounding factors, such as the risk of infant birth weight, with household

income and education (Kleinman and Kessel 1987, Collins and David 1990). These









studies indicate that the interrelationships among prenatal care, birth weight, and child

mortality document the importance of socioeconomic and other social variables on the

probabilities of low birth weight and the risk of child death. The relationships between

prenatal health care and maternal education have been found to be highly correlated with

each other (Cramer 1987, Hogue et al. 1987, Mangold and Powell-Griner 1991).

Socioeconomic variables, household income and education, serve as indicators of

knowledge, amount of medical services and level of household income. These variables

exercise an important effect on the ability to obtain medical provisions. Echeverarria and

Frisbie point out the potential of prenatal care practice: an increased utilization of a wider

array of postnatal health care services was found among mothers who practice an

adequate level of prenatal care (2001).

Preventive health care

Represented by breastfeeding and timely immunization, postnatal care includes

feeding children nutritious solid food and sanitation, and a hygienic way of living is vital

in preventing possible postnatal child deaths. Child immunization is one of the health

systems' principal interventions aimed at lowering child mortality. Preventive health care

has been improved dramatically since the Second World War such that relatively rapid

child mortality reduction in mid-20th century was primarily attributed to this health

technological improvement (Suwal 2001). The evidence, which supports the relationships

between accessibility to health services and child mortality, has been provided by Maison

and Sekeito (1987) and Chaulagai (1993). However, according to Cleland and van

Ginneken (1988), the nature of the interaction between accessibility and utilization of

health service influencing the reduction of child mortality is context-sensitive. Its nature

depends mainly on the level of development of the health infrastructure. Education and









geographical accessibility are substitutive, but maternal education has more potential to

compensate for the disadvantage of mothers' lack of accessibility to health services

(Rosensweig and Schultz 1982, Kadende 1994, Bicego and Boerma 1996). Education

equips mothers with knowledge of healthy living and encourages them to practice proper

health care.

Research examining the relation between ethnic groups and receipt of preventive

services, which are usually in the form of care such as pap smears, breast exams,

mammography, and cholesterol screening, found general differences in female preventive

health care use among racial and ethnic groups. Furthermore, each ethnic group has a

tendency to receive a particular preventive service (Corbie-Smith et al. 2002). For

narrowing the differences by race in preventive health care practices, it is necessary to

address racial differences in disease outcomes. Simultaneously, we may consider the

influence of a certain ethnic group's common tendency to receive health care as one of

their child care practices. The more mothers receive preventive health care within an

ethnic group, the more they become aware of the importance of having their children

receive immunizations appropriately.

Breastfeeding

Breastfeeding has been emphasized because of its significant influence on the well-

being of children. Children's health advantages are conferred by breastfeeding and,

conversely, there are detrimental effects of failure to breastfeed on the child deaths. A

large body of evidence indicates that children who are bottle-fed from birth run a higher

risk of health and developmental problems than do breast-fed children (Goldberg et al.

1984). The relationship between breastfeeding and child mortality has been examined

mainly with mothers' socioeconomic characters, education and income which are









considered as having strong predictive power, whether or not a mother breastfed. Mothers

with a higher formal education and a higher monthly income are more likely to breastfeed

(Goldberg et al. 1984, Huffman 1984, Trussell et al. 1992, Humphreys et al. 1998, Alan

Ryan 1998, Forste 2001). Marital status is also strongly associated with breastfeeding;

mothers being married are more likely to breastfeed than mothers having other union

status (Hirschman 1981, Forste 2001) which indicates that the support of the child's

father is important in the breastfeeding decision.

Residential differences are also predictive of breastfeeding. Usually breastfeeding

is considered to affect child mortality most strongly in the cities (Trussell et al. 1992).

Some findings indicate that urban settings are negatively associated with mothers'

breastfeeding decisions, due to general perceptions toward bottle-feeding, which is

considered as a modern, adequate, and convenient method. Additionally, for mothers

living in urban area, it is rare to have role models of breastfeeding practices (Huffman

1984) suggesting urban-rural cultural differences. Provided that living conditions in

urban areas have been improved such as access to safe drinking water, medical facility,

and health information, the impact of breastfeeding on child mortality has been reduced

in cities. Instead, Goldberg, et al. found that accessibility in rural areas have been left as it

is or is still behind the urban areas, therefore, association between breastfeeding and child

mortality becomes stronger and excess urban areas; a higher mortality risk is experienced

by the non-breastfeeding division in rural areas (1984).

Not only is the breastfeeding variable considered an indicator of quality of

childcare, but also it may explain individual mothers' perception of "time," which is

strongly influenced by their daily life orientation and its tempo and general health









attitudes and beliefs of the mother and those of her social network (St Clair PA et al.

1989, Butz et al. 1993). Mosley and Chen (1984) mention the importance of a mother's

"time" caring for her children and maintaining their living circumstances such as prenatal

visits, breastfeeding, food preparation, sanitation, and sickness care. Childcare time often

competes with time needed for income-generating work. They note that, "A mother's

time may also be required for other economically productive activities that may or may

not be related to child health" (Mosley and Chen 1984: 35). The physical accessibility to

modern ) health services is also largely determined by mother's socioeconomic standing;

thus, issues of accessibility to health institution and services extend to the variability

within education. Greater physical access to health services improves survival to a greater

extent among the children of less educated women than for children of more educated

women (Katende 1994, Bicego and Boerma 1996).

Type of place where the child is born

Place of delivery; whether or not children were born in a medical facility or private

home that may be unsafe and unhygienic, have been shown to influence child

survivorship. Whether or not the baby was delivered in a hospital serves as one extreme

indicator of lack of medical care (Gortmaker 1979). For mothers, the high risk of

neonatal tetanus deaths was found to be associated with home delivery (Foster 1983), and

the births taken place at a medical center and assisted by doctor, nurse, or midwife are

found to have a significantly lower risk of child mortality (Suwal 2001). Although

traditional birth attendants still deliver a considerable proportion of newborns in

developing countries and rural areas including middle developed countries, the

proportion has shown a tendency to decline (United Nations Population Fund 2000).









Having provided the persistent interrelations among socioeconomic background,

namely, income level and educational achievement, and demographic and health related

factors, it is evident that these factors do not independently determine child survival

chances. Eberstein et al. state that, "there are reasonable a priori theoretical grounds to

expect interaction among them that the effects of some of the variables may vary

depending on levels of the others (1990: 414). The effective variables and their strength

may also vary according to the social system of a given society. For a better

understanding of the association between socioeconomic factors and child mortality,

recent child mortality literature includes health related factors such as prenatal health

care, preventive health care, and breastfeeding which can indicate mothers' childcare

orientations and cultural influence derived from their race and ethnic background. On the

other hand, recent studies in the United States indicate that race is a modern social

stratifying agent coinciding with the emergence of slave trade. Hummer stated that "such

destructive exploitation and reasoning served as fertile ground for continued inequalities

in resources, status, power, and health between socially defined group" (1993: 533).

The phenomena remain at present and there is a general consensus that race and

ethnicity, as social stratifying agents, continue to affect child mortality (National

Research Council 1989, Hummer 1993, Mullings et al. 2001). As seen in the ethnic

context in the TT society, which embodies a number of dimensions that are indicative of

socioeconomic status, women who are from disadvantaged socioeconomic background

and from an ethnic group which experienced continued lower socioeconomic status in TT

should be considered higher child mortality risk group.







44


Infant and Child Mortality in Trinidad and Tobago

Infant and child mortality rates from 1960 to 2000 are presented in Figure 2. In

1990, 21 out of every 1,000 babies died before reaching the first birthday while 24 per

1,000 died before the fifth birthday (UNICEF 2003a). The decline of the infant mortality

rate from 61 to 21 deaths per 1,000 births between 1960 and 1990 represents a 65.6%

drop. An even greater decline of 72.6% is seen for under five mortality, which decreased

from 73 to 20. UNICEF provides the latest under five mortality rate of 20 and infant

mortality rate 17.15 These figures represent a very low level of mortality, approaching that




61
1960 7 102
153
49
1970 76
123
35
62
1980 62
84


1990 43
54

336
1995 36
43
7 h Infant Mortality TT
20 30 Infant Mortality LA&Caribbean
2000 E 20 Under 5 Mortality TT
36 R Under 5 Mortality LA&Caribbean


Source: UNICEF Statistics. UNICEF End Decade Database Child Mortality. hil,
childinfo.org/cmr/revis/dbl.htm, hliiip I L h.i.lii!....' 'o/cmr/revis/db2.htm.

Figure 2. Infant and under 5 mortality in Trinidad and Tobago and Latin American and
Caribbean, 1960 to 2000



15 UNICEF calculated infant and under five mortality rates based on an indirect estimation technique, the
Brass Method. The data used in the estimation are the mean number of children ever born to five year age
groups of women aged 15-49, and the proportion of these children who are dead, also for five year age
group of women. Hence, the infant mortality rate indicates the probability of dying before the first birthday.
The under five mortality rate is the probability of dying before the fifth birthday.









of developed countries. By comparison, the infant and child mortality rates in TT are

relatively lower compared with the other Caribbean and Latin American countries as a

whole.

