• TABLE OF CONTENTS
HIDE
 Front Cover
 Table of Contents
 List of Tables
 Foreword
 Introduction
 Objectives
 Previous work
 Theoretical and methodological...
 Empirical framework
 Data
 Empirical results: Expenditure...
 Empirical results: Nutritional...
 Summary, conclusions, and policy...
 Appendices
 Reference
 Back Cover














Title: Impact of socioeconomic characteristics on food expenditure pattern and adolescent nutritional status among low-income Florida households
CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00026902/00001
 Material Information
Title: Impact of socioeconomic characteristics on food expenditure pattern and adolescent nutritional status among low-income Florida households
Physical Description: Book
Publisher: Agricultural Experiment Stations, Institute of Food and Agricultural Sciences, University of Florida
Publication Date: 1983
Copyright Date: 1983
 Record Information
Bibliographic ID: UF00026902
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: ace7368 - LTUF
10680670 - OCLC
000401520 - AlephBibNum

Table of Contents
    Front Cover
        Page i
        Page ii
    Table of Contents
        Page iii
    List of Tables
        Page iv
        Page v
        Page vi
    Foreword
        Page vii
        Page viii
    Introduction
        Page 1
    Objectives
        Page 2
    Previous work
        Page 3
        Page 4
    Theoretical and methodological framework
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
    Empirical framework
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
    Data
        Page 20
    Empirical results: Expenditure component
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
    Empirical results: Nutritional component
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
    Summary, conclusions, and policy implications
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
    Appendices
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
    Reference
        Page 91
        Page 92
    Back Cover
        Page 93
        Page 94
Full Text

3?
December 1983 Bulletin 837



Impact of Socioeconomic Characteristics on
Food Expenditure Pattern and Adolescent
Nutritional Status among Low-Income
Florida Households


M. Moussie, C. G. Davis, L. B. Bailey, P. A. Wagner,
H. Appledorf, J. S. Dinning, and G. J. Christakis


















Agricultural Experiment Stations
Institute of Food and Agricultural Sciences
University of Florida, Gainesville
F. A. Wood, Dean for Research













IMPACT OF SOCIOECONOMIC CHARACTERISTICS
ON FOOD EXPENDITURE PATTERN AND ADOLESCENT
NUTRITIONAL STATUS AMONG LOW-INCOME
FLORIDA HOUSEHOLDS



M. Moussie, C.G. Davis, L.B. Bailey, P.A. Wagner,
H. Appledorf, J.S. Dinning, and G.J. Christakis
















M. Moussie is Assistant Professor of Agricultural Economics, Lincoln Uni-
versity, Missouri. He was formerly Research Associate, University of Florida.
C.G. Davis is Professor of Food and Resource Economics, University of Florida.
L.B. Bailey and P. A. Wagner are Associate Professors, H. Appledorf (deceased)
was Professor, and J.S. Dinning is Research Scientist of Food Science and Human
Nutrition, University of Florida. G.J. Christakis is Professor of Nutrition and
Medicine, University of Miami School of Medicine.
This report is one of a number of studies completed under Grant No. 5901-
0410-8-0122-0 from the Competitive Research Grants Office, Science and Educa-
tion Administration/U.S. Department of Agriculture.









CONTENTS
Page
FOREWORD .................................................. vii
ABSTRACT ............ ............................. .......... vii
INTRODUCTION: SOCIOECONOMIC DIMENSIONS OF
UNDERNUTRITION AND MALNUTRITION .................... 1
OBJECTIVES ....................................... .... ...... 2
PREVIOUS WORK ............................................. 3
THEORETICAL AND METHODOLOGICAL FRAMEWORK ....... 5
Theory ............... ......................... .. 5
Consumer Demand Theory ................................... 5
Household Economic Theory .................................. 7
Lancaster's Consumer Goods Characteristics Approach .......... 7
Becker's Approach .......... .............................. 8
Bagali's Nutritional and Nonnutritional Components of
Demand for Food ............ ....................... 10
Conceptual Framework of the Present Study ...................... 11
EMPIRICAL FRAMEWORK ................................... 13
Model .................. .......................... 13
Independent Variables ................ ........................ 14
Economic Factors ............................................ 14
Income .......................................... 14
Food Supplement Income Transfer ............................ 14
Noneconomic Factors ................ ....................... 14
Fam ily Size ................................ .... ......... 14
Educational Level ................ ........................ 15
R evidence ................................... ............ 15
Ethnicity ...................................... .......... 15
Nutrition Education Program ............... ................ 15
Employment Status of Homemaker ............................ 15
Age of the Homemaker .......................... .......... 16
Hypotheses ....................... .................... 16
Functional Form: Statistical Model Specification ................... 17
Food Expenditure Functional Form ............................. 17
Biochemical Nutrient Functional Form ........................... 19
DATA ................... .............................. .. 20
EMPIRICAL RESULTS: EXPENDITURE COMPONENT ........... 21
Descriptive Analysis ............................. ............ 21
Household Income and Poverty Characteristics ................... 21
Household Size and Other Sociodemographic
Characteristics ................................ ........ 24
Food Expenditure and Other Household Expenditures ............. 27
Food Expenditure Regression Analysis ............................. 34
Income Elasticity and Expenditure Propensity
C characteristics ..................... ............. ........ 34
Household Size and Other Socioeconomic Characteristics .......... 37


iii








EMPIRICAL RESULTS: NUTRITIONAL COMPONENT ............ 39
Descriptive Analysis ................... .......................... 40
Incidence of Nutrient Deficiency ................................ 40
". 1.. u Mean Nutrient Differences ............................... 45
Nutrient Regression Analysis ........................ ........... 48
Serum Folacin .................................... ..... 53
Serum Iron ................ ................... .......... 54
Red Blood Cell Folacin ............. . ................... .. 56
H em oglobin .................................... ...... 56
P ro tein ....................................... ......... 57
Vitamin C .................................................... 58
V itam in B12 ............. .............................. 58
Zinc ................ ............................. 58
SUMMARY, CONCLUSIONS, AND POLICY IMPLICATIONS ...... 59
APPENDICES ............ ................ ................ ... 69
REFERENCES ................................... .......... 91



LIST OF TABLES
Table Page
1 Summary of mean household values, selected socioeconomic variables,
Miami and Sumter County, Florida, 1980 .................... 22
2 Estimated poverty incidence among households by race and location,
Florida, 1980 ................................... ....... 24
3 Family size distribution among households by race and location, Florida,
1980 ................................................. 25
4 Mean monthly household expenditures by type and selected sociodemo-
graphic characteristics, Miami and Sumter County, Florida, 1980 28
5 Mean monthly food expenditure by income and food group category,
Miami and Sumter County, Florida, 1980 .................... 31
6 Monthly food expenditure as proportion of monthly income, by selected
socioeconomic characteristics, Miami and Sumter County, Florida,
1980 .............. ................... ...... .......... 32
7 Statistical summary of OLS monthly food expenditure equation, all
households by selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980 .............................. 35
8 Statistical summary of OLS monthly food expenditure equation, by
household categories and selected socioeconomic characteristics,
Miami and Sumter County, Florida, 1980 .................... 36
9 Statistical summary of food expenditure marginal propensity and income
elasticity, by selected household characteristics, Miami and Sumter
County, Florida, 1980 ..................................... 38
10 Incidence of adolescent nutrient deficiency by race and sex, Miami and
Sumter County, Florida, 1980 ............................ 41


iv








11 Incidence of adolescent nutrient deficiency by race and region, Miami and
Sumter County, Florida, 1980 .............................. 42
12 Mean nutrient differences among adolescents, by race and region, Miami
and Sumter County, Florida, 1980 .......................... 46
13 Mean nutrient differences among adolescents, by race and sex, Miami and
Sumter County, Florida, 1980 .............................. 46
14 Mean nutrient differences among adolescents, by sex and region, Miami
and Sumter County, Florida, 1980 .......................... 47
15 Mean nutrient differences among black adolescents, by sex and region,
Miami and Sumter County, Florida, 1980 .................... 47
16 Statistical summary of OLS nutrient equation, by nutrients and dummy
interaction variables, Miami and Sumter County, Florida, 1980. 49
17 Statistical summary of OLS nutrient equation by nutrients and selected
socioeconomic characteristics, Miami and Sumter County, Florida,
1980 .............................. ...... ... ............ 50
18 Statistical summary of OLS nutrient equation: serum folacin (SF), by
households and selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980 .............................. 54
19 Statistical summary of OLS nutrient equation: serum iron (IR), by house-
holds and selected socioeconomic characteristics, Miami and Sum-
ter County, Florida, 1980 ................................ 55
20 Statistical summary of OLS nutrient equation: red blood cell folacin (RF),
by households and selected socioeconomic characteristics. Miami
and Sumter County, Florida, 1980 .......................... 56
21 Statistical summary of OLS nutrient equation: hemoglobin (HGB), by
households and selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980 ............................ 57
22 Statistical summary of OLS nutrient equation: protein (PRTN), by house-
holds and selected socioeconomic characteristics, Miami and Sum-
ter County, Florida, 1980 ................................ 58
23 Statistical summary of OLS nutrient equation: Vitamin C (VITC), by
households and selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980 .............................. 59
24 Statistical summary of OLS nutrient equation: Vitamin B12 (VITB12), by
households and selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980 .............................. 60
25 Statistical summary of OLS nutrient equation: hair zinc (HAIRZN), by
households and selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980 .............................. 61








V








LIST OF APPENDIX TABLES
Tables Page
A-1 Poverty income guidelines for all states except Alaska and Hawaii, 1980 69
B-I Average monthly expenditure by type of expenditure and race, Miami
and Sumter County, Florida, 1980 ........................ 69
B-2 Statistical summary of OLS monthly food expenditure equation by
household category and selected socioeconomic characteristics,
Miami and Sumter County, Florida, 1980 .................. 70
C-1 Function of, problems associated with deficiency of, and food sources of
selected nutrients ................................... 72
D-1 Statistical summary of OLS nutrient equation by household category,
dependent variable, serum folacin (SF); Miami and Sumter
County, Florida, 1980 ................................... 74
D-2 Statistical summary of OLS nutrient equation by household category,
dependent variable, serum iron (IR); Miami and Sumter County,
Florida, 1980 ...................................... ..... 76
D-3 Statistical summary of OLS nutrient equation by household category,
dependent variable, RBC folacin (RF); Miami and Sumter
County, Florida, 1980 ................................. 78
D-4 Statistical summary of OLS nutrient equation by household category,
dependent variable, hemoglobin (HGB); Miami and Sumter
County, Florida, 1980 ................................... 80
D-5 Statistical summary of OLS nutrient equation by household category,
dependent variable, protein (PRTN); Miami and Sumter
County, Florida, 1980 ................................. 82
D-6 Statistical summary of OLS nutrient equation by household category,
dependent variable, Vitamin C (VITC); Miami and Sumter
County, Florida, 1980 ................................... 84
D-7 Statistical summary of OLS nutrient equation by household category,
dependent variable, Vitamin B12 (VITB12); Miami and Sumter
County, Florida, 1980 ................................... 86
D-8 Statistical summary of OLS nutrient equation by household category,
dependent variable, hair zinc (HAIRZN); Miami and Sumter
County, Florida, 1980 ................................. 88














vi













FOREWORD

This study represents a pioneering piece of interdisciplinary research in
the area of human nutrition. The disciplines represented are economics,
food science and human nutrition, and medicine. Many individuals and
institutions gave unselfishly of their time and resources to make this study
possible. The results have been made more meaningful for our knowl-
edge base by the combined efforts of these individuals. Sincere apprecia-
tion is extended to these individuals and institutions.


ABSTRACT

A multiple regression economic model was developed and used to
evaluate: (a) the relationship between Florida low-income household
socioeconomic characteristics and food expenditure behavior, (b) re-
sponsiveness of household adolescent's nutritional status to variations in
selected socioeconomic factors, and (c) appropriate food and nutrition
policy implications from the empirical findings. Adolescent nutritional
status was evaluated by eight biochemical parameters: serum folacin,
serum iron, red blood cell folacin, hemoglobin, vitamin C, protein,
vitamin B12, and zinc.
Empirical findings suggested that nutritional education played a key
role in increasing food purchasing power and nutritional status of adoles-
cents in low-income households. Although household income, household
size, and Food Stamp Program (FSP) participation had a significant
positive impact on the level of food expenditure, there was no evidence
that these variables significantly influenced the overall nutritional status
of adolescents as measured by the eight nutrient indicators. The highest
incidence of nutrient deficiency was among rural households, and rural
black households registered the highest incidence of poverty.
KEY WORDS: Multiple regression, malnutrition, poverty, food assist-
ance programs, nutrient deficiency, socioeconomic characteristics,
biochemical nutrient parameters, policy analysis.








vii














INTRODUCTION: SOCIOECONOMIC DIMENSIONS OF
UNDERNUTRITION AND MALNUTRITION

Despite increasing levels of public investment in food assistance pro-
grams, the nutritional impact of such investment on low-income house-
holds continues to be the subject of national debate. Although the
various welfare programs have reduced the incidence of poverty, there is
still no conclusive evidence whether the programs have improved the
nutritional status of low-income people [6,31]. Several empirical studies
have shown that income level and Food Stamp Program (FSP) benefits
exert a significant impact on household food expenditures [11,
13,16,18,21,26,35]. However, the general assertion that increased food
expenditure, via increased income transfer programs, has increased the
nutrient intake of low-income households is still being debated. It is
obvious that further research is needed to identify and quantify rela-
tionships between socioeconomic characteristics, food expenditure, and
nutritional status of low-income households.
Another issue pertains to the validity of policy recommendations based
on the most readily available nutritional assessment data. Most nutri-
tional assessment has used the 24-hour dietary recall survey procedure as
a source of data to determine household nutritional status. It has been
argued by many persons that this method gives conflicting and inconclu-
sive results for a number of reasons. First, it is argued that food intake
during the preceding 24-hours cannot be unequivocally taken as repre-
sentative of the food consumption habits of the household. For some
households, the recall survey might have been conducted close to a
shopping day when enough food, in terms of both quantity and quality,
was available for consumption. For others, it might have been conducted
on a day when food intake was below normal for one reason or another.
Second, as suggested by Madden et al. [19], the 24-hour' dietary recall
method is prone to over-reporting low food intakes and underreporting
high intakes. This pattern of reporting food intakes for the last 24 hours is
termed the "flat-slope syndrome" and implies a downward bias in the
number of subjects with extremely low and extremely high intakes. Thus
the validity of the 24-hour dietary recall method is a subject that needs
further research.


1. This procedure involves interviews conducted with each sample person
about his (her) total food and drink consumption during the past 24 hours.
Nutrient values are determined by measuring the respective nutrient intake from
each type of food.


1








Recently, there have been indications that the biochemical method2 of
collecting nutritional data is superior and preferable to the 24-hour
dietary recall [10,25,30,34]. This study accepts the notion that blood and
tissue specimens provide a better measurement of nutritional status than
a "one-time" food intake survey. Biochemical assessment, although
more expensive, is likely to provide a more solid basis for nutrition policy
analysis. In keeping with the demand for sound empirical analysis of the
economies of nutrition, this study seeks to identify some of the relevant
socioeconomic variables that explain low-income household food ex-
penditure behavior and nutrient levels of household members by means
of biochemical parameters. Specifically, the population of the present
study consists of adolescents from low-income households in Miami and
Sumter County, Florida. As a result of their rapid growth spurt, adoles-
cents are particularly vulnerable to nutritional problems. These problems
increase the need for sound empirical analysis of the dynamic rela-
tionships between the socioeconomic background of the families of
adolescents and the nutritional status of adolescents as a target popula-
tion.
There is a significant knowledge gap with respect to the nature and
dynamics of the links between: (a) food expenditure levels and nutrient
intake (particularly as evaluated by biochemical nutritional data), and (b)
household characteristics, food expenditure, and nutritional status. In
addition, the problem is complicated by the fact that there are inherent
dangers in generalizing the nutritional problems of the general popula-
tion, as identified in national surveys, to those of target populations.
National nutritional surveys run the risk of overlooking unique nutri-
tional problems of certain segments of the population. In order to reduce
the risk of such a problem, nutritional surveillance is necessary for
specific populations as a basis for effective policy analysis and program
implementation [34].

OBJECTIVES
The general objective of this study was to determine the impact of
selected socioeconomic characteristics on low-income household food
expenditures and the nutritional status of adolescents from these house-
holds in urban and rural areas of Florida. Specific objectives were to
determine:
(a) The interrelation between household socioeconomic characteris-
tics and food expenditure pattern.
(b) The impact of household socioeconomic characteristics on the
nutritional status of adolescents within the household.