A previous study by Heath et al. (1988) using TTDHS indicates the socioeconomic

characteristics of child mortality. The results show that both infant and child mortality are

lower in rural than in urban areas. This somewhat unexpected finding may reflect the

homogeneity of the socioeconomic conditions in the society: there is difficulty in

distinguishing urban from rural areas largely due to the developed transportation systems

throughout TT. As expected, mortality for children aged 1-4 decreases as the mother's

education increases. However, infant mortality appears highest among the best-educated

women. It is most likely because the rates for the highest and lowest education groups are

based on a small number of births.

The ethnic differences in child mortality in TT are gathered from a fertility study

in TT by Harewood (1978), a comparative study on child mortality between three

Caribbean nations, TT, Guyana, and Jamaica, carried out by Ebanks (1984), and the

survey on postnatal practices in Trinidad by Mahabir (1997). According to Harewood

who conducted research in 1970, the fertility was higher in the East Indian division

compared to African division -- the gap between the two groups was radically narrowing

over time and the situation is reversed in 1990 census. This is probably due to the natural

relationship between total children ever born and child mortality though, at the time of

Harewood's research, East Indian women were more likely to have experienced a child

loss. Mahabir reports that the infant mortality was higher among East Indian women than

Africans with respective proportions of 8% and 5% in 1994. Perhaps this is because of









the higher concentration of the East Indian population in rural areas. Generally, more

lower income families are found in rural areas than urban areas and Ebanks' findings

correspond to the disadvantaged income situation in rural areas indicating higher infant

mortality among East Indians in rural areas. Considering the account of Mahabir, the

higher infant mortality in the East Indian division may be situational.

Relatively lower child mortality in this nation is supported by such indicators as

large immunization coverage (DPT, Polio, measles), higher rate of perinatal care use, and

higher rate of receipt assistance at delivery with respective proportion of 90% (1995),

97.6% (1987), and 99% (1997) (Pan American Health Organization et al. 1998).

However, UNICEF (2003a) reports that only 1.8% of children under the age of four

months were exclusively breastfed. This is considerably lower than expected. Even

though immunization is free in all health centres (public sector) and the large

immunization coverage is reported, UNICEF indicates that the proportion of children

who have had all eight recommended vaccines in the first 12 months of life was

estimated to be relatively small amounting to 7.4% (2003a). Mothers' levels of formal

education differentiate the rates of immunization coverage for their children, that is

between mothers who had secondary or higher education and those who had only primary

education with respective proportions of 18% and 7.1%.

There are both public and private health care systems in TT. Health care is

provided privately for varying, but generally high rates of payment. Free health care is

provided by various state-owned and controlled public health facilities. Public facilities

include general hospitals, located in the main urban areas, and health centers, located in

all eight counties of the twin-island state. There are 101 health centres in total, as of 1996









(Mahabir 1997), under the control of the Ministry of Health, which plays a key role in the

delivery of health care to the nation. In this relatively small country, with good internal

transportation, many of the same doctors at the public facilities are involved in private

practice: many doctors work at public hospitals and operate their own clinics. Health care

delivery at state-administered institutions has been a contentious issue among members

of the public and among the providers themselves. The general population has

vociferously expressed numerous complaints condemning the quality of service offered at

public health facilities (Mustapha and Singh 2000). There is a perception that the services

of medical practitioners in a private practice are more efficiently delivered than those

offered at public health facilities which are characterized by the usual bureaucracy and

inefficiency that accompany state enterprises (Rathwell and Phillips 1986). Hence one

may be tempted to believe that the quality of service received is linked to patients' ability

to pay.

Hypotheses

The analysis of TTDHS is organized in order to empirically test hypotheses that are

derived from theoretical perspectives of Coleman's social system and based on the

literature review. Social system theory concerns the balance and dynamics of interests

and control over scarce resources between groups contributing to constructing the social

structure. Coleman's theory is informative because it gives people a way of viewing what

has occurred over socioeconomic and political allocations between the two major ethnic

groups in TT. Although his social system theory recognizes the interaction among an

individual's purposive actions, social networks, and social capital, ethnicity or race is not

considered as a pervasive criterion in a society. In the ethnic context in TT, ethnic

identity can be considered as a major motive and value to seek the common social and









economic benefits through which we can locate the status of individuals and their groups

in the social structure. The following hypotheses are based upon social system theory:

H1: Socioeconomic factors have the strongest and most persistent association with child

mortality among all explanatory variables.

H2: Ethnic identity will show an association with child mortality, but after controlling

for socioeconomic, demographic, and health care factors, ethnicity cannot maintain

its influence on child mortality and its statistical significance.

H3: Individuals' economic levels affect children's and mother's quality of health;

therefore, better health status and favorable maternity care reduce the risk of child

mortality.














CHAPTER 3
RESEARCH DESIGN AND METHODS

The previous two chapters presented the significance of the study of the

relationships between child mortality and ethnicity in TT. Many researchers have found

that there are economic, political, and educational inequalities in TT society. However

not only are the discourses conflictive but frequently "perceptions" held by each ethnic

group toward others cause conflictive claims about inequality in allocation over social

resources. Race/ethnicity, socioeconomic factors, and prenatal/postnatal care including

breastfeeding decision have been found to have a significant impact on child

survivorship, but they do not independently affect child mortality. In addition, the effects

of variables may vary depending on the other variables and on the social system of a

given society.

The purpose of this chapter is to describe the TTDHS that is used in the next

chapter to see if ethnic background plays a significant role in determining the probability

of child mortality in TT society and to understand the level of influence of factors on

child mortality to describe the characteristics of the social stratification system in this

nation.

Data and Sample Size Analyses

The TTDHS, a national-level sample survey, was conducted by the Family

Planning Association of Trinidad and Tobago in 1987. The sampling frame for the

TTDHS was based on the 1980 Population and Housing Census, one of the Continuous

Sample Surveys of Population used by the Central Statistical Office of the Republic of









Table 3-1. Distribution of women 15 to 49 by ethnic and type of place of residence in
1987 and 1990, Trinidad and Tobago

Characteristics 1990 Census 1987 Census

Ethnicity
African 39.6% 35.3%
East Indian 42.9% 47.0%
Mixed 15.4% 17.1%
Others 2.1% 0.7%

Type of Place of Residence
Urban 48.7% 44.4%
Rural 51.3% 55.6%

Source: Population and Housing Census 1990, Central Statistical Office
Demographic and Health Survey, Trinidad and Tobago, 1987 (Individual dataset).
Sample population = 3,807



Trinidad and Tobago. The TTDHS is primarily concerned with family planning, maternal

and child health and child survival. Information on household composition and housing

quality is also included. Additionally, there are data about ethnicity, religion, and

education, as well as economic indicators such as the presence or absence of consumer

goods. The TTDHS has three separate sets of data based on household information,

individual information, and child information. This study utilizes only one set, the

TTDHS individual dataset that contains a total of 3,807 cases (individual women). The

data is composed of women between the ages of 15 and 49. Table 3-1 shows the

characteristics summary of the TTDHS along with corresponding figures from the 1990

Census. The original data is composed of 1,342 African women (35.3%), 1,787 East

Indian women (47.0%), 649 Mixed women (17.1%), and 27 women (0.5%) claimed some

other ethnicity. Comparing to the census figures, African women make up a smaller

portion of the sample and East Indian women represent a larger portion of the sample.

The report of TTDHS presented by Family Planning Association of Trinidad and Tobago









partly explains the reason for the difference in the ethnic composition; it could be due to

an unintended over-sampling in areas where the East Indian population is heavily

concentrated and a higher response rate among this group. Since the majority of East

Indians have resided in the rural areas, a less urbanized sample than the actual Census

population may be a reflection of those differences in ethnic distribution of the TTDHS

from the actual Census numbers.

This study investigates three main questions: (1) Whether or not ethnic background

influences child survivorship, (2) whether or not quality of life and education influence

child survivorship, and (3) Whether or not child health care practices impacts on child

survivorship. Of particular concern in this study is how the ethnicity influences the

relationship between child mortality and its predictors. The data is first restricted to only

women who belong to the two ethnic groups, African descent and East Indian descent.

The second restriction includes only those women who had at least one child born. To

assess the overall well-being of women in TT, it is pertinent to restrict the sample

population to a narrow age gap as possible to avoid economic gaps over time; time lags

between the time of childbirths and the time of survey, and faltering ethnic differences

among the sample population. Hence the third restriction includes only women who had

children) within 10 years prior to the survey being conducted. The analysis of child

mortality differences between the two major ethnic groups deals with various factors,

such as demographic standings, quality of life created by variables of household

composition and housing quality, and mothers' child health care. Lastly, women who did

not answer any questions used in the analysis were excluded from the analysis. These

procedures reduce the sample size to 1,082 from 3,807. Consequently, the sample









population used for the analysis in this study is composed of 584 East Indian women

(54.3%) and 492 African women (45.7%). This represents 34.4% of all the respondents

who belong to the two ethnic groups in the TTDHS.