2. In this case, nutritional status is determined from biochemical measure-
ments in blood and tissue samples.

2








(c) Whether the level of household food expenditure and nutritional
status of adolescents differs significantly between race, sex, age,
and residential location.
(d) Appropriate food and nutrition policy implications resulting from
the empirical analysis.

PREVIOUS WORK
The dearth of an adequate and relevant econutrition information base
has seriously affected the ability of the U.S. government to develop and
implement effective food and nutrition policies and programs. However,
there is a small body of relevant studies that has led the way in providing
important information on the links between socioeconomic parameters
and the nutritional status of target populations. These studies are briefly
reviewed, since they provide important starting points for this study.
A 1972 U.S. Department of Health, Education, and Welfare study [33]
documented the location and distribution of hunger in U.S. populations.3
A major finding of the study was that a significant number of persons had
nutrient levels below the recommended standards for their age, size, and
weight. Nutrient deficiency was acute in low-income states, where 58
percent of the population had deficient or low values in at least one of six
biochemical parameters. In contrast, the nutrient deficiency rate was only
32 percent in high income states. The highest incidence of nutrient
deficiency was among low-income black families.
A 1974 study by the U.S. Department of Health, Education, and
Welfare [32] attempted to measure the nutritional status of representa-
tive samples of the U.S. population, and establish a continuing national
surveillance system. The study concluded that on the basis of both dietary
and biochemical parameters, there was a significant incidence of iron
deficiency in all race, sex, and age groups, regardless of income.
A few economists have undertaken pioneering research in the area of
integrated economic and nutritional analysis (econutrition analysis). In
1978, Lane [16] empirically estimated the impact of food distribution and
food stamps on food consumption and nutritional status of low-income
California households. Results showed that both the Food Stamp Pro-
gram (FSP) and the Food Distribution Program (FDP) affected nutrient
intakes. Nutritional response was associated with: (a) the increased cash
outlays made for food as a result of the income supplement of the FSP and
(b) the increased availability of food items through the FDP. All nutrient
intakes, with the exception of vitamin A and vitamin C, showed higher
values for FSP and FDP participating households, compared to non-

3. The states included in the study were: California, Kentucky, Louisiana,
Massachusetts, Michigan, New York, South Carolina, Texas, Washington, and
West Virginia.

3








participating households. Nutrient achievement levels were also found to
differ according to ethnic group, residence, family size, and the educa-
tional levels of households.
Adrian and Daniel [1] in 1976 used multiple-regression analysis to
estimate the impact of socioeconomic factors on selected nutrient intake
levels. This analysis was based on the 1965-1966 National Household
Food Consumption Survey. Although individual nutrient intake was
found not to be highly responsive to higher income levels, results of the
study indicated that income had a positive impact on the consumption of
all nutrients except carbohydrate. The results also showed that nutrient
consumption differs by race and location. For instance, farm households
consumed more of all nutrients, except vitamin A and vitamin C, than did
urban households. It was noted that black households consumed less
carbohydrates, calcium, and thiamine than did either white or other
racial types.
In another 1976 study, West and Price [35] analyzed a 1972-1973
sample of households containing 8 to 12-year-old adolescents in the state
of Washington. They used regression analysis to assess the impact of
income, assets, food program participation, and household size on food
consumption behavior. Results indicated that bonus food stamps, assets,
and length of pay period significantly increased the value of food con-
sumed. Income showed a relatively small effect on the value of food
consumed. Two explanations were given for this effect. First, the range of
observation did not include households with very low income, where food
expenditures may be more responsive to income. Second, the value of
food consumed included food from non-market sources, some of which
may be less sensitive to income changes.
In a 1972 Pennsylvania study, Madden and Yoder [18] used survey data
and multiple regression analysis to estimate the impact of the Commodity
Distribution Program (CDP) and two versions of the Food Stamp Pro-
gram (FSP) on food consumption behavior and nutritional status of
low-income families. Their findings suggested that: (a) low-income fami-
lies were most deficient in vitamin A and calcium, and least deficient in
phosphorus and protein, (b) CDP families had no better diets than
non-CDP families with similar socioeconomic characteristics, (c) FSP
participation enhanced the nutritional aspects of diets only if the families
had not received income for a period longer than two weeks, and (d) the
total value of household food purchases did not increase significantly with
FSP participation.
In a 1977 study, Neenan and Davis [21] empirically estimated the
impact of Food Stamp Program participation on low-income household
food consumption in rural Florida. Regression results showed that in-
come and family size explained a significant proportion of the variation in
food expenditure among FSP participating households and nonpartici-
pating households. For FSP households, there was a strong interaction
4








between the bonus value of the food stamps with both money income and
family size. This interaction suggested that food stamp bonus value was
more effective in increasing household food expenditure as family size
increased. Also, the food stamp bonus value was more effective in
increasing household food expenditure at lower-income levels. In a re-
lated 1979 study, Davis and Neenan [11] empirically estimated the impact
of the Food Stamp Program and a nutrition education program, the
Expanded Food and Nutrition Education Program (EFNEP) on food
expenditure and nutrient intake of rural Florida low-income households.
Findings suggested: (a) nutrient adequacy of protein, calcium, vitamin A,
and vitamin C were not consistently related to either money income or
bonus income from food stamps, (b) household participation in the FSP
and EFNEP impacted positively on total food expenditure and nutrient
adequacy of protein, calcium, iron, vitamin A, and vitamin C, (c) policies
which combined income supplement such as FSP with nutrition educa-
tion, such as EFNEP, were more effective than either program separately
in increasing the nutritional status of poor households, and (d) a joint
FSP-EFNEP participation mechanism was nutritionally superior to a
direct cash supplement or joint cash-EFNEP program mechanism for
low-income households.
In a 1979 study, Scearce and Jensen [26] used 1972-73 Consumer
Expenditure Survey data to estimate the effects of FSP on the nutrient
intake of southern low-income families, as well as the effects of other
socioeconomic factors on the availability of nutrients. The study con-
cluded that: (a) participation in the FSP significantly increased the availa-
bility of six of the nine nutrients examined, (b) family income and
educational level of the homemaker impacted positively on the amount of
nutrients purchased, (c) urban low-income families had lower levels of
purchased nutrients than did rural low-income families, and (d) without
the FSP, participating households would have had less energy (calories),
protein, calcium, iron, vitamin A, and vitamin B.
These studies represent important contributions to our understanding
of the economics of human nutrition. However, all of these studies, with
the exception of the U.S. Department of Health, Education, and Welfare
studies [32,33] were based on the 24-hour dietary recall procedure for
nutrient assessments.


THEORETICAL AND METHODOLOGICAL FRAMEWORK

Theory
Consumer Demand Theory
Traditional theory of consumer behavior is predicated on the view that
the consumer maximizes utility by direct consumption of goods and
services purchased in the market place, subject to a budget constraint.
5








The functional utility maximizing model is expressed mathematically as:
Maximize U=f(q,, q2 . qn)
Subject to
I piq, = M
i=1

where q, are goods purchased in the market, pi are their respective prices,
and M is the money income constraint. Using the first order conditions,
demand functions are derived from the above utility function and are
expressed in the functional form:
q, =f(p,,M)i =2 . .
To ensure the theoretical plausibility of empirical demand equations,
one has to make sure that the general restrictions of consumer demand
theory are satisfied [22]. The four general restrictions as stated by Phlips
[22] are homogeneity, symmetry, negativity of the own substitution
effect, and the Engel aggregation. Four methods of estimating demand
functions are identified by Bagali [3]: specific utility, traditional theory
constraints imposed on a specific form of demand function, total differen-
tials, and separability. According to Phlips, the easiest way to impose all
general restrictions simultaneously is to derive the demand equations
from a specified utility function.
The traditional consumer demand theory has been frequently modified
by economists to broaden its range of applicability. However, these
modifications have not altered the fundamental weaknesses of the tradi-
tional approach. The major weaknesses of traditional demand theory as
cited by Becker [4], Lancaster [15], and Bagali [3] are as follows.
(a) The traditional theory has generally been formulated in terms of
monetary prices and monetary income whose application has
tended to be restricted to the market sector. Decisions about the
choice allocation of nonmarket factors are not included in tradi-
tional theory. Also, the impact of such factors as family size,
ethnicity, sex, and time are not included in choice decisions.
(b) The theory lacks predictive power for all possible goods, including
new commodities and qualitative changes in commodities.
(c) All intrinsic properties of particular goods have been omitted from
the theory. Substitution and complementary relationships among
commodities are not given, but are rooted in the specific utility
functions of each individual consumer and are therefore expected
to be different for different consumers.
(d) The traditional theory does not handle explanations of changes in
consumer tastes, except as shifts in demand functions. To what-
ever extent income and prices do not explain observed behavior,


6








the explanation rests with variations in consumer tastes. However,
the traditional theory which the empirical researcher utilizes is
unable to assist him in choosing the appropriate taste proxies on a
priori grounds or in formulating predictions about the effects of
these variables on consumer behavior.

Household Economic Theory
The household approach enables us to deal with nonmarket or noneco-
nomic factors that affect consumer behavior. Schuh [28] emphasized that
recent extensions of economic theory would enable us to deal both with
conventional economic variables such as income and prices and with
nonconventional aspects of behavior such as fertility, marriage, divorce,
race, sex, schooling, and health. The new contribution to the theory
provides basically an economic theory of the family or, as it has been
called, the "New Household Economic Theory". According to Schuh,
the new theory has three essential ingredients that distinguish it from the
traditional theory:
(a) It views the household essentially as a firm or factory, with the
result that neotraditional theory of the firm enables us to under-
stand what takes place within the household. The traditional
theory, on the other hand, explains the activities of the household
in relation to its market activities.
(b) The important variable "time" is brought explicitly within the
scope of economic analysis, largely by recognizing that individuals
seldom consume only a market good, but consume some combina-
tion of that market good and some limited available time.
(c) The new theory attempts to help explain nonmarket activities that
have to do with schooling, investments in health, marriage, and
household size, to name a few.
Unlike the traditional approach, the new theory implies that the direct
consumption of goods is not the goal of the consumer, but is the means of
obtaining more basic needs, such as nutrition in the case of food con-
sumption, which become the real sources of satisfaction. Lancaster's [15]
and Becker's [4] new approach is predicated on the argument that goods
per se are not the direct objects of utility; instead, they are the means to
the final desire of satisfaction.
Lancaster's Consumer Goods Characteristics Approach.-In Lancas-
ter's approach to consumer theory [15] the utility function is defined in
terms of goods characteristics, rather than in terms of the goods con-
sumed, per se. In the traditional theory, goods which are easily quanti-
fiable are assumed to be the direct source of utility or satisfaction for the
consumer. According to Lancaster's theory, the assumption is that con-
sumption is an activity in which goods are inputs, and in which the output
is a collection of characteristics which become the arguments of the utility

7








function. A single good can have more than one characteristic, providing
multiple outputs from the consumption of a single good. Goods in com-
bination may also possess characteristics different from those pertaining
to the goods separately.
Lancaster used the following model which allows a more general
multidimensional function:
Maximize U(z)
Subject to: p'x :< m
z = Bx
x, x 0
where U is utility operating on characteristics space (C-space) and z is a
vector of characteristics. The budget constraint p'x m is defined in
goods space (G-space), where x represents quantity of goods, p the price
vector, and m income; z = Bx represents a transformation between
G-space and C-space. The matrix B, which is called the consumption
technology, represents qualitative properties of the goods; z = Bx is
linear, and B is a matrix of constants. The following important assump-
tions are made to obtain a viable model for the optimization problem.
(a) Objectivity.-All the relationships and equations are assumed to
hold for all individuals. The intrinsic properties of the goods, and
possibly the context of technological knowledge in the society, are
known.
(b) Linearity.-Each consumption activity produces a fixed vector of
characteristics, and that relationship is linear; i.e., r characteristics
are related to n consumer goods (n > r usually) by a linear fixed
technology matrix B = (bii), where bj, is the amount of the ith
characteristic in one unit of the jth good; i = 1, 2, .. n. Linearity is
assumed to simplify the problem without loss of generality.
(c) Maximization.-The consumer possesses an ordinal utility func-
tion on characteristics U(z), and will choose a situation which
maximizes U(z).
Lancaster's consumer theory has been very useful in dealing empiri-
cally with consumer durable goods and assets, new commodities, and
quality-differentiated products. The effects of product changes, advertis-
ing, and innovation in consumer goods can be analyzed in the framework
of the new theory, since the consumption technology, and not the utility
function, is expected to be affected by these factors. In the new model, a
new product simply means addition of one or more activities via con-
sumption technology. Traditional theory has limitations in dealing with
these factors.
Becker's Approach.-Becker's Theory of Allocation of Time and
Household Production Function Approach [4] has been helpful to empir-


8








ical studies regarding household economics. His theory has been applied
to a wide range of household problems such as fertility, marriage, popula-
tion growth, and investment in children.
Becker's consumer theory rests on the assumption that utility is
obtained from commodities which are produced by the consumer unit
itself through the productive activity of combined purchased market
goods and services with some of the household's own time. Goods and
services purchased through the productive system is the final output,
which is the argument of the utility function. Becker's household utility
maximization model is expressed as follows.
Maximize U = u(z1, z,, . z,)
Subject to zi = f(xi, ti), and the income constraint,

M = I pixi and a constraint on the household's available time
i-= 1
T= t+ I t
i= 1
where z, represents quantity of commodity i produced by the household
using a vector of market goods xi, and a vector of quantities of its own
time, ti; pi and xi are the price and quantity of the market good input used
in producing zi, respectively; t, and ti are the household's time spent in
the labor market and in producing z;, respectively.
The time and money income constraints were combined into a single
resource constraint on the household's "full income":
S = wT + V = X(wt + pixi)

where w is the wage rate and V is the household's nonwage income. The
utility function, therefore, is maximized subject to the constraints of
the production function, z =f(xi, ti), and resource constraint,
S = li(wti + pixi). The Lagrangian function is expressed as
L = U (zi,Z2 ... z,) [.(wti + xi) S].
i
The household production function framework gives emphasis to the
parallel service performed by firms and households as organizational
units. The neoclassical theory of the firm can easily be applied to house-
hold economics. Like a firm, a household unit maximizes its objective
function, subject to resource and technological constraints, including
labor and capital. The new approach incorporates numerous noneco-
nomic variables into the consumer demand function. This is accom-
plished by expanding the economist's theory of choice into the non-
market sector, thereby making the theory more applicable to the new
household economics. The approach has been a stimulant to a number of
studies such as the production theory of family health, children, mar-


9








riage, and schooling. An example in which Becker's production function
approach would be applicable is in the production of family health. Here,
the output can be measured in terms of nutrient values and the inputs
conceived as variables are food consumption and other socioeconomic
factors that affect the production-consumption process. For instance, the
level of education or employment can be introduced as an input in the
family health production function.
Bagali's Nutritional and Nonnutritional Components of Demand for
Food.-Bagali [3] used a model which decomposes the food demand
vector into nutritional (biogenic) and nonnutritional (social and psy-
chogenic) components within a framework of the traditional consumer
approach. The major objective of Bagali's empirical study was to develop
a method for measuring nutritional and nonnutritional components of
demand for meat products. Results of this study supported the hypoth-
eses that: (a) nutritional inefficiency exists, (b) nutritional and nonnutri-
tional components are competitive, and (c) increased food expenditures
augment the influence of nonnutritional factors. The result also showed
that higher income or food expenditure does not necessarily lead to
nutritional improvement.
According to Bagali, food purchasing behavior can be explained nei-
ther by traditional theory nor by the new approach. His argument lies on
the assumption that Lancaster's objective characteristics do not appear in
traditional theory. As such, nutritional factors could not be handled by
traditional consumer demand theory. Conversely, the new household
theory could handle objective characteristics (nutritional requirements)
but not the nonnutritional factors, which are inherently subjective. In
other words, there is no place for subjective characteristics in the tradi-
tional theory. His model was an attempt to integrate objective (nutri-
tional) and subjective (nonnutritional) characteristics into consumption
analysis.
In Bagali's study, household characteristics are the argument of the
utility function, since they explain consumers food purchasing behavior.
According to Bagali, consumption of food serves the purpose of satis-
fying two basic needs, biogenic (nutritional) and psychogenic (nonnutri-
tional).
The nutritional need is self explanatory, in the sense that it has an effect
on the physical growth, mental development, and maintenance of an
individual. The nonnutritional component, on the other hand, has histor-
ically been discounted in consumer demand theory. Bagali argues that
nonnutritional factors affecting the demand for food can be categorized
as religious, cultural, social, psychological, and sensory [3]. In many
societies, food habits are associated with religion. Certain religious and
ethnic groups do not eat pork or beef, while others are vegetarians. The
cultural habits of preparing and consuming food differ by ethnicity, sex,


10








age, and geographical area. Other social factors that influence the de-
mand for food are education and employment status of the homemaker
(mother) and the composition and size of the household. Therefore,
religious, cultural, and other social characteristics must be incorporated
into empirical analyses of consumer behavior.
The other major factor that influences the demand for food is economic
status of the household. Economic conditions influence food choices very
strongly. The purchasing power of the household is a major determinant
in the quantity and quality of diets. Income and asset ownership for the
household are the two major yardsticks of measuring the economic status
of the household.