Measures

There are three clusters of predictors of child mortality: socioeconomic,

demographic, and health related. The first cluster includes demographic factors such as

ethnicity, place of residence, and marital status. The second cluster includes

socioeconomic factors such as maternal educational attainment and quality of life. No

data was collected in the TTDHS (it is common to all other DHS-I surveys) on household

income per se, however, data intended to capture variations in household wealth and

disposable income were collected and they are useful for creating an indicator of

economic status which can be called "quality of life." The third cluster includes health

related factors such as prenatal care, type of place child born, quality of preventive health

care, and breastfeeding. These variables are useful indicators of whether or not the

respondent has access to a higher quality of health services and whether or not the

respondent provides appropriate child health care to her children. Differences in health

orientation and preference of medical facility, medical doctor, or medicine between the

two major ethnic groups have been observed (Mahabir 1997, Mustapha and Singh 2000,

Chijiwa 2001). Selections of the type of medical services serve as not only an indicator of

the quality of health of women but also a proxy for their cultural and ethnic vestiges.

Before proceeding to the descriptions of predictors mentioned above, the procedures of

creating the respondent variable; child mortality, are presented for first then variable

definitions and constructions are provided.









Child Mortality

The analysis for this study includes women who had at least one child born within

10 years prior to the survey being conducted. Therefore, child mortality of a woman is

determined in regards to her children who were born between 1978 and 1987.

Information on the child survival status of each woman at the time of the survey is drawn

from a variable "whether or not the child is alive" for computing child mortality in TT.

Each woman was asked about all her children she had ever had from the youngest to the

oldest. They were asked whether or not the youngest child is alive, whether or not the

second youngest child is alive, whether or not the third youngest child is alive, and so

forth. Additionally, women who had at least one child within ten years provided answers

on whether or not their children were alive up to the 11th child.

In constructing the variable of "child mortality," children who had died prior to the

survey are coded as 1, and children who are alive when the survey was conducted are

coded as 0, and then each woman's answers of all her children are combined to create the

variable; how many children the woman lost within last ten years. The analysis reveals

that 1,008 women out of 1,082 have never lost a child, 65 women have lost one child,

five women have lost two children, and four women declared having lost from three to

six children. Since child mortality distribution is heavily right-censored, the multivariate

analysis uses logistic regression techniques. The newly created variable of "child

mortality," which is the respondent variable in the logistic regression analysis, is

dichotomized: 1 = women (74 women) who have lost at least one child and 0 = women

(1,008 women) who have never experienced a loss of their child.















Table 3-2. Variable descriptions

Variable Type Value Value Label Proportion/Mean
(Frequency)


Childhood Mortality


Demographic Factors


Ethnicity


Type of Place of Residence


Marital Status


Categorical


Categorical


Categorical


Categorical


Have never had a loss of child
Have lost at least one child


0 (reference)
1 (dummy)

0 (reference)
1 (dummy)

0 (reference)
1 (dummy)
1 (dummy)


East Indian
African


Urban
Rural


Married
Separated / Divorced / Widowed
Cohabiting / Visiting Relations


93.2% (1,008)
6.8% (74)



54.5% (590)
45.5% (492)

40.8% (441)
59.2% (641)

58.4% (632)
6.4% (69)
35.2% (381)


Socio-Economic Factors

Years of Education

Quality of Life

Health Related Factors

Quality of Preventive Heath Care History


Prenatal Care History


Privatized Health Care History


Breastfeeding History


Categorical


Categorical


Categorical


0 (reference)
1 (dummy)

0 (reference)
1 (dummy)

0 (reference)
1 (dummy)


Adequate Prenatal Care
Inadequate Prenatal Care

Privatized Health Care Only
Public Hospital or Both

Adequate Breastfeeding
Inadequate Breastfeeding


Source: Demographic and Health Survey, Trinidad and Tobago,
Sample Population= 1,082


1987 (Individual dataset)


0-16


Discrete


Scale


Scale


0.00-6.65


0-1.00


1.8656


0.7316


83.6% (905)
16.4% (177)

8.9% (96)
91.1% (986)

80.7% (873)
19.3% (209)









Demographic Factors

Ethnicity

The original TTDHS contains four categories for ethnic background: African,

Indian, Mixed, and Other as we already have seen in the previous section. For the

purposes of this study, the ethnicity variable has been restricted to only women who

belong to the two ethnic groups of East Indian and African with respective proportions of

54.5% (590) and 45.5% (492). In the multivariate analysis, East Indian coded as 0

(reference) and African coded as 1 (dummy). Proportions for each ethnic group before

introducing the restrictions for determining the sample population in this study

(excluding women who are mixed and other ethnicity) were 57.1%, East Indian, and

42.9%, African.

Type of place of residence

The respondents were categorized into two types of residence: urban and rural.

Within the sample population, woman who lived in urban areas at the time of the TTDHS

survey are 441 (40.8%), and those who lived in rural areas are 641 (59.2 %). Women who

lived in urban area are in the reference category coded 0, and those who lived in rural are

in the dummy variable coded 1. The successive data from census in TT indicate that

urban residents have been better educated than rural residents. These data also tell us the

discrepancy in levels of educational achievement between urban areas and rural areas

have gradually been narrowed reflecting the improvements in educational systems, which

was on going when the TTDHS was conducted. In relations to the role of maternal

education in child survivorship and the significant residential differences between East

Indians and Africans, urban-rural distinction may influence in determining probability of

child mortality in the society of Trinidad and Tobago.









Marital status

The TTDHS has five categories for classifying the participants' marital status. The

actual questions concerning the respondent's marital status are stratified: "have you ever

been married?" (yes or no), "are you married now?" (yes or no), "are you living with a

common-law partner now?" (yes or no), "are you having a visiting relationship now?"

(yes or not). The respondents' answers to four questions were accumulated and the

participants were categorized into five categories: never married, married, living together,

visiting relation, and widowed/divorced/separated.

Definitions of "living together" and "visiting relation" are vague. In the case of

"living together," the participant may have a common-law relation with her partner, or

the participant's partner might have another relationship outside their living place. In the

case of "visiting relationship," the participant may marry the partner in the future, or she

might be the second wife or partner. Since men who are in the prime of life tend to go

abroad for a job, the population proportion of women to men is said to be 7 to 1 or more

disproportional. These common-law type relationships instead of the legal marriage are

commonly accepted and more generalized in TT. Therefore, marital status depends on

their self-definition and it varies in how single women perceive their situation. These two

categories exist in close relation to each other and it is difficult to distinguish between

them clearly. Frequency distribution indicates that there is only one single woman in the

sample population. To simplify the analysis the single woman is excluded from the

analysis. Consequently, four categories of marital status were collapsed into three

categories, married, cohabiting/visiting relationship, and widowed/divorced/separated

with respective proportions of 58.4%, 35.2%, and 6.4%. Women who are married are









coded as 0 (reference), and other forms of marital status are dummy variables coded as 1

respectively in the multivariate analysis.

Previous investigation indicated that differences in the dominant form of marital

union reflect significant cultural distinctions between the two ethnic groups. The

considerably low marriage rate among African women compared to East Indian women

reflects the predominance of other forms of unions, cohabiting relationship and visiting

relationship, within African population. Even when we controlled for the variable of

religion, ethnic identity has a very clear association with the dominant type of marital

union (Harewood 1978, Chijiwa 2001). Therefore it is meaningful to investigate whether

married women have lower child mortality; one of the two major ethnic groups, which

has a lower probability of being married is subject to the additional disadvantages of

higher incidence of child mortality.

Socioeconomic Factors

Maternal educational attainment

The TTDHS asked questions in terms of respondents' educational attainment as to

what level of education they ever attended (highest educational level) and, within that

level (educational level for TT), how many years (education in single years). This study

uses maternal education in a single year, which has a range between 0 and 16. The mean

years of schooling is 7.70 for the sample population.

Quality of life

The information about household composition and housing quality was drawn from

questions in the TTDHS to construct an index of quality of life. The index summarizes a

household component such as consumer durables as well as access water and electricity;

therefore, it serves as a proxy measure of level of modernization and level of hygiene









within household. Because, higher maintenance of sanitation may reduce a risk of

infectious disease, thus reducing mortality (Wood and Carvalho 1988, Perz 1997), higher

level of modernization indicates better quality of life, which may suggest income level.

Quality of life is a broad concept that includes variables such as the quality of drinking

water, toilet facilities, flooring materials, accessibility to electricity, and ownerships of

such things as a VCR, television, automobile, and refrigerator. However, some of these

variables may be more valid operational definitions of quality of life than others. To

uncover the latent structure (dimension) of a set of variables, to select the factors that are

considered adequate to explain the relationships among the observed variables, and to

search for a plausibly appropriate value for each variable, the eight economic indicators

were inspected using factor analysis.

Table 3-3 presents the results of the factor analysis. The eight variables were loaded

on two factors that suggest a conclusion: Quality of life (at least as measured by the

variables in this study) is not a unitary concept but rather, it is a concept that consists of

two different underlying concepts. The numbers presented in the two columns of Table 3-

3 represent the size of the correlation between the particular variable and the underlying

factor. Hence, the correlation between the variable "toilet facility" and Factor 1 is .744.

This correlation is higher compared to the factor loading for "drinking water" (.690).

Because the factor loadings represent the degree of correlation between the variable and

the underlying factor, the loadings can be used to weight each of the variables, and then

they are combined into a composite index. The variables toilet facility, drinking water,

floor material, VCR, and Car are highly correlated with factor 1. One might interpret this

factor as the principal measure of quality of housing (in the sample population) and to









Table 3-3. Rotated component matrix for 8 variables of household composition
Loadings Communality
Variable Factor 1 Factor 2 (extraction)

Toilet Facility .744 .242 .611
Drinking Water .690 .309 .572
Floor Material .619 -.002 .384
VCR .616 .168 .408
Car .587 .115 .358
Refrigerator .222 .838 .751
Electricity .138 .825 .699
TV .143 .778 .626

Variance (%)* 39.61 15.50

Source: Demography and Health Survey, Trinidad and Tobago, 1987 (child dataset)
Note: Extraction Method: Principal Component Analysis
Rotation Method: Varimax with Kaiser Normalization
KMO and Bartllett's Test: sampling adequacy = .801
* Rotation sums of squared loadings


construct a composite index for, namely, "luxury items." Factor 2, which is highly

correlated with refrigerator, electricity, and TV, is interpreted as a measure such as

"consumer durables."

In this study, however, the result of factor analysis is regarded as a means of

determining which variables can be adequate for creating an index of "quality of life."16

Considering the moderate communality (the proportion of variance explained by factors)

for each variable, I decided to leave all eight variables in the analysis. The weight in

Factor 1 for each variable has a positive direction indicating better quality of household

composition and housing quality; in addition, Factor 1 accounts for 39.61% of the


16 First of all, factor analysis was conducted with 11 variables; in addition to 8 variables, 'has stove,' 'has
radio,' 'any family member has house/apartment,' were included, and this analysis extracted three factors.
Since these three variables negatively correlated with other variables in the analysis, they were excluded
from the following analysis.









variability of the original eight variables, which has greater variability than Factor 2

(15.50% of the variability). Hence, the weights in Factor 1 for all eight variables are

combined into one index "quality of life" instead of combining them into each index7.

The new variable has a range between 0 and 6.65, an interval of 92. The mean score of

quality of life for the sample population is 1.8656.

Health Related Factors

Each woman provided health related information on prenatal care, type of place

child born, use of immunizations, and breastfeeding about her children from the youngest

to the 5th youngest (e.g., whether or not the respondent had prenatal care when she was

pregnant the youngest child, whether or not had prenatal care when she was pregnant the

second youngest child, and so on.) Hence, all respondents have five variables for each

question. Each variable are dichotomized into 1 and 0. Since the number of children is

different among mothers, five dichotomous variables of a certain question are combined

and averaged out for creating a variable. For instance, in the case of prenatal care, if a

woman had two children and had prenatal care when she was pregnant the youngest child

(coded 1), but she did not receive prenatal care for her second youngest child (coded 0),

and then her prenatal care history is 0.5. These newly created variables are considered as

child health care histories of mothers.






17 I once constructed two variables for each factor; Factor 1 = luxury items and Factor 2 = consumer
durables. However, the means test for luxury life by ethnicity indicated that East Indians are higher in
quality of life on luxury items than Africans (sig.=.031), and the means test for consumer durables by
ethnicity showed that Africans are slightly higher than East Indians with no statistical significance. To
simplify the analyses and to enhance the magnitude of 'quality of life' in addition to the reason which all
variables have the same positive directions in Factor 1, I decided to combine all variables' loadings for
creating the index of Quality of Life.









Prenatal care

The actual question for this variable was "when you were pregnant did you see

anyone for a check on this pregnancy?" and answer categories were "no one," "doctor,"

"trained nurse," "trained midwife," "traditional birth attendant," and "other." The

frequency distribution revealed no use of a trained midwife, traditional birth attendant, or

other. I conferred with some doctors whom I met during my research in TT. They

confirmed that generally there is no difference in medical care for pregnant women

between being taken care of by doctors and by trained nurses in the case of normal

pregnancy. However, in the case of unusual pregnancy or emergency, the probability of

the incidence of miscarriage (Fetal deaths are not included in this study) and the risk of

postnatal health and child health could be higher with no doctor in attendance. Hence

three values for prenatal care are collapsed into two categories: received prenatal care

from medical doctors (1) and did not receive prenatal care from medical doctors (0).

After combining the variables of five children, 134 women have never received

prenatal care from doctors (12.4%), 905 women received prenatal care from only doctors

whenever they where pregnant their children (83.6%), and 43 women have received

prenatal care from doctors or trained nurses (4.0%). The newly constructed variable,

which indicates "history of adequacy for prenatal care" is collapsed into two categories.

Women who received prenatal care from only doctors are coded as 0 (reference), namely

"adequate prenatal care," and those who have received prenatal care from doctors but not

always are combined and coded as 1 (dummy), namely "inadequate prenatal care."

Type of place where the child is born

The variable of the type of place where the respondents' children were born is used

to measure the respondents' preference for privatized health care. The respondents were









asked, "in what type of place was your child born?" for which respondents' answer fell

into four categories: government hospital, private hospital, private home, and other. In

TT, mothers are able to receive public medical services that are cheaper than medical

facilities, including free child delivery at public hospitals. The ability of a woman to

afford a private hospital or private doctor may be considerably influenced by her

financial status. In addition to this purely economic sense, if there is a greater utilization

of private medical facilities by certain ethnic group, and if the variable of privatized

health care is statistically significant after controlling for other factors, then one can

predict that there may be a cultural factor influencing the probability of child mortality

through their own cultural preferences and tendencies. Hence, the variable for type of

place child born is dichotomized into two categories; private doctor or facility

(0=reference) versus public hospital (1=dummy).

In the same manner as the creation of the variable of prenatal care history, variables

of "in what type of place was your child born" for all children (up to the fifth youngest)

of a mother were combined and averaged out. Of the sample population of 1,082, 956

women use only public hospitals (88.4%), 96 women use only private doctor or facility

(8.9%), and other 30 women use both (2.7%). Further, the variable is collapsed into two

categories for the final analysis; one is "public hospital only or both" coded as 1

(dummy), and the other is "privatized health care only" coded as 0 (reference).

Quality of preventive health care for child

Infants and children are extremely vulnerable to epidemic diseases and infectious

illness, hence, inadequate child preventive health care directly risks child survival and a

mother's choices in health care practices influence the health and survival of the child. To

assess mothers' health practices related to child preventive health care, a new variable of









"quality of preventive health care history" was created using a number of questions about

vaccination use for children in TTDHS. Children were assigned to either use or non-use

category based on vaccination information from the child health cards, such as poliol,

polio2, polio3, DPT1, DPT2, DPT3, yellow fever, measles, and if mothers have the child

health and had tetanus. In the case mothers did not have the child health card, they were

asked to recall whether or not the child had received a specific vaccination. The variable

whether or not a health card for hospital care is possessed by mothers serves as a rough

indicator of the accessibility for hospital care and the mother's attention to her children's

health. UNICEF reported that for children without child health cards, the proportion of

vaccinations given is smaller than for children with child health card (2003a). Overall,

these questions intended to capture mothers' care for their children and relations between

actual child health and frequencies of receiving the health services including vaccination.

Questions on ten vaccinations and the child health card are asked mothers about her

children from the youngest to the fifth youngest. All variables are dichotomous; yes (1)

or no (0), which are combined into one index, quality of preventive health care history.

The newly created variable has the range between 0 and 1. Of the sample population, 103

mothers had their all children receive all vaccinations and had child health cards of all

their children (9.5%), 40 mothers had never had any of their children receive vaccination

and had no child health card (3.7%), and the rest, 939 mothers were inconsistent to

provide their children preventive health care (86.8%). The index of preventive health care

history is used as scale variable, and the mean score is 0.7316 for the sample population.

Mahabir reported that East Indians tend to have eastern medical treatment or

"folk/traditional" medical treatment (1997). Although education contributes to improved









awareness of importance in receiving vaccination appropriately, one may presume that a

woman's skills in health care practices and health orientation are have been nurtured

influenced by her family orientation and may be eventually rooted in their cultural and

ethnic origin. Hence, the level of quality of preventive health care is considered as

important measures for both influence of cultural orientation on child's health and

accessibility to modern medicine.