Conceptual Framework of the Present Study
This section discusses how traditional consumer theory, the new house-
hold economics approach, and the nutritional and nonnutritional ap-
proach are applied to the present study.
Traditional consumer theory generally ignores household characteris-
tics-such as ethnic background, social class, family size, and residential
location of the household-in determining the variations in food con-
sumption habits between households of different social and economic
strata. This study attempts to include relevant household characteristics
in the general neoclassical model that describes the Engel relationship,
with no other deviation from the basic consumption theory. As such, the
study uses an extension of the neoclassical theory of consumer behavior.
This modification of the Engel curve can easily be applied in empirical
studies.
Extensive research has been done to modify the general Engel curve,4
because of the fact that the traditional consumption theory lacks applica-
bility to available data and empirical research. Allen and Bowley [2],
Prais and Houthakker [23], Lancaster [15], Becker [4], and Phlips [22]
have identified other factors, not included in the traditional model, that
affect the household's consumption habits and decisions.
In line with the above theoretical background, this study views the
household as one organizational unit. Within this framework the follow-
ing household optimization model is used to explain the interrelationship
between nutritional status, socioeconomic characteristics, and food ex-
penditure of the household.
Maximize U = U(z) subject to resource (total income) constraint,
Y= c(p, w, z, k)
where z is vector of nutrients (characteristics),

4. The Engel curve is the relationship between expenditures on a given com-
modity and the income of the household for a specified period of time.

11








U is utility,
C is cost function,
Y = pq + wlo (money income plus leisure income),
p is vector of market products,
10 is vector of labor inputs, utilized within the household
process,
w is vector of wage rates,
k is capital items (technology set).
Using Phlips' explanation on duality relation [22, pp. 27-31], the
preceding direct utility function can be altered to an indirect utility
function by substituting optimal quantities pO, wo, zo for p, w, z in
c(p,w,z,k). This transfer leads into equations stated as the minimization
of the cost function with respect to price, income, wage rate, and nu-
trient, given the respective quantities. By minimizing the cost function,
the following demand system is obtained by using the Lagrange Multi-
plier, and first and second order conditions:

qi = qi (p,w, z;k) -> demand for ith type of food items,
api

or = or (p,w,z;k) -- demand for rth type of labor,
aWr

w -= = wT(p,w,z;k) -- shadow price of jth nutrient (charac-
"azi teristic)

Restating the problem, the household nutrient optimization prob-
lem is written as:
Maximize U = u(z) (1)
subject to
Y= 'ri(p,w,z;k) zj

where rry are the shadow prices at the optimal levels of z, zj.
The above constrained maximization yields the following structural equa-
tion for Zj:

Z, = Zi (y, r) (2)
The above structural form (2) can be reduced into equations of the
following form:
Zj = aj (y,p, w;k) (3)
qi = qi (Y,P, w k) (4)
Equations 3 and 4, satisfied simultaneously, are solutions of the corre-
sponding structural forms that contain the demand functions. These
12








reduced forms are used as the basis for this empirical study. Reduced
form 3 is used to define the nutrient-household characteristics rela-
tionship, while reduced form 4 is used to describe the food expenditure-
socioeconomic characteristics relationship.

EMPIRICAL FRAMEWORK

Model
One objective of this study is to measure the impact of socioeconomic
characteristics on household food expenditure. The approaches of Lan-
caster's household characteristics [15], Becker's production function [4],
and Bagali's nonnutritional consideration [3] are used to determine the
impacts of socioeconomic characteristics on food expenditure and nutri-
tional status. Theoretical models, equations 3 and 4, are the bases of this
cross-sectional study. Following these theoretical models, the empirical
models of this study are expressed as:
Qf = f(I, FSP, HS, A, E, L, S) (5)
Nj = f(I, FSP, HS, A, E, L, S, V) (6)
where Qf is the household's monthly expenditure on food,
Nj is the value indicating selected adolescent nutrient level
(j = 1, . 8)
I is the average monthly income of the household,
FSP reflects participation of one or more members of the house-
hold in the Food Stamp Program,
HS measures the number of persons in the household,
A is the age of the homemaker,
E reflects the race of the household,
L denotes the residential location of the household,
S reflects the socio-demographic status of the homemaker with
respect to (a) the stocks of general educational attainment
(E), (b) awareness of food nutrient composition and food
purchasing and preparation efficiency (EDNT), and (c) em-
ployment (EMP)
V reflects if the household has a vegetable garden (technology
set).
The relationships expressed in equations 5 and 6 are in line with the
theoretical model zf, qf =f(y,p, w;k). Variables I and FSP explain the
economic conditions of the household. Variable HS was included in the
model to capture the effect of economies of size in food purchasing and
nutrient intake. Variable L indicates the proximity to markets and cap-
tures the cost of living effect of regional differences.
It was hypothesized that the household's food consumption behavior is
related to the household's tastes and preferences. It is assumed that tastes

13








and preferences are proxied by race, educational level, and employment
status of the homemaker. For this reason, variables A, E, and S were
included in the model to determine the effect of tastes and preferences,
which are likely to differ among households of different social strata. In
the nutrient model, variable V is included as a capital item to denote the
presence of a vegetable garden. The explanatory variables that explicitly
incorporate the socioeconomic characteristics of the household are ex-
plained in greater detail in the succeeding section.

Independent Variables
As stated in equations 5 and 6, the household's food consumption
behavior and adolescent nutrient level are influenced by socioeconomic
characteristics, subdivided into economic and noneconomic factors.

Economic Factors
Income.-The economic condition of the household is a major deter-
minant of the household's food expenditure. The quality and quantity of
food purchased mainly depend on the purchasing power of the household
[8, 16, 20, 21, 26, 35]. Income and price levels are major components of
purchasing power, which is the major economic component of the house-
hold food expenditure. In both the traditional consumer theory and the
household economics approach, income and price are key variables in
determining food consumption behavior. In this study these two eco-
nomic variables are also assumed to be major determinants of food
consumption behavior.
Food Supplement Income Transfer.-This variable is also one of the
economic factors that determine the value of food expenditure. Results
of empirical studies [1,11,16,18,35] suggest that there is a direct rela-
tionship between FSP bonus income and food expenditures. Davis and
Neenan [11] found that participation in the FSP increased household
discretionary income, which impacted positively on food expenditure and
nutrient intakes. FSP participation will therefore be evaluated as a vari-
able affecting food expenditure and nutrient levels.

Noneconomic Factors
The noneconomic variables discussed here are largely associated with
Bagali's nonnutritional factors [3]. The following characteristics are
hypothesized to be the major factors influencing food purchase and
nutrient consumption. Most of these noneconomic factors are included in
the empirical model to account for difference in tastes and preferences.
Family Size.-The number of persons in a household is expected to
affect the value of food expenditures [21]. There might even be econo-
mies of scale, since family packaged food items tend to be less expensive



14








than small size packaged food items. Most often, there are price dis-
counts when large quantities of food are purchased. This study will
attempt to assess changes in household food expenditure and nutrient
levels associated with size of households.
Educational Level.-Educational level of the homemaker is assumed
to be a source of information about health and nutrition. Persons with
higher levels of schooling are more inclined to scrutinize the product
contents of packaged foods for nutrient values. Scearce and Jensen [26]
found a positive relationship between the level of education of the female
homemaker and the nutritional status of the household. The mother or
guardian of the household in which the adolescent resides is assumed
to be the homemaker. It is expected that the educational level of this
homemaker will influence food expenditure and adolescent nutrient
levels.
Residence.-Residential location of a household is another factor that
is expected to affect consumption and nutritional status of households
[16]. Urban residents may benefit from the availability of a wider choice
of supermarkets and grocery stores, which would make possible a greater
variety of food items and choice in prices, quality, and quantity of food
items. Conversely, some rural households grow garden vegetables, which
are excellent sources of essential nutrients. The extent to which residen-
tial location influences food expenditure and nutrient status will be
analyzed.
Ethnicity.-Culture, ethnic customs, and traditions are factors which
help to explain food consumption behavior [3,11,16,21,26]. People of
different ethnic and racial backgrounds tend to prefer one kind of food
item to another. These consumption habits affect the nutritional status of
households.
Nutrition Education Program.-Nutrition education programs, such as
the Expanded Food and Nutrition Education Program (EFNEP) have
been found to be a major determinant of food expenditure and nutrient
intake of low-income households [11]. The objective of this particular
program is to help low-income households acquire knowledge about
nutrition, food economy, and meal planning and preparation.
Employment Status of Homemaker.-The employment status of the
homemaker is also one of the factors responsible for variation in food
consumption habits and nutrient status of the household [24]. For home-
makers employed outside the home there is a time constraint on food
preparation at home. In this situation, one of the following could happen:
(a) partly cooked or frozen food items are purchased so that the home-
maker takes less time to prepare home-cooked food, (b) food might be
consumed away from home, such as school, fast food restaurants, etc.
This consumption behavior will affect the value of household food ex-
penditures and the nutritional status of the adolescent.

15








Age of the Homemaker.-Younger families, particularly those with
children, tend to place a higher emphasis on food and nutrition than do
older families [18]. The type of food consumed will affect both the value
of food expenditure and the nutritional status of the adolescent.

Hypotheses
The federal government's income transfer programs were designed to
improve the nutritional status of poor people. Although increased in-
come transfers tend to increase the purchasing power of low-income
households, there is still no consensus as to whether or not these pro-
grams resulted in improved diet of poor people. This study will attempt to
answer some of the questions raised earlier by testing the following
hypotheses.
(a) The economic condition of the household is a major determinant
of variations in food expenditure. Increased income and participa-
tion in the Food Stamp Program (FSP) will impact positively on
the value of food expenditure.
(b) Families with large numbers of people in the household spend
more on food than small size households. However, the value of
food expenditure per person decreases as the size of the household
increases.
(c) Educational level and employment status of the homemaker will
impact positively on the value of food expenditure. No a priori
hypothesis is given as to how the homemaker's employment status
is going to affect the nutritional status of the adolescent. The
general educational level of the homemaker alone will not have a
beneficial effect on the dietary level of the adolescent within the
household. Special nutrition educational programs will have a
positive influence on the nutrient intake of the adolescent.
(d) Age of the homemaker will have a negative relationship on both
the value of aggregate food expenditure and the nutrient intake of
the adolescent. Older people tend to have difficulty in overcoming
traditional food patterns and generally place less emphasis on
nutritious food than do younger people.
(e) Generally, rural households who have less access to diverse types
of food stores are more malnourished than urban dwellers. The
difference of adolescent malnourishment between the two regions
is expected to be minimal, since rural households tend to consume
more fresh food items (such as garden vegetables) than do urban
dwellers.
(f) No a priori hypothesis is given with respect to differences in sex as
associated with variations in food consumption behavior and levels
of nutrient intakes.



16








(g) Generally, ethnic groups who have easy access to occupational,
educational, and economic opportunities have a better diet than
other groups. To this effect, it is hypothesized that whites are less
malnourished than blacks and hispanics.
Test results of the above hypotheses not only explain socioeconomic
impacts on food consumption behavior and nutritional status of the
adolescent but also help in discussing food and nutrition policy formula-
tions and implementations.

Functional Form: Statistical Model Specification
The type of functional form to be selected largely depends on the type
of empirical problem and the nature of the data collected for the study. In
choosing the functional form, the theoretical plausibility of the demand
equations should be ensured. Several studies used different functional
forms to estimate household expenditure models. Allen and Bowley [2]
used a linear functional relationship to estimate Engel curves. Prais and
Houthakker [23] tried linear and non-linear relationships to generate
improved estimates. They suggested that semi-logarithmic and double-
logarithmic functions give better results for estimate of food expenditure.
Salathe and Buse [27] suggested a quadratic function in which food
expenditure is a function of income squared and the square of household
size. Phlips [22] questioned the extent to which the type of functional
forms discussed above are theoretically plausible. He discussed the trade-
off between statistical (pragmatic) results and properties of economic
theory. He addressed the need for further research to find functional
forms that are both realistic and theoretically plausible.

Food Expenditure Functional Form
For this study, a double-logarithmic functional form would be the
appropriate model to explain variation in food expenditure resulting
from variation in the household's socioeconomic characteristics. This
functional form satisfies the conditions of the Engel curve, which is
convex to the origin, satisfies the Engel aggregation through a trans-
formation technique (intrinsically linear), and has a positive intercept to
avoid a possibility of negative expenditure values [22].
Studies reviewed earlier also indicate that food expenditures vary
nonlinearly with income and family size, and level off as these explana-
tory variables reach a certain limit. Following these arguments, the
statistical model that describes the relationship between food expendi-
ture and socioeconomic characteristics is specified as:
InQf = a + B Inl + B2 nHS + B3FSPI + B4A1 + BsEi + B6E2 (7)
+ B7R1 + BsR2 + B9L1 + BloEMPI + B IEDNTI + U,



17








where lnQf= log of household's monthly food expenditure in dollars
(including food stamp value)
Inl = log of household's monthly food expenditure in dollars
(excluding food stamp value)
InHS = log of household size-total number of individuals in the
household who depend on a common pool of income
FSP = participation in the Food Stamp Program (FSPI = 1 if
one or more members of the household received food
stamps)
A = age of the homemaker (A = 1 if age is greater than 40)
E = educational attainment of the homemaker
EB, = 1 if school grade is greater than 12
E, = 1 if school grade is 9-12
E2 = 1 if school grade is less than 9
R = ethnic background of the household
Ro = 1 if hispanic
Ri = 1 if white
R2 = 1 if black
L = residential location of the household
L, = 1 if rural (Sumter)
L1 = 1 if urban (Miami)
EMP = employment status of the homemaker
EMPI = 1 if employed, 0 otherwise
EDNT = nutritional education of the homemaker
(EDNTi = 1 if homemaker has basic nutritional educa-
tion, 0 otherwise
U, = error term assumed to be distributed normally with zero
mean and constant variance.
Zero-one dummy variables are used to analyze the impact of all explana-
tory variables, with the exception of income and family size. Log of
family size is included to allow economies of size for food expenditure.
The initial categories (0 subscripts) of the dummy variables are omitted to
avoid problems of perfect multicollinearity (singularity). Specifically, the
excluded 0 subscript variables whose coefficients will be picked up by the
intercept are FSPo, Ao, Eo, Ro, Lo, EMPo, and EDNTo, representing
nonparticipation in FSP, less than 40 years old, college level education,
hispanic origin, rural households, non-employed, and no basic nutri-
tional education, respectively. The FSP and EDNT variables are spec-
ified as binary variables because of measurement problems. Ideally, FSP
participation should have been expressed in terms of the bonus value of
coupons. As a general rule respondents refused to divulge this informa-
tion. The EDNTvariable was developed from the homemaker's response
to a series of questions regarding exposure to nutrition counseling and
nutrition information sources.

18








Biochemical Nutrient Functional Form
Lancaster's characteristics approach to demand theory [15] is used as a
theoretical basis for estimating the relationship between nutrient values
and socioeconomic variables. Within this framework, food expenditures,
per se, are not the arguments of the utility function. Rather it is the
nutrient value of food consumed which provides satisfaction to an indi-
vidual. Thus, following Lancaster, Becker, and Bagali, food expenditure
is a behavior characteristic which satisfies biogenic (nutritional) need.
Therefore, values of nutrient indicators are expressed as a function of
socioeconomic variables, including income.
As to the appropriate type of functional form for the nutrient model,
no substantial study has been done to define the nutrient-socioeconomic
relationship. For this reason, this study did preliminary experimentation
with different functional forms. The preliminary findings showed that the
linear model produced a better estimate in describing the nutrient-
socioeconomic relationship. This study, therefore, uses a linear OLS
regression model to estimate the impacts of household socioeconomic
characteristics on the nutritional status of the adolescent.
Writing the above reduced forms into statistical functional form, the
following linear regression is to be estimated by OLS:
Nj=a+B1+ B2HS + B3FSP + B4A + BsE + B6E2 + B7R
+ BsR2 + B9L + B1 EMP + E B11S + B12 V1 + U
where Nj = biochemically acceptable indicator level5 of the j'h nutrient,
with = 1, 2, .. 8.
1 = serum folacin (SF)
2 = serum iron (IR)
3 = red blood cell folacin (RF)
4 = hemoglobin (HGB)
5 = total protein (PRTN)
6 = vitamin C (VITC)
7 = vitamin B12 (VITB12)
8 = hair zinc (HAIRZN)
S = sex of the adolescent
(S = 1 if adolescent is male, 0 if female)
V = if household has vegetable garden
(V1 = 1 if household has vegetable garden, 0 otherwise)
I = household's monthly income (from all sources)
HS = household's size
The notations for the other independent variables are explained in the
expenditure model.

5. See Table 10 for biochemical criteria used to establish acceptable and
deficiency levels for indicators.
19








Although each nutrient may be affected differently by various factors,
the same socioeconomic relationship is used for each nutrient. There is
no justifiable reason to modify the model or to delete or include variables
for any one nutrient. The physiological function of, and problems associ-
ated with, deficiency of these nutrient indicators are given in Appendix
Table C-1.

DATA
Sample survey instruments were used to generate household socioeco-
nomic profile and nutritional data for adolescents. Urban samples were
obtained from the inner-city area of Miami, and rural samples were taken
from Sumter County in north central Florida over the 1979-1980 period.
The inhabitants of the inner city area of Miami are predominantly
low-income blacks. This region's population is thought to be representa-
tive of urban poor minority households in Florida. Sumter County, on the
other hand, is predominantly rural with no town with a population
greater than 2500. The estimated population of the county in 1977 was
21,586. The per capital income of this county in 1977 was $4,433, which is
low compared to the averages for the state ($6,728) and nation ($7,019)
[7].
. The total adolescent sample consisted of 381 subjects-202 from urban
Miami and 179 from rural Sumter County. Two schools from each area
were selected, and a stratified random sample of adolescents (age 12-16)
were selected from each school. Each adolescent received a complete
physical examination by the survey team physician. Fasting blood sam-
ples were taken by venipuncture for analysis of nutrient indicators. These
nutrient indicators are the dependent variables in the nutrient level
model. The indicators included values from serum and red blood cell
folacin, serum iron, total protein, hemoglobin, vitamin C, vitamin B1,
and hair zinc. The samples were stratified by poverty level, sex, and
ethnic group, and contained a larger number of blacks than would have
been obtained by a random sample of the state's population.
The individual adolescent is the unit of observation for the biochemical
nutrient indicators. However, for purposes of explaining the impact of
socioeconomic factors on adolescent nutrient indicators, sample survey
data were obtained via a separate survey instrument administered to the
household units to which the adolescents belong. The total household
sample consisted of 305 units-157 from the Miami area and 148 from
Sumter County. The socioeconomic characteristics of these households
are used as explanatory variables against the adolescent levels of nutri-
tional indicators in the operational models.




20








EMPIRICAL RESULTS: EXPENDITURE COMPONENT
In this section the empirical results of the expenditure component of
the study are presented. The discussion of the findings are presented in
two parts. The first part is a descriptive analysis of the socioeconomic and
demographic characteristics of the households. A discussion of the inci-
dence of poverty and the distribution of total household expenditure by
type of household group is presented in this section. In the second part
the findings of the selected OLS regression analyses of the food expendi-
ture model are discussed. The impact of selected socioeconomic variables
on the value of household food expenditure elasticities for all families in
each housenold category are also presented.

Descriptive Analysis

Household Income and Poverty Characteristics
Table 1 presents the mean level of selected socioeconomic variables,
including average household income by race and region for all house-
holds sampled. The average monthly income for the entire sample
population (303 households) was $887 per household. There was an
average of 4.78 persons per household, which translated into an average
monthly income of $186 per person. Disaggregation of the data by race
and region shows rural white households accounting for the highest group
level income. They reported an average monthly income of $1154 per
household ($256 per person). Rural blacks reported the lowest level of
income, with an average monthly household income of $493 ($96 per
person). Cuban Americans (hispanics) had an average income of $600
per household ($124 per person).
Among rural households, the average income for a white person was
about 2.7 times the average income of a black person ($256 versus $96).
Further indication of the income gap between groups of households is
reflected in Table 2, which shows the estimated distribution of income
below and above the national poverty income levels. The numbers shown
in Table 2 are calculated by dividing the annual income of the household
by the corresponding poverty threshold of the 1980 Federal Register's
Poverty Income Guideline [12]. The guideline reports the poverty
threshold for every family size and family type (Appendix Table A-i).
The incidence of poverty (percentage of families below the poverty
threshold) was highest among rural black households and lowest among
rural white households. As shown in Table 2, 83 percent of rural black
households were below the threshold. Of this number, 56 percent could
be classified as being chronically poor, since their income was below 75
percent of the poverty level. Also, 67 percent of rural white households
could be classified as "less poor," since their average income was greater


21











Table 1. Summary of mean household values, selected socioeconomic variables, Miami and Sumter County, Florida, 1980.

Household category

Urban (Miami) Rural (Sumter)
Cell
Socioeconomic count Total Black Hispanic Black White
variable (n) n = 305 n = 128 n=24 n = 48 n = 100

---Dollars-
Monthly income 303 887.00 881.00 600.00 493.00 1154.00
(excluding FSP income) (37.7)a (54.1) (57.8) (36.5) (80.1)
Monthly food exp. 304 254.00 224.00 262.00 237.00 294.00
(including food stamp (6.97) (9.7) (26.6) (13.5) (13.7)
purchases and away from
home purchases)
Total monthly exp. 303 664.00 584.00 637.00 528.00 781.00
(16.91) (20.9) (54.4) (37.1) (34.3)
Numbers
Household size 305 4.78 4.82 4.83 5.15 4.50
(0.10) (0:17) (0.40) (0.29) (0.23)
Age (homemaker) 305 41.2 42.1 41.3 42.6 39.7
(0.52) (0.89) (1.74) (1.50) (0.20)
Education level
(homemaker) 276 11.2 11.6 6.9 11.2 11.8
(0.10) (0.24) (0.88) (0.20) (0.20)






Table 1. (cont.)
Household category

Urban (Miami) Rural (Sumter)
Cell
Socioeconomic count Total Black Hispanic Black White
variable (n) n = 305 n = 128 n=24 n = 48 n = 100

Education level (homemaker) Percent
<9th grade 12.0 10.2 65.0 4.6 5.1
(1.96) (2.92) (10.94) (3.18) (2.23)
9-12th grade 72.1 68.5 30.0 88.6 78.6
(2.78) (4.49) (10.51) (4.84) (4.17)
>12th grade 15.9 21.3 5.0 6.8 16.3
(2.21) (3.96) (5.00) (3.84) (3.75)
Employment 279 68.1 75.5 40.9 60.5 69.1
(homemaker) (2.80) (4.12) (10.73) (7.54) (4.71)
FSP participation 277 28.0 33.0 36.4 53.7 9.1
(2.70) (4.53) (10.50) (7.88) (2.90)
Nutritional education 279 41.0 62.7 78.3 16.7 16.3
(homemaker) (2.95) (4.63) (8.79) (5.82) (3.75)
Marital status 305 65.0 55.5 75.0 47.0 85.0
(married) (2.72) (4.41) (9.02) (7.02) (3.59)
Vegetable garden 295 18.6 9.9 9.1 17.4 33.3
(2.27) (2.72) (6.27) (5.65) (4.76)
EFNEP participation 278 9.7 8.1 4.6 7.1 3.3
(2.78) (2.00) (4.54) (4.02) (3.44)
"Numbers in parentheses are standard errors of mean.








than 125 percent of the threshold. Only 4 percent of the rural black
households reported an income level above 125 percent of threshold
level.
Among urban households, the incidence of poverty was highest in the
hispanic group (Table 2). Sixty-two percent of hispanic households reg-
istered an average income below the poverty income level. Of this
number, 46 percent were below 75 percent of the threshold. The preva-
lence of poverty for urban blacks was 43 percent, of which 23 percent
could be classified as chronically poor. Also, 41 percent of urban black
households could be classified as being "less poor," since their level of
income was greater than 125 percent of the poverty threshold. Only 17
percent of the hispanic households were estimated to be in the "less
poor" category. The incidence of poverty in Florida families was esti-
mated to be 10 percent for whites, 38 percent for blacks, and 18 percent
for hispanics [7, p. 136]. The findings of this study, therefore, show a
higher estimated incidence of poverty in all ethnic groups than that
reported by official sources. There are at least two possible explanations
for the higher poverty incidence found in this study. First, this study used
a more recent poverty income guideline than the official sources. The
more recent guidelines are adjusted to account for inflationary changes.
Second, the households and regions selected for this study were selected
on the basis of observable low-income characteristics. It is, therefore,
expected that the incidence of poverty for these households would be
somewhat greater than the state average.

Household Size and Other Sociodemographic Characteristics
The size of the household plays an important role in determining the
value of food expenditure [21]. In this study, 305 households responded


Table 2. Estimated poverty incidence among households by race and location,
Florida, 1980.
Poverty thresholds

Cell <75% 75-100% 100-125% >125%
Race/ count of poverty of poverty of poverty of poverty
Residence (n) level level level level
Urban:
Black 128 22.6 20.2 16.0 41.1
Hispanic 24 45.8 16.7 20.8 16.7
Rural:
Black 48 56.3 27.1 12.5 4.1
White 100 21.0 4.0 8.0 67.1
Note: For Poverty Income Guidelines, see Appendix Table A-1.


24








to the socioeconomic questionnaire, and all reported the number of
persons living in the household. The average household size for the total
sample was 4.78. Among ethnic groups, rural whites reported an average
family size of 4.50, while the average size for rural blacks was 5.15. There
was no significant variation among urban households. Urban blacks had
an average of 4.82 persons per household, compared to 4.83 for hispanics
(Table 1).
The distribution of family size by race and residence is presented in
Table 3. Forty-nine percent of urban black households registered 2 to 4
persons per household. The corresponding figure for hispanics was 46
percent. Rural black and rural white households reported 42 percent and
52 percent, respectively, for the same family size group (2 to 4 persons).
One-half of the hispanic group responded that number of persons living
in the household was in the range of five to seven. Forty-five percent of
urban blacks, 48 percent of rural blacks and 46 percent of rural whites
reported the same number of persons (5 to 7) in the household. Ten
percent of rural blacks reported the number of persons in the household
as greater than seven. The corresponding figures for the other sample
groups were much lower (Table 3).

Table 3. Family size distribution among households by race and location,
Florida, 1980.
Cell Family size
Race/ count
Residence (n) 2-4 5-7 >7
Urban:
Black 128 49.2 45.3 5.5
Hispanic 24 48.8 50.8 4.2
Rural:
Black 48 41.7 47.9 10.4
White 100 52.0 46.0 2.0


There are a number of problems encountered in generating social
survey data that are absent in other surveys. Many such problems are
associated with the complex social psychology of the human subject. In
this study, one such problem existed with the accuracy of household
income and expenditures. Households have a tendency to under-report
income and over-report expenditures. This general tendency is reflected
in Table 1, where hispanics reported mean total household expenditure
higher than their total income, by an average of $37. The same was true
for rural black households, who reported mean total expenditure greater
than mean total income by an average of $35. These apparent inconsist-
encies could be the result of selectivity in data reporting, or it could be


25








related to the fact that food expenditures include the value of food stamp
coupons, while income does not.
The average educational level for urban hispanic homemakers was
only 6.9 years of schooling, compared to 11.6, 11.2, and 11.8 for urban
blacks, rural blacks, and rural whites, respectively (Table 1). The lowest
employment rate was registered among hispanic homemakers. Forty-one
percent of hispanic homemakers reported that they were working outside
the home. In contrast, the employment rate for urban black homemakers
was about 76 percent. The corresponding percentages for rural blacks
and rural whites were 61 percent and 69 percent, respectively (Table 1).
Twenty-eight percent of the sample population participated in the
Food Stamp Program (FSP). Among urban ethnic groups, 33 percent of
black households and 36 percent of hispanic households responded that
one or more members of the household participated in the FSP. Among
rural black households, 54 percent were registered FSP participants,
compared to only 9 percent among rural white households (Table 1). This
result is not surprising, since 83 percent of the rural black households
were classified as poor, compared to 25 percent among their white
counterparts (Table 2).
Conceptually, the basic nutritional education levels of the homemak-
ers was hypothesized to play a significant role in affecting the nutritional
status of the adolescent. Also, nutritional education is one of the so-
cioeconomic variables postulated to have an influence on the value of
food expenditure. Specifically, this variable is conceived as affecting the
efficiency of food procurement, food storage, and meal preparation. The
specific nature of the nutrient and food expenditure response to this
variable is discussed in the regression analysis sections. However, from a
wide sociodemographic viewpoint, the distribution of nutritional educa-
tion among household groups appears to be a logical starting point for
further analysis of this variable. From the data it would appear that urban
homemakers had a relative advantage in this regard. Specifically, 63
percent of urban black households and 78 percent of urban hispanic
households indicated that they had some knowledge of diet and food and
nutrition requirements as a result of some type of nutrition education
programs. In contrast, the corresponding figures for rural blacks and
rural whites were only 17 percent and 16 percent respectively (Table 1).
Of some interest is the fact that participation in the Expanded Food and
Nutrition Education Program (EFNEP) was minimal in each of the
groups (Table 1). Thus, most of the households might have obtained
nutritional education from other sources, such as news media, pamph-
lets, books, friends, physicians, and extension agents who were not
directly involved in EFNEP. It should be noted, also, that the relatively
lower level of participation in EFNEP among rural households might



26








have been related to the non-availability of the program in the particular
rural area surveyed.6
It was felt that a vegetable garden might make a difference in the
nutritional status of the adolescent. For this reason, one of the questions
asked was whether or not the household had a vegetable garden during
any period of the growing season. Thirty-three percent of rural whites, 17
percent of rural blacks, 10 percent of urban blacks, and 9 percent of urban
hispanics indicated having vegetable gardens ranging from one to four
seasons (Table 1). It was not possible to determine what percentage of
the vegetable products were actually consumed or sold. If the greater part
of the products grown at home were consumed, rural households would
have an advantage in access to some of the nutrients studied.

Food Expenditure and Other Household Expenditures
Table 4 presents mean monthly household income and expenditure
categories disaggregated on the basis of selected socioeconomic and
demographic characteristics. In the aggregate, the largest monthly aver-
age expenditure category per household was for food ($254). The next
largest expenditure category was for housing ($209). On the average, $87
was spent for transportation, $71 for clothing, $59 for medical care, and
$33 for recreational activities. The remainder of monthly expenditure
allocations went into tobacco products, alcoholic beverages, and miscel-
laneous expenses, in that order.
Average monthly household income and expenditure varied substan-
tially across socioeconomic and demographic characteristics of the house-
hold. Among racial groups, whites spent more, in absolute dollars, on
food, housing, and transportation than the other racial groups. However,
in terms of respective income levels, whites spent relatively less on every
type of expenditure category except transportation (Table 4). Relatively
speaking, average expenditures were proportionally higher for hispanics
and rural blacks. In each type of expenditure category, the ratio of
expenditure to income was highest among hispanics and rural blacks
(Appendix Table B-1).
Average expenditures also varied substantially by residential location
(Table 4). The average monthly expenditure of urban households on
housing, clothing, recreation, and alcoholic beverages was higher than
those of rural households, both in absolute and relative terms. In con-
trast, rural households spent more on food and transportation than did
urban households. Since there is no local public transportation, rural


6. The EFNEP was not operational in rural Sumter County at the time of the
survey. Households registering participation in the program did so in some other
location at a prior date.


27









Table 4. Mean monthly household expenditures by type and selected sociodemographic characteristics, Miami and Sumter County,
Florida, 1980.