Breastfeeding

Breastfeeding have been found to be an important factor in infant survival, even

after controlling for other variables that affect child mortality; infants who are bottled-fed

from birth run a higher risk of health and development problems than do breast-fed

children (McCann et al. 1981, Goldberg et al. 1984). Forste et al. (2001) reported that

after controlling for socio-economic background and birth characteristics, race remained

a strong predictor of breastfeeding. If differences in breastfeeding practices between

African women and East Indian women would be observed in the sample, and if women,

who did not breastfeed, have experienced more loss of their children than their counter

part, then we can assume that one of the two major groups, which practice less adequate

breastfeeding than the other racial group is subject to the additional disadvantage of

higher child mortality.

The variable of months of breastfeeding is used for creating the variable,

"breastfeeding history." It is a discrete variable including two special codings for

inconsistent and never breastfeed. Similar to the three health related variables explained

above, mothers provided answers how long they breastfeed their children from the

youngest to the fifth youngest child, and never breastfeed is recorded as 0 and one or more

months are recorded as 1. After combining the variables of five children and calculating









the average months of breastfeeding, 209 mothers have never breastfed (9.9%), 873

mothers breastfed all their children (80.7%), and 67 mothers are inconsistent in

breastfeeding (9.4%). This variable, namely "breastfeeding history," is dichotomized into

two categories; mothers who breastfed all their children is coded as 0 (reference), namely

"adequate breastfeeding," and mothers who never breastfed or were inconsistent in

breastfeeding is coded as 1 (dummy variable), namely "inadequate breastfeeding."

Procedures of Data Analysis

The first stage of the analysis is to use descriptive statistics stratified by ethnicity to

summarize the basic characteristics of the sample. This step is meaningful for assessing

the basic relationships between ethnicity and other factors such as demographic,

socioeconomic, and health-related factors, and for simply assessing how well ethnicity

can be a significant factor and predictor in the analysis of this study. The second step will

be bivariate analyses to examine the relationships between all predictors in the later

logistic regression analyses and the variable of child mortality. First, the proportion of

women who have lost at least one child for all explanatory variables are presented,

followed by odds of woman who experienced a child loss for all variables in order to

check individual predictive power, and correlation coefficients between ethnicity and

other predictors in order to demonstrate the influence of ethnic background on each

factor.

The last stage in the analysis will consist of an examination of multivariate logistic

regression models to determine whether or not influences of mothers' ethnic background

on child mortality can be observed after controlling for all other socioeconomic and

health care factors. The Generalized Linear Model used will be the logistic regression

model where the random component is B(X) the probability that a mother, who has at









least one child in the past 10 years, has lost a child, and X is the vector of explanatory

variables, namely ethnicity = X1, place of residence = X2, unmarried (a dummy variable

of marital status) = X3, cohabitating/visiting relations (another dummy variable of marital

status) = X4, years of educational attainment = X5, quality of life = X6, quality of

preventive health care history = X7, inadequate prenatal care (dummy variable of

adequacy of prenatal health care history) = Xs, public hospital use only (dummy variable

for privatized health care history) = X9, inadequate breastfeeding (dummy variable of

breastfeeding history) = X10. The full logistic regression model has the form;

Logit(T)) = c + P1X1 + 32X2 + 33X3 + 34X4 + f5X5 + 36X6 + f7X7 + P8X8 + 39X9 + PloX0o.

The first logistic regression model of the first phase will include only mother's

ethnic background (Model 1). The second regression model introduces variables for

demographic factors, place of residence and marital status, in the equation (Model 2),

followed by the inclusion of socioeconomic factors, years of education and quality of life

(Model 3). The next is the full model introducing the four health related indicators

(Model 4). The multiple logistic regression analysis attempts to provide the combined

effects of all variables on the likelihood of mother who have experienced a child loss, and

to indicate the potential variables playing a significant role in determining the probability

of child mortality in the society of TT. A series of nested logistic regression models

systematically provide ethnic background variation for estimating direct or indirect effect

on the child mortality through other factors determining child survival.














CHAPTER 4
DATA ANALYSIS

Ethnic Differentials in Child Mortality and Its Determinants

The previous research have found that race and ethnicity as well as urban-rural

differences, marital status, education, income (in this study, quality of life), and health

related variables are important predictors of child survival chance and child mortality.

Since a major concern of this study is whether mothers' ethnic background matters to the

incident of child mortality after controlling for other predictors, sample population

distribution by all other variables are presented first. Descriptive examinations of the

dependent and independent variables by ethnicity are presented in Table 4-1, which show

how ethnicity connects with other factors predicting child mortality in this study. Reading

across each row, we can compare the proportions of African and East Indian within each

category of dichotomous independent variables and the means years of education, quality

of life, and quality of preventive health care for African and East Indian.

Demographic Characteristics

Ethnic differences in child mortality presented in the first two rows of Table 4-1

indicate that approximately 7% of the mothers have experienced a loss of their child. The

proportion of mothers with children born within 10 years who have lost at least one child

is slightly higher among African mothers (8.3%) compared with East Indian mothers

(5.6%). The differences between the two ethnic groups are moderately significant at 0.09

statistically.













Table 4-1. Characteristics of mothers with children born in last 10 years by ethnicity
All Women African East Indian

[1] Childhood Mortality(*)
Never Lost Child 93.2 % 91.7 % 94.4 %
Have Lost at Least One Child 6.8 % 8.3 % 5.6 %

[2] Type of Place of Residence**
Urban 40.8 % 55.5 % 28.5 %
Rural 59.2 % 44.5 % 71.5 %

[3] Marital Status**
Married 58.4 % 34.1 % 78.6 %
Separated / Divorced / Widowed 6.4 % 8.7 % 4.4 %
Cohabiting / Visiting Relations 35.2 % 57.1 % 16.9 %

[4] Years of Education (0-16)**
Mean 7.70 8.00 7.45
(Standard deviation) (2.300) (2.124) (2.410)

[5] Quality of Life (0-6.65)**
Mean 1.866 1.762 1.952
(Standard deviation) (1.151) (1.182) (1.119)

[6] Preventive Health Care History (0-1.00)(*)
Mean 0.735 0.745 0.720
(Standard deviation) (0.241) (0.230) (0.250)

[7] Prenatal Care History
Adequate Prenatal Care 83.6 % 83.7 % 83.6 %
Inadequate Prenatal Care 16.4 % 16.3 % 16.4 %

[8] Privatized Health Care History**
Privatized Health Care Only 8.9 % 4.9 % 12.2 %
Public Hospital or Both 91.1 % 95.1 % 87.8 %

[9] Breastfeeding History**
Adequate Breastfeeding 80.7 % 85.6 % 76.6 %
Inadequate Breastfeeding 19.3 % 14.4 % 23.4 %

Source: Demographic and Health Survey, Trinidad and Tobago, 1987 (Individual dataset)
Sample Population = 1,082
Statistical significance for the association between ethnicity and variables: (*) 0.10










Ethnic differences in demographic background are noted in the next two panels,

[2] type of place of residence and [3] marital status. Both variables are strongly

associated with ethnic background as indicated by the statistical significance ofp<0.01.

The TT population is historically considered to be divided into two groups; roughly,

Africans inhabiting urban areas and East Indians inhabiting rural areas. The finding from

the TTDHS represents this socially constructed residential difference. The majority of the

East Indian division live in rural areas (71.5%) while a greater number of the African

division live in urban area (55.5%). The distinct ethnic difference is also observed in the

marital patterns. The marriage rate of East Indian mothers (78.6%) is more than double

the marriage rate of African mothers (34.1%). The comparatively smaller marriage

proportion among African mothers reflects the predominance of cohabiting/visiting

relations within this population division; the proportion of African mothers who have

cohabiting/visiting relations is substantially higher than the comparable figures for East

Indian mothers with respective proportions of 57.1% and 16.9%. The conspicuous ethnic

differences in marital status are found controlling for religious affiliations in the separate

analysis [data not shown]. Although the influence of religion on the marital pattern are

observed, there are still clear association with the dominant type of marital status -

cohabiting or visiting relationship for African mothers, married for East Indian mothers.

This shows that ethnic identity on marital status permeates religious influence and it

could manifest a distinctive cultural notion for each ethnic group.

Urban-rural residence is regarded as a crude proxy measure for physical access to

modern health services and thus, the potential for variation in the effect of education on

child mortality risk across accessibility is expected. In considering this point and while









remembering that the majority of East Indians reside in rural areas, the East Indian

division could be considered the higher risk group in child mortality in terms of

accessibility to modem health facilities and maternal education.