Cell Average
count monthly Recre- Transpor-
Characteristics (n) income Food Housing Clothing Medical ation station Alcohol Tobacco

Dollars
Aggregate 305 887 254 209 71 59 33 87 16 21
Race:
White 205 1143 295 237 56 62 34 133 14 27
Black 176 772 228 190 85 51 34 60 15 15
Hispanic 24 600 262 224 47 97 22 50 30 19
Location:
Urban 159 838 233 213 78 58 34 52 19 16
S Rural 148 940 276 204 65 61 33 123 13 24
Household size:
2-4 persons 149 853 217 200 61 57 34 84 16 22
5-7 persons 142 912 277 216 76 62 30 88 15 20
> 7 persons 16 989 401 222 116 61 55 93 0 13
Income:
<75% of poverty level 88 369 224 152 67 54 19 51 18 16
75-100% of poverty level 48 569 215 208 79 50 22 69 8 17
100-125% of poverty level 42 750 243 201 70 77 21 54 14 26
>125% of poverty level 125 1411 293 251 71 59 41 123 17 24






Table 4. (cont.)
Cell Average
count monthly Recre- Transpor-
Characteristics (n) income Food Housing Clothing Medical ation station Alcohol Tobacco

Education (homemaker):
<9th grade 33 650 248 189 53 75 23 61 20 19
9-12th grade 199 853 257 202 71 57 35 96 15 21
> 12th grade 44 1306 282 272 72 49 42 106 18 21
Employment (homemaker):
Participants 77 508 251 171 64 60 30 51 19 17
Nonparticipants 200 1003 265 228 69 61 34 103 15 21
Basic nutrition education
Yes 114 866 242 219 67 65 35 64 20 17
S No 165 939 275 209 71 59 33 105 12 21








households generally incur higher transportation costs from use of pri-
vate transportation. In terms of food expenditure, urban households tend
to enjoy a more favorable proximity to markets, which would be ex-
pected to provide a wider variety of food items at relatively lower (dis-
counted) prices. Thus, the choice for rural households is one of additional
expenditure for transportation to secure a comparable bundle of food
items avialable to urban households (at relatively lower prices), or be
satisfied with less and a narrower choice of food items at a relatively
higher price.
Rural black households registered the highest percentage of monthly
income allocation for food (48 percent of income). In contrast, rural
white households spent only 26 percent of their monthly income on food.
Comparable allocations for urban black and hispanic households were 25
percent and 44 percent, respectively. Given the average family size of the
household categories (Table 1), rural black households had monthly per
capital outlay for food equal to $46, compared to $65 for whites. Monthly
per capital food outlay for urban black and hispanic households were $46
and $54, respectively. However, given the significantly higher incidence
of poverty among rural black households (Table 2), it might very well be
that locational factors might have forced these households to be satisfied
with less and narrower choice of food items than their rural white coun-
terparts, as well as their urban counterparts.
Generally, average household expenditures in each expenditure cate-
gory of Table 4 were: (a) higher for larger household sizes, (b) higher for
households where the homemaker had relatively higher level of educa-
tion, (c) higher for households where the homemaker was employed
outside the home, (d) lower for households in which one or more mem-
bers of the household participated in FSP, than for households which
reported nonparticipation, and (e) higher for households whose average
household income was higher than for households with lower average
income. Also, households with homemakers having no basic nutritional
education tended to spend more on food, clothing, and transportation. It
is conceptually plausible that income, family size, educational level,
employment status, and FSP participation are, to some degree, interre-
lated in their aggregate impact on food expenditure and nutritional
status. As such, the net effect of each variable would not be clear from the
descriptive analysis. The results of the two regression equations (expend-
iture model (7), and nutrient model (8)), discussed in the succeeding
sections, are used to clarify the impact and relationship of each variable
on food expenditure and nutrient level.
As additional background to subsequent analysis, Tables 5 and 6 also
summarize the monthly food expenditure of household categories by
broad food groups and socioeconomic categories. The largest food group
share went to meat products, followed by fruits and vegetables, dairy


30








Table 5. Mean monthly food expenditure by income and food group category,
Miami and Sumter County, Florida, 1980.
Household category

Total Urban (Miami) Rural (Sumter)
Black Hispanic Black White
n= 304 n= 128 n=24 n=48 n= 100
Dollars
Monthly food expenditure 254.00 224.00 262.00 237.00 294.00
(including FSP purchases
and away from home pur-
chases) Percent
Monthly income allocation 29.0 25.0 44.0 48.0 26.0
for food (excluding FSP
income)
Monthly food expenditure Percent
to food groups:
Meat products 43.0 48.8 44.3 44.2 35.4
Dairy products 16.1 15.1 12.6 16.8 17.6
Fruit and vegetables 18.0 18.2 22.0 14.9 18.2
Grain products 12.9 12.0 16.1 12.2 13.8
Miscellaneous 10.0 6.9 5.0 11.9 15.0
Total 100.0 100.0 100.0 100.0 100.0


products, grain products, and miscellaneous products, in that order
(Table 5). Urban black households spent almost half of their food alloca-
tion for meat products (49 percent), while they spent 15 percent of their
food expenditure on dairy products. For rural whites, the corresponding
share were 35 percent and 18 percent. Hispanics spent 44 percent of their
food expenditure on meat, while the relative share for fruits and vege-
tables was high, compared to other racial groups.
Table 6 shows the proportion of monthly income allocated to food
expenditure by selected subgroups. A number of interesting characteris-
tics are evident. Black and hispanic households had food expenditure-
income ratios higher than that of all households, while those of whites
were lower. The same was true for rural households compared to urban
households, households of five or more persons, compared to households
of two to four persons, and households with less than 75 percent of the
1980 poverty income threshold, compared to households with 100-125
percent of the threshold. Households with income less than 75 percent of
the poverty threshold allocated 61 percent of their mean monthly income
to food. This percentage was more than twice the average allocation for
all households. Households participating in the FSP allocated 49 percent
of mean monthly income to food compared to 26 percent for non-


31








Table 6. Monthly food expenditure as proportion of monthly income, by selected socioeconomic characteristics, Miami and Sumter
County, Florida, 1980.

Mean monthly
Mean food expenditure Food
Number monthly (including FSP expenditure
of income coupon purchases as
households (excluding and away from proportion
Characteristic (n) FSP income) home purchases) of income

Dollars Percent
All households 300 887 254 28.6
Race:
White 100 1154 295 25.6
Black 176 772 228 29.5
Hispanic 24 600 262 43.7
Location:
Urban 152 838 233 27.8
Rural 148 940 276 29.4
Household size:
2-4 persons 146 853 217 25.4
5-7 persons 139 912 277 30.5
> 7 persons 15 989 401 40.5
Poverty income status:
<75% of poverty level 88 369 224 60.7
75-100% of poverty level 47 569 215 37.8
100-125% of poverty level 39 750 243 32.4
> 125% of poverty level 126 1411 293 20.8





Table 6. (cont.)

Mean monthly
Mean food expenditure Food
Number monthly (including FSP expenditure
of income coupon purchases as
households (excluding and away from proportion
Characteristic (n) FSP income) home purchases) of income

Education of homemaker:
< 9th grade 33 650 248 38.2
9-12th grade 199 853 757 30.1
> 12th grade 44 1306 282 21.6
FSP Participation:
Participant 77 598 251 49.4a
Nonparticipant 200 1003 265 26.4
Nutrition education:
Yes 114 866 242 27.9
No 165 939 275 29.3

"aThis money income-food expenditure ratio is an "unadjusted" ratio, and as such, might overstate food expenditure. This overstatement stems from the fact
that mean monthly income does not include the income supplemental value of food stamp coupons, while mean monthly food expenditure includes
purchases made with coupons. Ideally, the food expenditure of FSP participants should be adjusted to reflect the cash value of food stamp coupons.
Respondents were willing to divulge food expenditure from money income and food stamp income, but would not divulge the monthly dollar value of food
stamp income supplement. The components of money income are described in the discussion on data base.








participants.7 The reverse was noted for households with basic nutrition
knowledge. In this subgroup, those households with nutrition knowledge
had income-food expenditure ratio lower than the aggregate ratios, while
those who did not, had a higher ratio.
The above descriptive analysis is intended to provide a broad contex-
tual framework for analysis of the econometric results of the expenditure
and nutrient models. In reviewing the statistical (regression) findings, it is
imperative that the results be viewed within the broader data set if policy
analysis is to be meaningful.

Food Expenditure Regression Analysis
The regression results of the food expenditure model (7) are presented
in this section. These results statistically explain the general impact of
selected socioeconomic characteristics on the value of household food
expenditure. Specifically, the responsiveness of food expenditure to
changes in the level of household income, household size, and other
discrete household characteristics, such as race, educational status, FSP
participation, to name a few, are discussed. Detailed model specification
for equation 7 was given earlier. A summary of regression parameters for
selected socioeconomic variables for the aggregate sample (total house-
holds) are given in Table 7. Summary of regression estimates for the four
household categories (race-location groups) are given in Table 8. In
Table 9, estimates of marginal propensity to spend (MSP) for food, and
food expenditure-income elasticity by selected household characteristics
are presented.

Income Elasticity and Expenditure Propensity Characteristics
Since the set of data for the present study was composed of four distinct
race-location groups, it was necessary to test whether the subgroup
regressions were significantly different from the aggregate regression.
For this purpose, two hypotheses were tested: (a) a test of homogeneity
of the regressions in which the intercepts and slopes were hypothesized to
be equal for all subgroups, and (b) a test for equality of the slope
coefficients for the subgroups. Using the F-test, the regression analysis
rejected both hypotheses.
A regression analysis was used to estimate a single equation (with
intercept shifters) from the data, since the slope coefficients of the
subgroup regressions were not significantly different from the aggregate.

7. This money income-food expenditure ratio actually overstates the food
expenditure level of FSP participants, since it does not reflect the income value of
food stamp coupons, but reflects food purchases from food stamp coupons. FSP
participants refused to divulge information relating to the cash value of cou-
pons so it was impossible to compute an "adjusted" ratio. Also, see footnote to
Table 6.


34









Table 7. Statistical summary of OLS monthly food expenditure equation", all
households by selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980.

Socioeconomic Total Regression Standard
variable n 165 coefficient errors t-value

Intercept 2.78 0.321 8.69**
Household income (Inl)a 0.329 0.044 7.43**
Household size (InHS)5 0.529 0.077 6.86**
Age of homemaker (A) 0.044 0.051 0.86
Race (R)b:
Urban black -0.393 0.098 4.03**
Rural black -0.254 0.120 2.11*
Rural white -0.208 0.106 1.97*
Educational level (E)C
9th-12th grade -0.037 0.071 0.52
<9th grade -0.179 0.107 1.67
Nutritional education (EDNT) -0.105 0.059 1.78*
FSP participation (FSP) 0.152 0.068 2.23*

R2 = .3929
F = 16.44"*

"'Food expenditure, household income, and household size are expressed in logarithmic
form. See page 18 for model specification.
"hHispanic group omitted.
'College level group omitted.
*P < 0.05 (coefficient significant at 95% level).
"**P < 0.01 (coefficient significant at 99% level).


Table 7 presents coefficient estimates of the aggregate data. The table
indicates that income exerted a positive and significant impact on the
value of monthly food expenditure. Since expenditure and income are
expressed in logarithmic form, the value of the income coefficient is the
income elasticity for food expenditure. Food expenditure-income elas-
ticity is defined as the additional percentage change in food expenditure
resulting from a 1 percent increase in income, when all other variables are
held constant. The income elasticity estimate for the aggregate sample
was 0.329. The interpretation of this number is that for every 1 percent
increase in monthly household income, monthly food expenditure in-
creased by 0.329 percent. This finding is consistent with the income
elasticity of 0.32 reported by Buse and Salathe [9] from their analysis of
the 1960-61 BLS Consumer Expenditure Survey, and by Smallwood and
Blaylock from their analysis of the 1977-78 USDA Nationwide Con-
sumption Survey [29]. However, West and Price [35] reported a lower
aggregate income elasticity of 0.04 in their study of low-income house-
holds (including blacks and Mexican-Americans).


35








Table 8. Statistical summary of OLS monthly food expenditure equation', by
household categories and selected socioeconomic characteristics,
Miami and Sumter County, Florida, 1980.

Household category

Socioeconomic Urban blacks Rural blacks Hispanics Rural whites
variable n= 103 n=40 n=19 n=97

Regression results
Intercept 2.87** 3.40** 2.79 1.65**
(5.19)b (3.66) (1.86) (3.21)
Household income (Inl)" 0.34** 0.31 0.39 0.36**
(3.86) (3.00) (3.80) (2.83)
Household size (InHS)* 0.48** 0.47** 0.79* 0.46**
(3.86) (3.00) (3.80) (2.83)
Age of homemaker (A) -0.06 0.27* -0.05 0.07
(0.66) (2.37) (0.26) (0.80)
Education level of
homemaker (E):
< 9th grade -0.23 0.03 -0.86 0.25
(1.39) (0.08) (1.99) (1.06)
9-12th grade -0.20 -0.28 -0.67 0.16
(1.81) (1.26) (1.36) (1.35)
Nutritional -0.08 -0.07 -0.07 -0.19
education (EDNT) (0.93) (0.47) (0.27) (1.59)
FSP participation 0.23** 0.10 -0.20 0.36**
(FSP) (2.11) (0.79) (0.81) (2.12)
R2 = .3548 .4591 .7150 .4226
F = 7.46** 3.88* 3.94** 9.30**

"Food expenditure, household income and household size are expressed in-logarithmic
form. See page 18 for model specification.
"t-values in parentheses.
*P < 0.05 (coefficient significant at the 95% level).
"**P < 0.01 (coefficient significant at the 99% level).

Table 9 presents the estimated marginal propensity to spend (MPS)
and income elasticity for food, by selected sample groupings. The MPS
and income elasticity for the aggregate sample are calculated from the
aggregate regression and aggregate sample means. The MPS and elastic-
ity for the subgroup are calculated from the subgroup regressions and the
subgroup means. The estimated income elasticity for each sample group-
ing is the coefficient of the income variable for the sample grouping
regression." The MPS is calculated by multiplying the estimated group

8. The procedure involved in generating these results was to first partition the
sample and then apply the empirical model to the various sample partitions. The
income coefficients for the subgroup regressions are the income elasticities, since


36








income elasticity by the corresponding ratio between mean monthly food
expenditure and mean monthly money income for each group. The MPS
is the additional expenditure resulting from an increase in income of
$1.00, when all other variables are held constant. The MPS for food
ranged from a low of 0.064 for all urban households to a high of 0.170 for
hispanic households. The MPS for the aggregate sample was 0.094. This
result means that for all households, for every one dollar increase in
income the value of food expenditure increased by an average of 9.4
cents.

Household Size and Other Socioeconomic Characteristics
Household size was also important in explaining food expenditure
variations in the aggregate sample (Table 7). The size coefficient estimate
(0.529) is positive and significant. Since household size and monthly food
expenditure are specified in logarithmic terms, the size coefficient can be
interpreted as the elasticity of household size for food expenditure. Size
elasticity can be interpreted as the percent increase in food expenditure
as household size increases by 1 percent. Thus, a 1 percent increase in
mean household size could result in a 0.529 percent increase in monthly
food expenditure across a range of sizes for the aggregate sample. The
household size coefficient of 0.529 was computed at the aggregate sample
mean of 4.78 persons (Table 1). The coefficient suggests economies of
size in food expenditure at mean family size, since the condition for scale
economies would require that the household size coefficient (HS) assume
a value between zero and one.' The existence of economies of size in food
expenditure is further supported by the values of the elasticity and MPS

the empirical model specifies expenditure and income in logarithmic form. In the
double logarithmic functional form, the income elasticity is constant for all
income levels. However, since the income elasticity estimates of the subgroups in
Table 9 are derived from subgroup regressions, they vary at subgroup means, but
are constant across the income range for the subgroups.

9. The mathematical test for scale economies is:
InQ = c + p InHS + ax
Q = (ea+aXP)HS


OHS


aHS \HS e
aHS aHS
,a+ax < Ofor p > 0 (economies)
Se HS2 = 0 for 3 = 0 (constant)
> 0 for p < 0 (diseconomies).


37








coefficient in Table 9. The estimated food expenditure elasticity coef-
ficients from subgroup regressions were 0.400 and 0.320, for 2- to 4- and
5- to 7-person households, respectively. The food expenditure elasticity
coefficient at the 2- to 4-person level (0.400) was higher than that at the
sample mean of 4.78 persons (0.329), and at the 5- to 7-person levels
(0.320). This suggests that as household size increased, food expenditures
increased at a decreasing rate. This finding is consistent with results
reported by West and Price [35].
The qualitative (dummy) variables included in the expenditure model
were age of the homemaker, race, education level, nutritional education,


Table 9. Statistical summary of food expenditure marginal propensity and in-
come elasticity, by selected household characteristics, Miami and
Sumter County, Florida, 1980.