Socioeconomic Characteristics

The next two variables presented in Table 4-1 are maternal socioeconomic

background, years of education and a measure of quality of life. The mean number of

years of education among African women is 8.00 and East Indian women's educational

attainment is lower than African women by 0.55 years. This difference is highly

statistically significant with p-value less than 0.01. Referring to the growth ratio of

educational attainment comparing younger generation (15-34) and older generation (35-

48) in the separate analysis, younger East Indian women have increased their educational

accomplishments by 31.8% and the comparable ratio among younger African women is

5.7% [data not shown]. The two numbers suggest that years of schooling had increased

for both ethnic groups, and East Indian women had experienced much greater relative

improvement. Consequently, the gap in the average years of education between the two

major ethnic groups decreased from 1.79 years to 0.37 year, but still East Indian women

were slightly disadvantaged in educational attainment as of the time the TTDHS was

conducted. Quality of life is another indicator that enables us to observe mothers

socioeconomic status of the mothers as an alternative of income in this study. The higher

mean of quality of life observed for East Indians (1.952) compared with Africans (1.762).

The difference is highly significant less than 0.01.

Contradictory arguments made by each ethnic group in TT over which ethnic group

is economically advantageous may be nested in the contradictions of two important

factors representing socioeconomic status; educational achievement and economic









advantage. It reflects a unique influence of ethnic backgrounds on probability of child

mortality in TT. The probability of child mortality that a mother would lose at least one

child also can be used as a measure to examine which of the two factors is a more

plausible reality expressing well-being in TT.

Child Health Care Practices

This study is also interested in whether or not mothers' ethnic background

influence health care practices. If there is an association between ethnicity and health care

factors, then interest is furthered as to what degree and how mothers' ethnic/cultural

orientation involves their health care practices related to their children's survivorship.

Preventive health care history measures mothers' practices on whether they have their

children receive vaccinations against infectious diseases and illness based on the records

of their children born within 10 years prior to the survey. The mean of preventive health

care for African women is higher than for East Indian women with respective means of

0.745 and 0.720. The difference is statistically significant moderately at 0.093. The next

indicator related to health care is prenatal care history whether or not women had

adequate prenatal care history based on the records of their children born within 10 years.

The proportions of women who had adequate prenatal care history for African mothers

and East Indian mothers are almost same with respective proportions of 16.4% and

16.3% and there is no statistical significance.

Accessibility to appropriate and necessary medical care is another important

dimension of a population's level of well-being. Mothers can receive public medical

services in TT that are cheaper than private medical services, particularly, child delivery

at public hospital is free. Therefore, one could assume that a person who can afford

medical care at private hospitals should experience a higher quality of health care. But it









may not only be influenced by their socioeconomic level. For example, their selection as

to whether to receive vaccination or western medicine, should be taken into consideration

by paying careful attention to the cultural context peculiar to the society of TT.

The sample population is cross-classified by privatized health care only versus

public hospital or both for each ethnic group (Table 4-1 [8]). Percentage for East Indian

mothers who tend to practice privatized health care within ethnic group is 12.2%, which

appears to be roughly 2.5 times larger than African mothers (4.9%). Referring to the

growth ratio of proportion of privatized health care for younger generation (15-34)

compared to older generation (35-40) in the separate analysis [data not shown], the

proportions of women for privatized health care are lower among younger age; 3.8% for

African and 9.7% for East Indian, compared to among older age; 6.7% for African and

16.7% for East Indian. The gap between the two ethnic groups slightly narrowed over

time (-4.1 point); however, a greater use of private clinics among East Indian mothers is

observed in both age groups. This situation implies that although delivery is free in public

hospitals for all women in TT, East Indian women have tended to use private clinics.

Last two panels present breastfeeding selection indicating that about 19.3% of the

mothers adequately breastfed their children. The percentage is considerably higher for

African mothers (85.6%) compared with East Indian mothers (76.6%). The difference is

highly statistically significant with p-value less than 0.01. African mothers are more

likely to have breastfed adequately compared to East Indian mothers.

Based prior literature and examination of associations between ethnicity and

maternal characteristics, East Indian mothers are more likely to have characteristics

associated with higher incidence of child mortality living in rural areas, lower









educational attainment, lower quality of preventive health care history, lower level of

breastfeeding history than are their African counterparts. However, observed in this

study, African mothers seem also disadvantageous in terms of marital status, quality of

life, and privatized health care history. Not only can we not satisfactorily determine

which of the factors is most influential on child mortality but also, whether privatized

health care is mainly influenced by purely economic condition or whether it is more

likely a the matter of cultural preference. From the findings, the socioeconomic section

however, we observed an unusual inversion of the standard relationship between two

indicators of human capital; educational and economic achievement. Both have been

considered important determinants of child survivorship. This unusual situation may

facilitate TT society to conceive an unfeasible sphere in which factors interact uniquely.

For furthering our understanding of nature of explanatory variables on child mortality in

TT, the individual effects of these variables on child mortality are examined next.

Influence of Maternal Characteristics on Child Mortality

The first column in Table 4-2 shows the proportions of mothers who have lost at

least one child within 10 years prior to the TTDHS, and mean years of education, quality

of life, and preventive health care for those mothers. The second column reports the odds

ratio of mothers who have experienced a child loss for each of the factors. The third

column presents the correlation coefficients indicating the associations and their

directions between ethnicity and all independent variables.

Influence of Demographic and Socioeconomic Factors

The proportions of mothers who have experienced a child loss within categories of

each of the demographic variables appear in panels [1], [2], and [3] in the first column of

Table 4-2. The percentage is higher for rural areas (7.0%) compared with urban areas









(6.6%). More women who have cohabiting or visiting relationships have experienced a

child loss (7.9%) than separated/divorced/widowed women (4.3%) and married women

(6.5%). These results are somewhat unexpected and contradictory to previous research,

since the support of the children's father is important for child survival and health

economically, practically, and emotionally, the category of married women is normally

expected to be the lowest disadvantaged group in child mortality. In addition, the group

of separated/divorced/widowed women who are normally considered insufficient in

maternity assistance has a relatively small proportion of mothers experiencing child loss.

However, mothers in the category of separated/divorced/widowed mothers do not seem to

associate with higher child mortality. This may be caused by the considerably smaller

number of separated/divorced/widowed mothers; 69 cases, 3.1% of the sample

population.

The individual effect of each demographic factor on child mortality is reported in

the second column of Table 4-2. With the exception of ethnicity, each of demographic

factors does not significantly influence the likelihood of being women with a child loss.

Ethnicity correlates with urban-rural differences and marital standings as shown in the

third column. 18

Education level is an important achieved human capital variable because it

covariates with economic life chances, and therefore with child health and survivorship.

The mean years of education for mothers with an experience of child loss is slightly

shorter (7.59years) compared with their counterparts (the mean for all alive =7.70 years),



18 The variable of marital status is dichotomous in the examination of correlation with ethnic identity.
Married is 0 and other two categories, separated/divorced/widowed and cohabiting/visiting relationship, are
collapsed into one category as 1.














Table 4-2. Proportions and odds for mothers who have lost at least one child and
correlations between ethnicity and maternal characteristics
.Char % of Mothers who Bivariate Correlation
Have Lost Child(ren) (Odds) with Ethnicity

[1] Ethnicity
East Indian (0) 5.6 %(*)
African (1) 8.3 % 1.53 (*)

[2] Type of Place of Residence -0.274 **
Urban (0) 6.6 %
Rural (1) 7.0 % 1.07

[3] Marital Status 0.120 *
Married (0) 6.5 %
Separated / Divorced / Widowed (1) 4.3 % 0.66
Cohabiting / Visiting Relations (1) 7.9 % 1.23

[4] Years of Education (0-16) 0.99 0.450 **
Mean 7.59
(Mean for All Alive) (7.70)

[5] Quality of Life (0-6.65) 0.79 -0.083 **
Mean 1.580 *
(Mean for All Alive) (1.887)

[6] Preventive Health Care History (0-1.00) 0.35 ** 0.051 (*)
Mean 0.664**
(Mean for All Alive) (0.737)

[7] Prenatal Care History 0.049 (*)
Adequate Prenatal Care (0) 6.2 % *
Inadequate Prenatal Care (1) 10.2 % 2.99 (*)

[8] Privatized Health Care History 0.128 **
Privatatized Health Care Only (0) 2.1 % *
Public Hospital or Both (1) 7.3 % 3.70 (*)

[9] Breastfeeding History -0.113 **
Adequate Breastfeeding (0) 4.2 % **
Inadequate Breastfeeding (1) 17.7 % 4.86 **


Source: Demographic and Health Survey, Trinidad and Tobago, 1987 (Individual dataset)
Sample Population = 1,082
Statistical significance: (*) 0.10








but the two groups are not statistically independent. The odds of mothers with a child loss

indicates that the higher the educational attainment, the lower the odds of being a mother

who has a child loss. But the difference is too small and there is also no statistical

evidence.

Another measure of socioeconomic status, mean score of quality of life, indicates

that level of quality of life for mothers with a child loss is also lower (1.580) than those

with no child loss (1.887). A t-test for independence showed that the mean score of

quality of life for mothers with a child loss and those with no child loss differs

significantly with p-value less than 0.01. The odds of having experienced a loss of child

for quality of life is 0.79 (b-coefficient = -.239) means that the higher the level of quality

of life, the lower the likelihood of child mortality. Moreover, the quality of life influences

the likelihood of child loss significantly at thep-value of 0.028. These results support the

previous studies that quality of life, which is considered a factor having an abstract

concept substituting for income and housing quality in this study and having a significant

association with child mortality.

As we observed in the previous section, there is an unusual inversion in educational

attainment and level of quality of life between the African division and the East Indian

division. The unexpected non-significant influence of education on child mortality may

be a reflection of this unique relationship. In terms of likelihood of child mortality, we

may expect that the influence of quality of life is stronger than that of education.

Health-related Proximate Factors

Lastly, this study is interested in the important inquiry involving possible

modification in the child mortality pattern by the particular role of health-related









orientation. The level of accessibility to modem health services is expected to narrow the

differences in the higher and the lower child mortality group.

The mean score of quality of preventive health care for women who have

experienced a child loss (0.664) is lower compared with that for women who have no

experience of child loss (0.737) that is exhibited in the panel [6] in the first column of

Table 4-2. A t-test for independence demonstrates that differences between the mean

preventive health care score between the two groups are statistically significant less than

0.01.

Inadequate prenatal care represents that poorer prenatal care associates with higher

probability of child mortality. The group of mothers who are inadequate for prenatal care

has a larger proportion of mothers with a child loss (10.2%) compared to their

counterparts (6.2%). This association is moderately significant at 0.071. The odds of

having a child loss for mothers having received inadequate prenatal care is 2.99 times

greater than that for mothers having received adequate prenatal care, and is moderate

statistically significant at 0.09. The public hospital category composes 7.3% of mothers

with a child loss, and the odds of using public hospital only for mothers with a loss of

child is 3.70 times larger than that of their counterpart with a moderate statistical

significance at 0.071. Breastfeeding history, which showed a strong association with

ethnic identity in the previous section, also presented a significant association on the

likelihood of child mortality. The proportion of mothers with a child loss within the

group of inadequate breastfeeding is 17.7% that is larger that of adequate breastfeeding

(4.2%). As clearly shown in the second column, inadequate breastfeeding is 4.86 times

higher probability of child mortality than adequate breastfeeding. The association is









highly statistically significant less than 0.01. Hence, mothers in higher quality of

preventive health care, adequate prenatal care, privatized health care, and adequate

breastfeeding appear to have characteristics of lower child mortality. These results

indicate that generally favorable treatments for the pregnancy and delivery correspond

with a greater chance of child survival.

While ethnicity, quality of life, and health related factors demonstrate significant

influences on child mortality, type of place of residence, marital status, and years of

education seem to have no influence on child mortality. A review of previous studies

provides substantial evidence of causal connections in the level of strength among child

survival and geographical settings concerning access to basic health resources, services

and maternal education. Particularly, education has been stressed in playing a pivotal role

in decreasing child mortality. However, overall, the level of educational attainment itself

could not explain it conclusively. Instead, specific knowledge and awareness of

appropriate maternity and child health care, rather than formal education, may be more

proximate factors of child survival. The degree of strength in causal connections would

also depend on the geographical location of the nation and the extent of the public

transportation system within the nation. Living on small and relatively wealthy islands in

the Caribbean, TT society has been able to develop an extensive cheap private

transportation system. If people have fairly equal access to the modern health services,

the differences between African and East Indian in use of preventive health care, prenatal

care, privatized health care, and decision to breastfeed can be considered a purely ethnic

conventional preference influenced by their community and their culture.









Multivariate Logistic Regression Models

I presented bivariate analyses for understanding the individual relationship between

child mortality and each of independent variables in multivariate analyses. Some findings

in terms of rural-urban differences, marital status, and educational attainment did not

support previous research. Simultaneously, there are paradoxes when the associations

between ethnic background and other mothers' characteristics were employed. To

examine the net of these various factors' influence on the probability of child mortality,

the relationships between child mortality and indicators by means of the multivariate

logistic regression analysis, which allows us to obtain more definite comparisons between

the two ethnic groups in terms of child mortality is explored next.

Ethnic Influence on Child Mortality

Table 4-3 presents five nested logistic regression analyses. Model 1 includes only

ethnicity. The coefficient for ethnicity is .428. By taking the antilog of the coefficient of

ethnicity, the probability that African mothers have lost at least one child is 1.534 times

more than East Indian mothers. Demographic factors and socioeconomic factors are

introduced into Model 2 and Model 3 respectively. The coefficients for ethnicity in the

two models indicate that the East Indian division consistently has a lower probability of

having experienced a child loss than the African division after controlling for

demographic factors in Model 2, and after controlling for socioeconomic factors in

Model 3. Model 4 includes all explanatory variables. The odds ratio of Africans increases

to 2.177.

The non-significant factors are removed from the model to produce the best fit with

the fewest variables, and this final model is presented in the last column. Differences

between the chi-squares reported for Model 4 and Model 5 are not significant (/2=2.727









with df-3, p>0.250) indicating that the simple model fits the data as well as the full

model. Not only can ethnicity be considered as a powerful indicator of the probability of

child mortality in this society indicated as p-value = 0.014, but also its Exp(b) for

Africans keep its value as 2.026 accounting the probability having experienced a child

loss in this model.

The crucial question for this study if ethnic background persists in child mortality

controlling for socioeconomic, demographic, and health-related factors in the TTDHS -

is answered; ethnic background is statistically significant. The East Indian division

appears to have a lower probability in child mortality than the African division after

controlling for maternal characteristics. As variables are included in order, the differences

between the two divisions seem to widen and strengthen.

Urban-Rural Setting, Marital Status, and Scio-economic Influences

Demographic factors, place of residence and marital status are introduced in Model

2. After controlling for ethnicity and marital status, mothers living in rural areas are

21.1% more likely to have experienced a child loss compared with their counterparts.

After controlling for ethnicity and place of residence, separated/divorced/widowed

mothers are 76.3% (e.567 = 1.763) more likely and mothers who have cohabiting/visiting

relationships are 0.1% more likely having had a child loss compared with married

women. But the influence of disadvantage for rural settings and forms of union (married

and separated/divorced/widowed) in child mortality is no statistically significant.

Model 3 introduces socioeconomic variables of years of education and quality of

life. Mothers who have a higher educational attainment are less likely to have

experienced a child loss, but this is not statistically significant. Instead, the second socio-

economic variable, quality of life statistically significant, therefore, mothers who have a














Table 4-3. Probability of having lost at least one child controlling for demographic, socioeconomic, and health care factors
(Logistic Regression)


Model 1
b coef. Exp(b)

-2.612 **


.428 (*) 1.534


Variables/Values


Constant
Ethnicity
East Indian (ref.)
African
Place of Residence
Urban (ref.)
Rural
Marital Status
Married (ref.)
Separated / Divorced / Widowed
Cohabiting / Visiting Relations
Years of Education
Quality of Life
Preventive Health Care History
Prenatal Care History
Adequete Prenatal Care (ref.)
Inadequate Prenatal Care
Privatized Health Care History
Privatized Health Care Only (ref.)
Public Hospital or Both
Breastfeeding History
Adequete Breastfeeding (ref.)
Inadequete Breastfeeding


-2Log likelihood 536.684
2 (df) 3.146(1)
Model p -value .076
Source: Demographic and Health Survey, Trinidad and Tobago, 1987 (Individual dataset)
Sample Population = 1,082
Statistical significance: (*) 0.10



Model 2 Model 3
b coef. Exp(b) b coef. Exp(b)

-2.89 ** -2.393 **


.499 (*) 1.647


.191 1.211


-.567
.001


.484 (*) 1.623


.078 1.082


.567 -.692
1.001 -.115
-.029
-.232 *


.501
.891
.733
.793


Model 4
b coef. Exp(b)

-1.521 *


.778 (*) 2.177


Model 5
b coef. Exp(b)

-1.079 *


.709 2.026


.230 1.259


-1.357 (*)
-.140
-.016
-.217 (*)
-.983 *


-1.367 (*) .255
-.106 .899


-.266 *
-.977 *


1.477 4.378 1.491 4.444


.922 2.514


535.019
4.810 (4)
.307


530.654
9.176 (6)
.164


1.718 ** 5.574
476.683
63.147 (10)
<.001


1.713 ** 5.548
479.410
60.420 (7)
<.001









lower level of quality of life potentially have a higher risk of child loss. After controlling

for ethnicity, place of residence, marital status, years of education, when quality of life

decreases by one point, the probability to have lost a child is multiplied by 1.26 (e0.232).

Model 3 is not significantly stronger than Model 2 (4.366 with df-2, 0.25>p>0.1).