Marginal Food Standard
propensity expenditure error of
to spend income income
Grouping (MPS)a elasticity elasticity t-value

Aggregate 0.094 0.329 0.044 7.43*
Race (R):
White 0.092 0.360 0.067 5.37**
Black 0.090 0.308 0.065 4.74**
Hispanic 0.170 0.390 0.206 1.89***
Region (L):
Urban 0.064 0.230 0.076 2.94**
Rural 0.117 0.400 0.054 7.38**
FSP participation
(FSP):
Participants 0.074 0.150 0.055 2.64**
Nonparticipants 0.097 0.390 0.058 6.78**
Educational level
homemaker (E):
<9th grade 0.127 0.460 0.141 3.28**
9-12th grade 0.104 0.280 0.049 5.71**
Family size (HS):
2-4 persons 0.102 0.400 0.076 5.23**
5-7 persons 0.097 0.320 0.065 4.82**

"Product of the estimated subgroup income elasticity and the ratio of subgroup mean
household monthly food expenditure and mean monthly money income. Expenditure-
income ratios were computed from averages in the sample groupings. See footnote 8 of text.
"This is the log-income (Inl) coefficient for the sample grouping food expenditure regres-
sions. Since food expenditure is also specified in logarithmic form, the log-income coef-
ficient is the income elasticity for the grouping. See footnote 8 of text.
*P < 0.01 (coefficient significant at 99% level).
"**P < 0.05 (coefficient significant at 95% level).
***P < 0.10 (.., .. ... at 90% level).


38








and FSP participation. All the dummy variables, except age and educa-
tional level, were significant in explaining variations in the level of food
expenditure (Table 7). As indicated by the coefficients of the race-
location dummy variables in Table 7, substantial food expenditure varia-
tion existed between the groups. The negative parameter estimates for
blacks (both urban and rural) and whites (rural) indicate they spent
relatively less on food than hispanics. This result suggests that cultural
and locational differences can be important factors in determining the
value of food expenditure.
General educational level of the homemaker (E) showed no significant
impact on the level of aggregate food expenditure (Table 7). Although
the general educational level of the homemaker made no difference in
the value of food expenditure, the homemaker's nutritional knowledge
was significant in explaining expenditure variations among household
groups. Specifically, the regression coefficient for the homemaker's stock
of nutritional education (EDNT) was negative. This suggests that house-
holds where the homemaker had some type of basic nutrition knowledge
spent less money on food than similar households whose homemaker did
not possess such a knowledge. This finding suggests that the general
educational level (in terms of years of schooling) might be a poor proxy
for the stocks of knowledge required to achieve efficiencies in food
purchases and consumption. It might very well be that this type of
knowledge is so specialized that it can only be communicated through
channels quite different from the traditional educational channels. The
negative impact of the EDNTvariable on food expenditures suggests that
there is an interaction between the homemaker's knowledge of food
nutrient composition and the level of economic literacy, with respect to
food purchasing and preparation. Such an interaction tends to improve
food purchasing efficiency.
Food Stamp program participation had a significantly positive impact
on household food expenditures (Table 7). The FSP coefficient had a
value of 0.152 and was significant at the 95 percent level. However, for
reasons given earlier, the binary form of specification of this variable
would tend to underestimate the extent of the dynamic impacts of coupon
value on food expenditure patterns. In spite of this shortcoming, it is
evident that this variable is a major determinant of food expenditure
patterns as reflected in the discussion on the characteristics of subgroup
MPS and income elasticities.
EMPIRICAL RESULTS: NUTRITIONAL COMPONENT
In this section the empirical results of the nutritional component of the
study are discussed. The discussion is presented in a three-part sequence.
The first part contains a descriptive analysis of the incidence of nutritional
deficiency in terms of comparisons between races, sexes, and regions. In


39








the second part there is a discussion of nutrient indicator variations
between sample groups. OLS regression results of the nutrient model and
discussions of how each nutrient indicator is affected by selected so-
cioeconomic variables are presented in the third part.

Descriptive Analysis

Incidence of Nutrient Deficiency
The focal point of this analysis is the prevalence of nutrient deficiency
among adolescents of different racial background, residential location,
and sex. Tables 10 and 11 show the incidence of selected nutrient de-
ficiency by race, sex, and region. Specifically, the tables show the percent
of sample population in each group with nutrient levels below the estab-
lished norm as specified in Table 10. An individual is considered to be
nutrient deficient if the level of that nutrient indicator is below a certain
threshold. The U.S. Department of Health, Education, and Welfare
study [32], reviewed earlier, used similar threshold levels for classifying
biochemical nutrient parameters.
Due to the adolescent's rapid rate of growth, the requirement for
nutrients increases as the synthesis of lean body tissues and blood volume
increases [14]. If this requirement is not met with an adequate consump-
tion of nutrients, adolescents would develop nutritional deficiency, which
is likely to affect their health. The major functions, the health problems
associated with deficiency, and the major food sources of the nutrients
studies are presented in Appendix Table C-1.
One particular nutrient that has been extensively studied is iron.
Several studies have reported that there is a high incidence of iron
deficiency in the 12 to 16 age group in the United States. Three U.S.
Department of Health, Education, and Welfare studies [20, 32, 33],
suggested that a high incidence of iron deficiency exists in most of the
nation's adolescents. In this study, the level of iron status is determined
by two nutrient indicators-hemoglobin and serum iron. Both indicators
are generally used as criteria to identify an individual's iron status [25].
As indicated in Table 10, the minimum hemoglobin threshold is 13 g/100
mL for males and 11.5 g/100 mL for females. If serum iron is used to
measure the level of iron, the minimum threshold is 60 pg/100 mL for
males and 40 pxg/100 mL for females [25]. These two threshold guidelines
are used in the discussion of the findings of the present study.
Tables 10 and 11 give the percent of sample populations in each
category with nutrient levels below the norm. Considering the level of
hemoglobin concentration as a measure of iron status, white adolescents
generally had a higher incidence of iron deficiency than blacks or his-
panics. Forty percent of white adolescents had low levels of hemoglobin,
in contrast to 16 percent for blacks and 12 percent for hispanics. Forty-


40






Table 10. Incidence of adolescent nutrient deficiency by race and sex. Miami and Sumter County, Florida, 1980.

Adolescent category

Blacks Whites Hispanics

Criteria for Totala Male Female Totala Male Female Totala Male Female
low value n= 221 n= 112 n= 109 n= 128 n=73 n=55 n=31 n= 15 n=16

Percent
Serum folacin (SF)
(ng/mL) <3.0 14.0 10.8 16.7 27.9 20.0 36.6 14.0 0.0 21.4
Serum iron (IR) <60 (Male)
(ig/100 mL) <40 (Female) 7.0 22.0 3.6 12.0 18.4 0.0 0.0 0.0 0.0
R.B.C. folacin (RF)
(ng/mL) <140 45.6 40.5 50.0 48.8 31.1 70.3 36.5 25.0 42.9
Hemoglobin (HGB) <13.0 (Male)
(g/100 mL) <11.5 (Female) 16.0 14.5 16.4 40.4 48.1 32.0 12.5 11.0 13.3
Protein (PRTN)
(g/100 mL) <6.0 16.0 14.5 16.4 40.4 48.1 32.0 12.5 11.0 13.3
Vitamin C (VITC)
(mg/100 mL) <0.2 4.1 1.6 5.8 5.0 0.0 2.7 0.0 0.0 0.0
Vitamin B12 (VITB12)
(pg/100 mL) <200 1.8 3.4 0.0 1.3 0.0 2.7 0.0 0.0 0.0
Hair zinc (HAIRZN)
(ig/g) <100 23.0 27.0 20.0 21.4 25.0 17.0 8.3 11.0 6.7
"Totals are averages of male and female values.








eight percent of white males were deficient in iron, in contrast to 32
percent of white females. The corresponding percentages for black males
and black females were 14.5 and 16.4, respectively (Table 10).
Rural adolescents had a higher incidence of low iron nutrient value
than their urban counterparts. Nineteen percent of blacks and 5 percent
of whites living in rural Sumter County had low levels of hemoglobin. In
contrast, low level of hemoglobin among urban black adolescents was 14
percent. There were no subjects in the urban white or hispanic category
registering low values of hemoglobin (Table 11).
The other criterion generally used to interpret the prevalence of iron
deficiency is the level of serum iron status in the adolescent. Considering
this indicator as a criterion, only 7 percent of blacks and 12 percent of
whites were deficient in iron (Table 10). Male adolescents showed higher
deficiency rate in serum iron than their female counterparts. Twenty-two
percent of black males and 18 percent of white males had low values of
serum iron. The corresponding numbers for black females and white
females were 3.6 and 0.0 percent, respectively. Hispanics also showed no
sign of iron deficiency by this criterion (Table 10). The serum iron
nutrient indicator also suggests that rural adolescents suffered a relatively

Table 11. Incidence of adolescent nutrient deficiency by race and region, Miami
and Sumter County, Florida, 1980.

Adolescent category

Blacks Whites Hispanics

Urban Rural Urban Rural Urbana
Percent
Serum folacin 11.7 21.1 20.0 28.4 14.0
(ng/mL)
Serum iron 5.0 10.0 0.0 13.0 0.0
(ng/mL)
R.B.C. folacin 40.0 63.2 40.0 49.4 36.4
(ng/mL)
Hemoglobin 14.0 19.0 0.0 5.0 0.0
(g/100 mL)
Protein 10.3 40.0 20.0 42.6 12.5
(g/100 mL)
Vitamin C 1.7 14.8 0.0 5.3 0.0
(mg/100mL)
Vitamin B12 2.8 0.0 0.0 1.3 0.0
(pg/100 mL)
Hair zinc 20.0 30.4 20.0 21.4 8.3
(pvg/g)
"aNo hispanic sample population in rural Sumter County.


42








high incidence of iron deficiency. Ten percent of rural blacks and 13
percent of rural whites had serum iron levels below the norm. The
corresponding percentages among urban adolescents were 5.0 and 0.0 for
blacks and whites, respectively (Table 11).
In summary, the serum iron criterion suggests that the incidence of iron
deficiency was highest among blacks, males, and rural adolescents. As to
the choice of criterion, there seems to be no consensus as to whether
hemoglobin or serum iron is a superior determinant of iron nutriture.
Both criteria are widely used by researchers. As such, it is not unusual to
find conflicting results and policy recommendations.
Another important nutrient examined is folacin. Although the inci-
dence of folacin deficiency has not been extensively studied, it is recog-
nized that this nutrient is a dietary essential for humans [14]. In this study,
the concentration of folacin in red blood cells and serum folacin concen-
tration are used to identify the incidence of folacin deficiency. It has been
suggested, however, that the concentration of folacin in red blood cells is
a better indicator of folacin than serum folacin [10]. Using red blood cell
folacin as an indicator for folacin deficiency, the findings of this study
indicate that a high percent of adolescents had folacin levels below the
minimum threshold in all categories (Table 10). Forty-six percent of
blacks, 49 percent of whites, and 36 percent of hispanics exhibited low
levels of red blood cell folacin. Female adolescents showed the highest
incidence of folacin deficiency. Fifty percent of black females, 70 percent
of white females, and 43 percent of hispanic females were below the
accepted nutrient levels. The prevalence of red blood cell folacin de-
ficiency was highest among rural adolescents. Sixty-three percent of rural
blacks and 49 percent of rural whites were deficient in folacin nutriture.
The corresponding figures for urban blacks, urban whites, and hispanics
were 40 percent, 40 percent, and 36 percent, respectively (Table 11).
In terms of serum folacin, the findings show that 14 percent of blacks,
28 percent of whites, and 14 percent of hispanics were deficient in folacin.
There was a consistent pattern of relatively high incidence of serum
folacin deficiency among females and rural adolescents (Tables 10 and
11).
Another nutrient that is widely known and extensively studied is
protein. Protein is vital in physical growth and regulation of body proces-
ses and is used as a source of energy [14]. Protein deficiency causes a
disease that severely affects both the physical and mental health of a
person. The findings of this study showed that all categories were affected
by low levels of protein (Tables 10 and 11). The incidence of protein level
below accepted norms was highest among white adolescents. Forty per-
cent of whites had levels of protein below the norms. In contrast, the
prevalence of low protein levels among blacks and hispanics was 16
percent and 13 percent, respectively (Table 10).


43








Rural white male adolescents exhibited the highest incidence of low
protein levels (48 percent versus 15 and 11 percent for black and hispanic
counterparts). Also, a high percentage of rural adolescents had a high
prevalence of low protein values. Forty percent of rural blacks and 43
percent of rural whites showed levels of protein below the norms, com-
pared to only 10 and 13 percent for urban blacks and urban hispanics,
respectively. The prevalence of low protein levels found among these
sample groups is comparable to levels found in the National Center for
Health Statistics study [20], in which it is reported that more than 20
percent of adolescents suffered from some degree of protein deficiency.
The present study also attempted to identify the zinc status of low-
income adolescents. The role of zinc in human nutrition is increasingly
being recognized. Researchers have found cases where retarded physical
development was attributed to a low dietary intake of zinc [14]. Both hair
tissue and serum analysis are used to provide a data base from which zinc
status can be evaluated [30]. However, within the socioeconomic context
of this study, the decision was made to focus on hair zinc analysis. If 100
Rtg/g is taken as the minimum norm threshold for hair zinc status, the
results suggest prevalence of low hair zinc values in all adolescent groups.
As shown in Table 10, hispanics had less prevalence of low values of hair
zinc than blacks and whites. Eight percent of hispanics registered low
values of hair zinc, compared to 23 percent and 21 percent for blacks and
whites, respectively. Male adolescents had the highest incidence of low
zinc status. Twenty-seven percent of black males, 25 percent of white
males, and 11 percent of hispanic males were deficient in zinc. Compara-
ble figures for black, white, and hispanic females were 20 percent, 17
percent, and 7 percent, respectively. Rural blacks registered the highest
incidence of low zinc status. Thirty percent of rural black adolescents
registered deficiency in zinc status in contrast to only 21 percent among
their white counterparts (Table 11).
Although this study analyzed only hair zinc status, comparable serum
zinc analysis was also undertaken by Wagner et al. [34]. As a point of
interest, Wagner and associates reported serum zinc deficiency among 13
percent of rural blacks, 5 percent of rural whites, 3 percent of urban
blacks, and 0 percent of urban hispanics. By combining serum and hair
zinc parameters to determine the zinc status of adolescents, they reported
that 36 percent of rural blacks, 24 percent of rural whites, and 21 percent
of rural blacks, and 10 percent of urban hispanics had low levels of serum
and/or hair zinc.
The other nutrients examined in this study were vitamin C and vitamin
B12. Numerous studies indicate that a large segment of the U.S. popula-
tion had inadequate intakes of vitamin C. Studies focusing on the adoles-
cent population consistently report vitamin C levels below acceptable
standards among low-income groups. The National Center for Health

44








Statistics (NCHS) study [20] reported more than 50 percent of adoles-
cents in the nation were affected by a high incidence of vitamin C
deficiency. In contrast, the findings of this study indicate a high incidence
of vitamin C and vitamin B12 deficiency did not exist among the aggregate
adolescent sample population (Table 10). However, among specific sub-
groups, 15 percent of rural black adolescents and 5 percent of rural white
adolescents were deficient in vitamin C. No vitamin B12 deficiency was
registered among rural black adolescents, and only 1 percent of rural
whites had such deficiency. About 3 percent of urban blacks had vitamin
B12 deficiency, and about 2 percent had vitamin C deficiency (Table 11).
In summary, the findings of this study suggest that blacks, females, and
rural adolescents were more malnourished than the other groups. In most
of the nutrient indicators examined, these groups had the highest inci-
dence of nutrient deficiency. This finding is similar to those of three
important national studies [20, 32, 33] which documented high preva-
lence of low nutrient levels for these same groups.

Group Mean Nutrient Differences
This section presents the mean value of each nutrient indicator by
household category. Tables 12 to 15 show mean nutrient differences by
race, sex, and location. Although a simple comparison of means using the
t-test may not produce a reliable statistical inference, it may still be of
interest to note the mean nutrient differences between the relevant
subgroups.
Table 12 shows the mean values of each nutrient indicator for urban
blacks, hispanics, rural whites, and rural blacks. Among these groups,
rural blacks and rural whites generally showed low mean values. In Table
13, the mean values of nutrient by race and sex are presented. In five of
the eight nutrient indicators, female adolescents from all three races
showed low mean levels when compared to their male counterparts.
However, the mean levels of protein, vitamin B12, and hair zinc for
female adolescents were equal or slightly higher than that of male adoles-
cents in each race category.
The mean nutriert differences of each sex in each of the locations are
given in Table 14. Among male adolescents, regional difference existed
in mean levels of serum folacin, serum iron, and protein. In all three
nutrients, urban adolescents showed high mean levels. Among female
adolescents, the urban group had higher mean levels in protein and
vitamin C than that of rural female adolescents.
The differences in the mean levels of nutrients for black adolescents is
presented in Table 15. As shown in the table, urban male adolescents
showed high mean values of serum folacin, serum iron, RBC folacin, and
protein. Among black female adolescents, there were regional differ-
ences in protein and vitamin C. In both cases, urban females had higher
mean levels than that of rural females.
45








Table 12. Mean nutrient differences among adolescents, by race and region,
Miami and Sumter County, Florida, 1980.
Urban Rural
Total
Black Hispanic White Black
Nutrient n= 353 n= 163 n = 31 n=101 n=58

Serum folacin 7.7 8.4 9.1 7.4 5.2
(ng/mL) (0.35)a (0.58) (.69) (0.76) (0.47)
Serum iron 100.60 108.5 106.3 95.7 86.7
(tg/100 mL) (2.19) (3.52) (10.2) (4.67) (5.17)
R.B.C. folacin 172.4 169.1 197.5 185.9 134.3
(ng/mL) (5.26) (6.86) (24.0) (13.4) (9.90)
Hemoglobin 13.6 13.3 14.3 14.2 13.2
(g/100 mL) (0.07) (0.10) (0.20) (0.12) (0.19)
Vitamin C 1.01 1.10 1.00 0.90 0.70
(mg/100 mL) (0.02) (0.04) (0.09) (0.05) (0.09)
Vitamin B12 412.8 450.5 390.0 382.5 442.1
(mg/100 mL) (6.80) (14.5) (13.6) (11.8) (8.8)
Zinc 142.6 141.7 151.0 154.0 137.9
(pg/g) (3.37) (6.4) (12.6) (6.7) (10.9)
"Numbers in parentheses are estimated standard errors of mean.