However, quality of life seems to be a valuable determinant for child mortality

directly/indirectly, combining the findings in the bivariate analysis and the multivariate

analysis. Since the level of quality of life appears to be higher for the East Indian

division, both ethnic background and economic situation work towards East Indian

mothers preferably.

Influence of Health-related Factors on Child Mortality

Model 4 assesses the effect of ethnic background on child mortality by adding

health-related factors. In this model all factors are included. Two findings stand out in the

model. Ethnicity achieves its statistical significance at 0.01, and odds ratio for African

mothers increases from 1.623 in Model 3 to 2.177 in Model 4. This implies that the effect

of ethnic differences with socio-demographic factors only and the combined effect

including health-related factors account for approximately 55% of the child mortality

difference between Africans and East Indians.

The inclusion of four health related factors adds a significant strength in the

inclusive model. Differences between the chi-squares reported for Model 3 and Model 4

are highly significant (V2=51.147 with df-4, p<0.001). Effects of the health-related

indicators on the probability of having a child loss follow the findings in the bivariate

analysis. The odds for dummy variables for prenatal care and privatized health care show

that mothers who have received adequate prenatal care, who have practiced privatized

health care, and who have breastfed adequately are less likely to have experienced a loss









of their child with respective odds of 4.378, 2.514, and 5.574. Prenatal care and

breastfeeding are statistically significant less than 0.05 but privatized health care has no

statistical significance. The probability having experienced a loss of child decreases with

a increase of quality of preventive health care score; for women who have a higher score

of quality of preventive health care are less likely to have had a child loss; when

preventive health care increases one point, child mortality decreases by 2.67 (e983). This

is statistically significant less than 0.01.

With the exception of years of education, the inclusion of health-related factors

increases the effects of ethnicity, urban-rural settings, marital status, and quality of life,

indicate that they are not working through health-related factors. On the other hand,

educational difference is weakened by the inclusion of health-related factors suggesting

that education potentially works through health-related factors to influence child

mortality. Notably, ethnic background continues to have an independent effect on child

mortality. To examine the effect of ethnicity further, the interaction terms between

ethnicity and the variables in the simplified model (Model 5) was conducted separately,

however, none of the interaction terms were significant suggesting that the factors

influencing the probability of child mortality in reported in Model 5 do not vary by ethnic

background.














CHAPTER 5
CONCLUSION

The purpose of this thesis has been to explore the economic and cultural differences

between African women and East Indian women in TT by means of examining ethnic

differentials in child mortality. There are a variety of factors relating to child mortality

and to each other as well. All the factors are considered to be affected by mothers' socio-

economic standings hence, the issue of child mortality has an aspect of being socially

determined similar to the context of ethnic issues.

Discourse concerning inequality in socio-economic standings is split along ethnic

lines in TT. Differences in the historical experiences of each population of Africans and

East Indians and the development of ethnic identity through discriminatory relationships

between old-timers and new-comers are found in the contemporary representation of

political identity, socio-economic position, and residential isolation. As such, an unusual

association between educational achievement and economic advantage is considered due

to the unique distribution of political power and economic power sharing between the

African division and the East Indian division. The juxtaposition of strikingly different

ethnic group identities has resulted in a construction of crossing perceptions about the

inequalities in general socio-economic standings toward others.

Ethnicity in the Analysis of Child Mortality

In order to explore ethnic inequalities in the health context, as a vital human capital,

this thesis made use of the variable of child mortality constructed from the TTDHS data.

Quantitative analysis in this study included three clusters of child mortality factors:









demographic, socio-economic, and health-related factors. The findings are summarized as

follows. First, Africans and East Indians were distributed quite differently in terms of

residential and marital patterns. The ethnic difference in place of residence followed the

historical evidence that the majority of African women reside in urban areas while the

majority of East Indian women appeared to be rural residents. Also African women

markedly differ from East Indians in terms of marital status. African women are more

likely to have cohabiting or visiting relationships, while most East Indian women secure

matrimonial relationships. The proportion of women who are married in the East Indian

division is as twice as much as that of the African division.

Second, an important contrast emerged in the socio-economic variables. The

findings demonstrate manifest relationships between the level of educational attainment

and the economic standings peculiar to the historical context of the TT society. African

women had a higher level of educational attainment however, they had a lower score on

quality of life compared to East Indian women. The differing livelihoods secured by the

women in the African division compared to East Indian division women reflected the

historical background held by each ethnic group.

Third, in the cluster of health service use, strong differences were also found in

terms of privatized health care history and breastfeeding history. East Indian women

tended to use privatized health care compared to African women of whom less than 5%

women use only private facility for delivery. In breastfeeding selection, African women

are more likely to have breastfed compared to East Indian women. They also differed in

preventive health care use. Proportionally, African women are more likely to have their

children vaccinated compared to East Indian women, however, the difference is not so









striking. Lastly, there was no evidence that these women are different in terms of prenatal

care use. This result corresponds to the high attainment of prenatal care in TT.

Unfortunately, there is no previous record to provide the information about the ethnic

differences between Africans and East Indians,

The results illustrated the heterogeneity between Africans and East Indians. Though

the findings did not always indicate specific ethnic group advantages to maternal and

child health, ethnic differences in residential and marital union patterns manifest that

ethnic identity is historically and culturally constructed in the livelihood of Trinibagonian

mothers. The unusual situation in which the wealthier East Indians are less educated

compared the economically disadvantaged Africans may contribute to level off the socio-

economic distinction between the two ethnic groups. At the same time, it may also cause

an unclear relationship between health care factors and ethnic background.

The second section of analysis presented the independent effect of each variable.

How each of the mothers' characteristics proportionally differentiates the child mortality

is summarized as follows. First, the association between ethnic background and child

mortality was observed. Africans have the larger proportion of mothers having

experienced a child loss compared to East Indians. Second, the variations in demographic

factors influence the slight difference in child mortality. But the associations are not

always in the expected direction. Rural settings disfavor mothers in child mortality,

which supports previous research. Though the maternal union patterns did not follow the

common consensus of researchers, the married status did not seem to be the favored

marital status, instead, the groups of mothers who have been separated, divorced, or

widowed appear to compose the smallest proportion of which mothers have experienced









a child loss. Third, years of school also did not appear to be a significant variable in

determining child mortality, which is dissimilar to the previous findings in child mortality

studies. On the other hand, the quality of life indicator showed its strength in influencing

the child mortality. In TT, urban-rural distinction, marital status, and level of educational

achievement do not seem to differentiate the child mortality.

Fourth, health related factors indicated that higher standards of maternal and child

health care show potential for reducing child mortality. The presence or absence of use in

preventive health care, prenatal health care, and breastfeeding, as well as the delivery at

private facility evidently yield the disparity in child mortality. An examination of the

independent effect of each variable on child mortality showed that mothers who are of

East Indian descent have higher scores of quality of life and have had a child exclusively

at a private facility, both of which indicate affordability to access a better quality of

health care. On the other hand, African women have characteristics of women who have

no child loss in terms of better preventive health care history and adequate breastfed.

In sum, referring prior research on determinants of child mortality, African women

are more likely to have a lower risk of child mortality in terms of urban settings, higher

education, higher score of preventive health care history, adequate prenatal care, and

breastfeeding history, while East Indian women have been shown to have a lower risk of

child loss in terms of married status, higher score of quality of life, and privatized health

care history. The economically advantaged ethnic group, East Indian, is found to be lower

usage of appropriate health care.

The final analysis presents logistic regression models that elaborate the findings in

the previous two analysis sections. The findings continuously showed that East Indian









women had experienced lower levels of child mortality than African women after

controlling for demographic, socio-economic, and health related factors. There is strong

evidence that ethnic background differentiates the child mortality as the ultimate outcome

of Trinibagonian mothers' livelihood. Quality of life played a significant role throughout

all models. The roles of health-related factors, with the exception of privatized health

care history, were strongly significant. Specifically, breastfeeding and prenatal care use

are very important factors to reduce the risk of child death. In the meantime, ethnic

background conspicuously appeared to be a unique and significant factor related to child

mortality. The findings indicate that ethnic identity pervades each of the variables.

Characteristics of each of the factors relating to ethnic identity accumulated as clusters

were added to the equation. This finding supports the previous research regarding the

steadfast causal pathway of household socio-economic standings impacting child survival

chances. On the other hand, ethnic background somehow plays a significant role to

determine child survival chances. Interestingly, here ethnicity seems to behaving

somewhat independently; although analyses do control for all other factors.

The effect of ethnic background on child mortality gains its strength every time that

the equation includes other variables with the exception of Model 3 in which socio-

economic factors were included. Comparing Model 2 and Model 3, the effect of ethnic

background on child mortality was slightly reduced from b-coefficient of 0.499 to that of

0.488, or reduced the association between ethnic background and child mortality by 3.7%

(the odd ratio increased from 1.647 to 1.623). This result implies that the socio-economic

factors contribute to reducing the risk of child mortality; i.e., the ethnic difference in

child mortality is dependent upon the distribution of socio-economic factors. However,