Table 13. Mean nutrient differences among adolescents, by race and sex, Miami
and Sumter County, Florida, 1980.
White Black Hispanic

Male Female Male Female Male Female
Nutrient n=56 n=47 n=79 n=92 n=9 n=15

Serum folacin 8.8 6.1 8.4 6.9 10.5 8.3
(ng/mL) (1.2)a (0.86) (0.81) (0.51) (3.6) (1.8)
Serum iron 94.3 99.0 109.2 96.9 130.0 87.3
(pig/100 mL) (5.8) (7.0) (4.7) (3.8) (10.6) (13.8)
R.B.C. folacin 217.8 146.1 166.3 155.8 253.1 165.8
(ng/mL) (17.2) (17.0) (8.3) (8.2) (53.4) (19.4)
Hemoglobin 14.6 13.6 13.7 12.8 14.8 14.0
(g/100mL) (0.14) (0.13) (0.14) (0.10) (0.32) (0.24)
Protein 5.8 6.2 7.4 7.3 7.0 7.1
(g/100 mL) (0.26) (0.18) (0.21) (0.18) (0.54) (0.35)
Vitamin C 1.00 0.80 1.10 1.00 1.10 1.00
(mg/100 mL) (0.06) (0.08) (0.06) (0.04) (0.18) (0.10)
Vitamin B12 359.2 404.7 447.4 447.4 353.6 409.6
(pg/100 mL) (12.4) (19.7) (14.6) (18.3) (15.7) (15.2)
Zinc 143.0 163.5 134.7 145.7 142.6 156.1
(xg/g) (8.0) (10.4) (9.3) (0.4) (22.0) (15.6)
"Numbers in parentheses are estimated standard errors of mean.


46








Table 14. Mean nutrient differences among adolescents, by sex and region,
Miami and Sumter County, Florida, 1980.

Male (n 200) Female (n = 180)
Nutrient Urban Rural Urban Rural
Serum folacin 9.9 7.5 7.1 5.6
(ng/mL) (0.82) (0.69) (0.55) (0.60)
Serum iron 118.8 95.3 95.9 87.9
(Lg/100 mL) (3.66) (4.25) (4.25) (4.77)
R.B.C. folacin 185.5 192.0 153.4 155.9
(ng/mL) (8.82) (13.16) (7.39) (13.09)
Hemoglobin 13.9 14.3 13.10 13.2
(g/100 mL) (0.13) (0.12) (0.11) (0.12)
Protein 7.5 6.0 7.5 6.1
(g/100 mL) (0.21) (0.21) (0.16) (0.15)
Vitamin C 1.1 1.0 1.1 0.80
(mg/100 mL) (0.05) (0.05) (0.04) (0.06)
Vitamin BI2 418,4 378.5 430.7 434.9
(pg/100 mL) (14.43) (10.12) (14.40) (15.46)
Zinc 136.0 134.0 145.2 158.0
(p/g/g) (7.67) (5.78) (5.01) (8.49)
"Numbers in parentheses are estimated standard errors of mean.

Table 15. Mean nutrient differences among black adolescents, by sex and region,
Miami and Sumter County, Florida, 1980.
Male (n= 200) Female (n= 180)
Nutrient Urban Rural Urban Rural
Serum folacin 9.5 5.7 7.4 4.8
(ng/mL) (0.85)a (0.62) (0.63) (0.66)
Serum iron 117.9 89.6 99.4 86.2
(ptg/100 mL) (4.19) (6.48) (4.70) (7.27)
R.B.C. folacin 177.4 124.2 158.8 135.5
(ng/mL) (8.21) (8.76) (8.69) (14.35)
Hemoglobin 13.7 13.7 12.9 12.8
(g/100 mL) (0.13) (0.22) (0.10) (0.23)
Protein 7.7 6.7 7.1 6.2
(g/100 mL) (0.21) (0.18) (0.19) (0.17)
Vitamin C 1.1 0.9 1.1 0.7
(Rg/100 mL) (0.05) (0.15) (0.04) (0.10)
Vitamin B12 434.1 422.4 441.3 453.7
(pg/100 mL) (17.53) (17.46) (19.30) (28.88)
Zinc 135.7 117.0 142.5 152.4
(xg/g) (9.13) (8.46) (5.30) (17.44)
"aNumbers in parentheses are estimated standard errors of mean.


47








Nutrient Regression Analysis
This section presents the results of the nutrient regression model
(equation 8) which statistically explains the impact of the households
socioeconomic characteristics on the nutritional status of adolescents.
Included in this section is a regression analysis of interaction terms
between all combinations of race, sex, and location. The purpose of this
regression equation is to determine whether the group classification is
appropriate or not. In other words, the finding will suggest if there was
significant difference between the subgroups in each nutrient indicators.
Table 16 presents the regression results of the interaction terms. As
indicated by the F values for each nutrient indicator, the finding showed
that there were significant differences between subgroups which suggests
that nutrient classification by race, sex, and location can be appropriate.
Except for hair zinc, the F values of all nutrients were significant. Insig-
nificant F values means that there were no nutrient differences among
subgroups.
For the aggregate sample, the regression results of the nutrient model
(equation 8) for each nutrient indicator are presented in Tables 17 to 25.
The detailed model specification for equation 8 is given earlier. Also,
results of subgroup samples are presented in Appendix Tables D-I to
D-8. "
An explicit hypothesis was that the nutritional status of the adolescent
would be affected by the household's socioeconomic characteristics.
Household income was one of the economic factors hypothesized to
positively influence nutritional status. The findings of this study suggested
that in the aggregate, no significant relationship existed between house-
hold income and the level of nutrient indicators (Table 17). This finding is
different from that of Adrian and Daniel [1] which utilized 24-hour
dietary recall method. They reported that income had a positive and
significant impact on the consumption of nutrients, although nutrient
consumption was found to be negatively responsive to incremental in-
come changes at higher income levels.
The size of the household was another factor hypothesized to affect
nutritional status. No consistent relationship was registered between
family size and nutrient indicator levels. The relationship between family
size and nutrient indicators, except serum iron, was not significant, as
indicated by the regression coefficients and the respective t-values. The
level of serum iron declined with household size at the 90 percent sig-
nificance level (Table 17).


10. These tables are presented only to give a broad information across the
sample group; the statistical properties of the data dictate caution in drawing firm
conclusions.

48





"Table 16. Statistical summary of OLS nutrient equation, by nutrients and dummy interaction variables, Miami and Sumter County,
Florida, 1980.

Interacting Serum Serum RBC Hemo- Vitamin Hair
variables" folacin iron folacin globin Protein Vitamin C B12 zinc

Intercept 7.91** 117.3** 241.0" 14.7** 5.9** 1.11** 356.6** 158.9**
(3.76)" (22.9) (53.0) (0.65) (0.9) (0.26) (71.7) (40.1)
Interaction:
White and female -2.28 49.6** -48.8 0.23 0.69 0.07 -83.0 -37.5
(3.70) (23.2) (52.2) (0.59) (0.90) (0.27) (67.5) (38.2)
Black and female 0.19 35.9 53.6 0.07 -0.20 -0.01 -84.0 -19.9
(2.98) (19.1) (41.9) (0.48) (0.66) (0.19) (56.0) (31.2)
Female and urban -1.84 -5.63 -58.9* 0.37 0.50 0.22 -74.9* -38.7*
(2.18) (12.9) (30.7) (0.35) (0.58) (0.19) (40.4) (22.4)
Race:
"White 0.70 -23.8 -24.0 -0.09 -0.20 -0.12 2.75 -15.0
(3.72) (22.7) (52.4) (0.64) (0.85) (0.25) (71.3) (39.7)
Black -2.64 -26.2 124.7** -0.86 0.65 -0.17 61.2 -43.4
(3.83) (23.3) (54.0) (0.66) (0.89) (0.27) (72.1) (40.8)
Sex (female) -0.26 -44.5** 19.5 1.22** -0.28 -0.27 132.6** 59.4
(3.51) (22.0) (49.5) (0.56) (0.84) (0.26) (64.3) (36.4)
Location (urban) 2.56 12.7 12.1 0.11 1.13 -0.05 -3.01 -16.4
(3.03) (18.4) (42.7) (0.54) (0.72) (0.21) (58.9) (32.9)

R2 0.0588 0.0887 0.1157 0.2874 0.1989 0.0953 0.1224 0.0347
F 2.02** 3.05** 4.15"* 14.1** 6.52** 3.20** 3.49** 1.30

"The following dummy variables were omitted to avoid problem of multicollinearity: Hispanic, male, and rural adolescents.
"Standard error in parentheses.
*P < 0.10 (significant at 90% level).
**P < 0.05 (significant at 95% level).












Table 17. Statistical summary of OLS nutrient equation by nutrients and selected socioeconomic characteristics, Miami and Sumter
County, Florida, 1980.

Serum Serum RBC Hemo- Vitamin Hair
Independent folacin iron folacin globin Protein Vitamin C B12 zinc
variables n=220 n=215 n=216 n=240 n= 177 n=211 n=177 n=249

Intercept 6.73** 100.81** 206.80** 15.34** 6.5** 1.08** 360.93** 154.07**
(2.46)a (6.21) (5.21) (39.50) (10.02) (6.03) (7.10) (5.70)
Income 0.0008 0.0003 -0.011 0.0001 0.0024 0.0001 -0.002 -0.004
(0.98) (0.05) (0.93) (0.87) (1.27) (0.97) (0.13) (0.57)
Household size 0.32 -2.94* 1.31 -0.03 0.79 -0.02 0.17 -1.47
S(1.09) (1.78) (0.31) (0.76) (1.10) (1.17) (0.02) (0.50)
Age (homemaker) -1.98** -1.21 6.04 0.08 -3.01 -0.10 -17.84 4.04
-41 (2.10) (0.22) (0.44) (0.59) (1.25) (1.51) (1.02) (0.44)
Sex (adolescent) -2.18** -12.0** -38.11** -1.02** -0.009 -0.13** 20.96 14.75*
female (2.42) (2.28) (2.89) (7.79) (0.004) (2.04) (1.23) (1.67)
Race:
RI White -0.53 3.21 -14.61 -0.53* -6.24 -0.06 11.86 1.19
(0.25) (0.24) (0.46) (1.71) (1.24) (0.45) (0.31) (0.05)
R2 Black -0.31 5.61 -53.48* -1.21** 5.21 0.06 94.28** -5.95
(0.15) (0.44) (1.83) (4.16) (1.15) (0.49) (2.50) (0.30)





Table 17. (cont.)

Serum Serum RBC Hemo- Vitamin Hair
Independent folacin iron folacin globin Protein Vitamin C B12 zinc
variables n= 220 n= 215 n=216 n= 240 n = 177 n=211 n= 177 n= 249

Educational level
E, 9-12th grade 0.23 13.94** 9.29 -0.19 -4.61 0.03 10.27 3.01
(0.18) (1.84) (0.50) (1.02) (1.29) (0.30) (0.39) (0.23)
E2- < 9th grade 3.32* 7.17 2.64 -0.26 -0.63 0.16 71.37* 17.06
(1.67) (0.61) (0.09) (0.91) (0.12) (1.22) (1.77) (0.86)
Nutritional educa- 1.92* 10.29* 32.11** 0.090 4.96** 0.07 -32.58* -19.22*
tion (homemaker (1.93) (1.78) (2.19) (0.67) (2.04) (0.95) (1.75) (1.96)
participated)
Food stamp pro- -0.79 3.92 -4.53 0.02 -4.31 -0.02 -25.78* -5.39
gram (partici- (0.66) (0.57) (0.27) (0.11) (1.47) (0.24) (1.74) (0.46)
pated)
Vegetable garden 0.32 -10.51 9.60 -0.08 -0.38 0.007 22.10 5.81
(planted) (0.28) (1.59) (0.56) (0.50) (0.11) (0.09) (1.04) (0.53)
F 2.31** 1.85* 2.35** 10.07** 3.33* 1.18 2.23** 1.05
R2 0.1087 0.091 0.1127 0.3270 0.1818 0.614 0.1294 0.0465

"aNumbers in parentheses are values of Student's t test.
*t-values significant at P < 0.10
"**t-values significant at P < 0.05








Age of the homemaker was also hypothesized to have a negative
impact on the nutritional status of the adolescent. The relationship
between age and the level of nutrient indicators turned out as hypothe-
sized for only one nutrient (serum folacin). In general, there was no
consistently significant relationship between the age of the homemaker
and the nutritional status of adolescents (Table 17). This finding is quite
different from that of Madden and Yoder [18], where household Nutrient
Adequacy Ratio of 10 nutrients declined with age of homemaker. A
recent study by Blanciforti, Green, and Lane [5] reported evidence that
older people tended to have a greater appreciation for more nutritious
foods.
Sex of the adolescent consistently showed significant nutrient varia-
tions. For five of the nutrients, the finding indicated female adolescents
had lower levels of nutrients than did their male counterparts. Specifi-
cally, the coefficients for serum folacin, serum iron, red blood cell fola-
cin, hemoglobin, and vitamin C were significant and negative. This
suggests that female status was associated with low levels of nutrients
(Table 17).
It was also hypothesized that the racial characteristics of the household
would have differential impact on the nutritional status of the adolescent.
Adrian and Daniel [1] found that black households consumed less cal-
cium, thiamine, vitamin C, and iron than did either white or hispanic
households. The finding of this study was that in the aggregate, there was
no consistently significant relationship between the race of the home-
maker and the nutritional status of adolescents (Table 17). Regression
results suggest that black adolescents had low nutrient levels for only one
indicator (hemoglobin), which is one indicator of iron deficiency.
Empirical results of this study showed the general educational level of
the homemaker not to be consistently significantly related to the nutri-
tional status of the adolescent (Table 17). One possible reason for the
insignificance of education could be the fact that the sample contained a
small percentage of households with a college level education. This is
related to the fact that this study was oriented towards relatively poor
households whose educational level beyond high school was rare. It
should be noted, however, that this finding is consistent with Madden and
Yoder study [18], where no significant relationship was found between
general educational attainment and dietary levels of the family. Adrian
and Daniel [1], noted an inverse relationship between educational attain-
ment of the housewife and carbohydrate, fat, iron, and thiamine intakes.
Unlike general educational attainment, it was postulated that special-
ized nutrition educational attainment of the homemaker would have a
beneficial effect on the dietary level of the adolescent. Findings of Davis
and Neenan [11 suggested that the Expanded Food and Nutrition Educa-
tion Program (EFNEP) impacted positively on the nutritional status of


52








low-income households. In this study, nutritional education of the home-
maker was found to consistently influence the nutritional status of the
adolescent. As indicated in Table 17, nutrient indicators such as serum
folacin, serum iron, red blood cell folacin, and protein were highly
responsive to nutritional education. The impact was positive and statisti-
cally significant. An inverse and significant relationship was registered
between nutritional educational and vitamin B12 and zinc. No similar
relationship existed for vitamin C and hemoglobin.
Participation and non-participation by households in the FSP had no
significant impact on adolescent nutritional status (Table 17). However,
this finding cannot be conclusive, since data were not available to relate
how the FSP participants fared nutritionally before they effectively par-
ticipated in the program. However, the findings suggest that adolescents
from households which participated in the FSP were nutritionally equiva-
lent to adolescents from households which did not participate in the FSP
(Appendix Tables D-1 to D-8). This could mean, however, that FSP
participation enabled adolescents to achieve comparable nutritional sta-
tus to those from nonparticipating households.
Adolescents from households possessing vegetable gardens were ex-
pected to have a higher nutritional level than those without gardens.
However, regression results indicated no significant relationships be-
tween the household having a vegetable garden and the nutritional status
of the adolescent (Table 17).

Serum Folacin
A summary of the response for the aggregate sample population
appears in Table 18. For the aggregate sample, the regression model
explained fairly well nutrient variation associated with socioeconomic
characteristics as indicated by the F-statistic, which is significant at the 95
percent level. However, when variables are considered separately, serum
folcain showed no significant responsiveness to changes in income and
family size (low t-values). Likewise, FSP participation by households had
no significant impact on the level of adolescent serum folacin. House-
holds with vegetable garden had no effect on the nutritional status of the
adolescent. As far as serum folacin is concerned, the dummy racial
coefficients indicated that no significant difference existed between races
(Table 18).
Age of the homemaker had a negative and significant impact on the
level of serum folacin. The age coefficient suggests that the level of
adolescent's serum folacin was reduced by 1.98 ng/mL when the age of
the homemaker was greater than 40 years. This number is about 26
percent of the sample mean nutrient value (Table 12), which suggests that
the homemaker's age was a major determinant of adolescent consump-
tion of folacin based food groups (Table 18).


53








Table 18. Statistical summary of OLS nutrient equation": serum folacin (SF), by
households and selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980.

Socioeconomic Regression Standard
variable coefficient error t-value

Intercept 6.730 2.74 2.46**
Household income (1)" 0.0008 0.0008 0.98
Household size (HS)a 0.318 0.29 1.09
Age of homemaker (> 41 years)(A) -1.980 0.94 2.10**
Sex of adolescent (female)(S) -2.180 0.90 2.42**
Race (R):
White -0.530 2.15 0.25
Black -0.310 2.30 0.15
Educational level homemaker (E)
> 9th grade 0.231 1.30 0.18
< 9th grade 3.20 1.99 1.67*
Nutritional education (EDNT) 1.920 0.99 1.93*
FSP-participation (FSP) -0.790 1.19 0.66
Vegetable garden (V) 0.320 1.15 0.28

R = 0.1087
F =2.31**

"Nutrient level, household income, and household size are expressed in linear form. See
page 19 for model specification.
"*P < 0.10 (coefficient significant at 90% level).
**P < 0.05 (coefficient significant at 95% level).

As indicated in Table 17, the homemaker's nutrition knowledge was a
major factor associated with increasing levels of folacin, iron, and pro-
tein. In the case of serum folacin, the nutrient level increased significantly
by 1.92 ng/mL (25 percent of sample mean) for those adolescents whose
homemakers had some kind of nutrition education (Tables 12 and 18).
The general educational level of the homemaker also showed positive
and significant impact on serum folacin when the homemaker's educa-
tional level was lower than ninth grade (Table 18).

Serum Iron
A summary of results obtained from the analysis of the aggregate
sample is presented in Table 19. Household income and FSP participa-
tion had no significant effect on the level of serum iron. Household size
had a significantly negative effect on the level of adolescent serum iron.
As indicated by the regression coefficient, the level of serum iron was
reduced by 2.94 p.g/mL as household size increased by one person (Table
19). This result implies that a sample mean value of 100.60 xg/100mL
(Table 12) of iron status was probably rationed among adolescents of the
household. This nutrient rationing would have had a more severe impact

54








Table 19. Statistical summary of OLS nutrient equation: serum iron (IR), by
households and selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980.

Socioeconomic Total Regression Standard
variable n=215 coefficient error t-value

Intercept 100.81 16.24 6.21**
Household income (I)a 0.0003 0.004 0.05
Household size (HS)a -2.94 1.67 1.78*
Age of homemaker (< 40 years)(A) -1.21 5.53 0.22
Sex of the adolescent (female)(S) -12.00 5.25 2.28**
Race (R):
White 3.21 13.29 0.24
Black 5.61 12.62 0.44
Educational level homemaker (E):
a 9th grade 13.94 7.59 1.84*
s 9th grade 7.17 11.75 0.61
Nutritional education (EDNT) 10.29 5.78 1.78*
FSP-participation (FSP) 3.29 6.86 0.57
Vegetable garden (V) -10.51 6.60 1.59

R2 = 0.0910
F = 1.85*

"Nutrient level, household income, and household size are expressed in linear form. See
page 19 for model specification.
*P < 0.10 (coefficient significant at 90% level).
**P < 0.05 (coefficient significant at 95% level).

on larger size households with the higher incidence of poverty.
Female adolescent status had a significantly negative effect on serum
iron levels. Specifically, the regression coefficient indicates that serum
iron levels for female adolescents was lower by 12 pxg/100 mL than that of
male adolescents (Table 19).
Nutritional education of the homemaker significantly increased the
level of adolescent serum iron in the aggregate sample. The nutrition
education regression coefficient for the entire sample indicates that the
level of serum iron increased by 10.3 pgg/100 mL if the adolescent's
homemaker had participated in nutritional education programs (Table
19). At sample mean nutrient value of 100.60 ig/100 mL (Table 12),
nutrition education of the homemaker was associated with increased
nutrient level of the adolescent of approximately 10 percent.
Serum iron was another nutrient for which the general educational
attainment of the homemaker had a significantly positive impact on the
adolescent nutrient level. As indicated by the nutrition education regres-
sion coefficient, the level of serum iron was increased for adolescents
whose homemaker level of education was above the ninth grade (Table
19).

55








Red Blood Cell Folacin
RBC folacin regression coefficients for the aggregate sample appear in
Table 20. As indicated by the respective coefficients for the aggregate
sample, household income and household size had no significant impact
on the level of adolescent RBC folacin.
The significant coefficient of nutritional education indicates that the
level of RBC folacin increased by 32 ng/mL, or 18.6 percent of sample
mean (Table 12), if the adolescent's homemaker had participated in
nutritional education programs. Female status was associated with a
significant decline (38 ng/mL) in the RBC folacin level of the adolescent,
as was belonging to a black household (53 ng/mL).

Hemoglobin
Hemoglobin summary response for the aggregate sample is presented
in Table 21. For this nutrient indicator, income, household size, age,
education, FSP participation and nutritional education of homemaker
registered no significant impact. The only variables that registered signifi-
cant impacts on the level of hemoglobin were sex of the adolescent and
racial background. Racial composition had a significantly negative im-

Table 20. Statistical summary of OLS nutrient equation: red blood cell folacin
(RF), by households and selected socioeconomic characteristics,
Miami and Sumter County, Florida, 1980.

Socioeconomic Total Regression Standard
variable n= 216 coefficient error t-value

Intercept 206.80 39.71 5.21**
Household income (I)" -0.011 0.011 0.93
Household size (HS)a 1.314 4.22 0.31
Age of homemaker (> 40 years)(A) 6.044 13.69 0.44
Sex of adolescent (female)(S) -38.11 13.16 2.89**
Race (R):
White -14.61 31.06 0.47
Black -53.48 29.25 1.83*
Educational level homemaker (E):
a 9th grade 9.29 18.74 0.50
< 9th grade 2.64 29.25 1.83*
Nutritional education (EDNT) 32.11 14.65 2.19**
FSP-participation (FSP) -4.53 17.10 0.27
Vegetable garden (V) 9.60 16.91 0.57

R2 = 0.1127
F = 2.35**

"Nutrient level, household income, and household size are expressed in linear form. See
page 19 for model specification.
*P < 0.10 (coefficient significant at 90% level).
**P < 0.05 (coefficient significant at 95% level).

56








pact on adolescent hemoglobin level. The greatest impact on the level of
hemoglobin was registered among blacks, where coefficients indicated
that the nutrient level would be lower by 1.2 g/100 mL (9 percent of the
sample mean), if the adolescent was black, compared to 0.53 g/100 mL (4
percent) for whites.

Protein
A summary of the regression analysis for this nutrient indicator is given
in Table 22. The only variable that showed a significant impact was the
nutritional education of the homemaker. The level of adolescent protein
showed a higher level when the homemaker had some kind of nutritional
education. The nutritional education coefficient suggests that adolescent
protein level would have increased by 4.96 gm/100 mL (71 percent of
sample mean) (Table 12) if the homemaker had participated in some type
of nutritional education program. Thus, as expected, nutritional educa-
tion of the household played a major role in improving the nutritional
status of the adolescent. Household income, family size, FSP participa-
tion, and vegetable garden had no significant impact on the level of
adolescent protein.

Table 21. Statistical summary of OLS nutrient equation: hemoglobin (HGB),
by households and selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980.

Socioeconomic Regression Standard
variable coefficient error t-value

Intercept 15.34 0.39 39.50*
Household income (I)a 0.0001 0.0001 0.87
Household size (HS)' -0.03 0.04 0.76
Age of homemaker (< 40 years)(A) 0.08 0.13 0.59
Sex of adolescent (female)(S) -1.02 0.13 7.79**
Race (R):
White -0.53 0.31 1.71*
Black -1.21 0.29 4.16"*
Educational level homemaker (E):
a 9th grade -0.19 0.18 1.01
< 9th grade -0.26 0.29 0.91
Nutritional education (EDNT) -0.09 0.14 0.67
FSP-participation (FSP) 0.02 0.17 0.11
Vegetable garden (V) -0.08 0.16 0.50
R2 = 0.3270
F = 10.07**

"Nutrient level, household income, and household size are expressed in linear form. See
page 19 for model specification.
*P < 0.10 (coefficient significant at 90% level).
**P < 0.05 (coefficient significant at 95% level).

57








Table 22. Statistical summary of OLS nutrient equation': protein (PRTN), by
households and selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980.

Socioeconomic Total Regression Standard
variable n 177 coefficient error t-value

Intercept 6.5 0.64 10.02**
Household income (I)a 0.002 0.002 1.27
Household size (HS)a 0.79 0.72 1.10
Age of homemaker (< 40 years)(A) -3.01 2.41 1.25
Sex of adolescent (female)(S) -0.009 2.31 0.004
Race (R):
White -6.24 5.02 1.14
Black 5.21 4.55 1.29
Educational level homemaker (E):
> 9th grade -4.61 3.57 1.29
< 9th grade -0.63 5.07 0.12
Nutritional education (EDNT) 4.96 2.43 2.04**
FSP-participation (FSP) -4.31 2.92 1.47
Vegetable garden (V) -0.38 3.32 0.11

R' = 0.1818
F =3.33**

"Nutrient level, household income, and household size are expressed in linear form. See
page 19 for model specification.
*P < 0.10 (coefficient significant at 90% level).
**P < 0.05 (coefficient significant 15 95% level).

Vitamin C
Table 23 shows the relationships between the level of vitamin C and the
various socioeconomic characteristics. The only statistically significant
variable was the sex of the adolescent. For this variable, female adoles-
cents showed lower levels of vitamin C than did male adolescents.

Vitamin B12
A summary of the regression analysis for this nutrient is presented in
Table 24. For this nutrient, blacks and households with less than a ninth
grade education showed positive and significant impact. Nutritional
education and FSP participation affected adolescent vitamin B12 nega-
tively.

Zinc
Zinc status responses are presented in Table 25. Regression results of
the aggregate sample showed no significant responsiveness to any of the
variables except sex of adolescent and nutritional education of the house-
hold. Female adolescents had a positive impact on zinc status, but nutri-
tional education of the household had a negative impact on this nutrient.

58








Table 23. Statistical summary of OLS nutrient equation: vitamin C (VITC), by
households and selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980.

Socioeconomic Total Regression Standard
variable n=211 coefficient error t-value

Intercept 1.08 0.17 6.03**
Household income (I)a 0.0001 0.00005 0.97
Household size (HS)a -0.02 0.01 1.17
Age of homemaker (< 40 years)(A) -0.09 0.06 1.51
Sex of adolescent (female)(S) -0.13 0.06 2.04**
Race (R):
White -0.06 0.14 0.45
Black 0.06 0.13 0.49
Educational level homemaker (E):
> 9th grade 0.03 0.09 0.30
< 9th grade 0.16 0.13 1.22
Nutritional education (EDNT) 0.07 0.07 0.95
FSP-participation (FSP) -0.02 0.08 0.23
Vegetable garden (V) 0.007 0.07 0.09

R2 = 0.0614
F = 1.18

"aNutrient level, household income, and household size are expressed in linear form. See
page 19 for model specification.
*P < 0.10 (coefficient significant at 90% level).
**P < 0.05 (coefficient significant 15 95% level).

SUMMARY, CONCLUSIONS, AND POLICY IMPLICATIONS
Results of this study showed that in the aggregate, household income,
household size, and FSP participation exerted a significant impact on
household monthly food expenditure. There was also a significant rela-
tionship between household nutritional education and monthly food
expenditure. The amount of food expenditure was consistently lower for
households whose homemaker had some type of nutritional education,//
relative to those who had none.
Household size and FSP participation showed no consistently signifi-
cant impact on the nutritional status of the adolescent. The one variable
that showed a consistent pattern was nutritional education of the home-
maker. In four of the eight nutrients (serum folacin, serum iron, RBC
folacin, protein), nutritional education had a positive effect that was
statistically significant.
The strong (significant) negative relationship registered between nutri-
tional education and food expenditure did not appear to have adversely
affected the nutritional status of adolescents. On the contrary, as noted
previously, the nutritional education relationship increased the level of
nutrients in 50 percent (four) of the indicators as showed by the param-
59








Table 24. Statistical summary of OLS nutrient equation: vitamin B12 (VITB12),
by households and selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980.

Socioeconomic Total Regression Standard
variable n= 177 coefficient error t-value

Intercept 360.93 50.84 7.09**
Household income ()a' -0.002 0.016 0.13
Household size (HS)* 0.17 5.85 0.030
Age of homemaker (-< 40 years)(A) -17.84 17.58 1.02
Sex of adolescent (female)(S) 21.00 17.01 1.23
Race (R):
White 11.86 38.75 0.39
Black 94.28 37.80 2.50**
Educational level homemaker (E):
9th grade 10.27 26.26 0.39
< 9th grade 71.37 40.33 1.77
Nutritional education (EDNT) -32.58 18.66 1.75*
FSP-participation (FSP) -25.78 22.61 1.74*
Vegetable garden (V) 22.10 21.20 1.04
R2 0.1294
F = 2.23**

"Nutrient level, household income, and household size are expressed in linear form. See
page 19 for model specification.
*P < 0.10 (coefficient significant at 90% level).
**P < 0.05 (coefficient significant 15 95% level).


eter estimates for the aggregate sample. This implies that nutritional
education made households more efficient in terms of food expenditure
per dollar, as well as nutrient value per dollar of expenditure. In other
words, a dollar's worth of food purchased (from the standpoint of both
quantity and quality) was greater for households with nutritional educa-
tion than it was for those with no nutritional education. This finding is
consistent with, and has similar policy implications to that of, the Davis
and Neenan study [11], which found that: (a) policies which combined
income supplement programs such as FSP, with a nutrition education
programs, such as EFNEP, were more effective than other programs
taken individually for increasing the nutritional status of low-income
households, and (b) joint FSP-EFNEP participation was nutritionally
superior to a direct cash supplement program among low-income house-
holds.
In addition, further insight was gained as to the importance of econo-
mies of size within a dynamic household framework. The regression
analysis suggests that household size economies existed for aggregate
food expenditures. This relationship was suggested by the values of the


60








Table 25. Statistical summary of OLS nutrient equation: hair zinc (HAIRZN),
by households and selected socioeconomic characteristics, Miami and
Sumter County, Florida, 1980.

Socioeconomic Total Regression Standard
variable n =249 coefficient error t-value

Intercept 154.08 27.01 5.70**
Household income (I)a -0.004 0.008 0.57
Household size (HS)' -1.47 2.91 0.50
Age of homemaker (- 40 years)(A) 4.04 9.17 0.44
Sex of adolescent (female)(S) 14.75 8.83 1.67*
Race (R):
White 1.19 21.17 0.06
Black -5.97 19.95 0.30
Educational level homemaker (E):
> 9th grade 3.00 12.93 0.23
< 9th grade 17.06 19.82 0.86
Nutritional education (EDNT) -19.22 9.81 1.96*
FSP-participation (FSP) -5.39 11.71 0.46
Vegetable garden (V) 5.81 11.01 0.53

R2 = 0.0465
F =1.05

"Nutrient level, household income, and household size are expressed in linear form. See
page 19 for model specification.
*P < 0.10 (coefficient significant at 90% level).
**P < 0.05 (coefficient significant 15 95% level).


size-expenditure and MPS coefficients for household sizes above and
below the sample mean household size. Also, as shown in Appendix
Table B-2, nutritional education impacted highly on large size house-
holds and less educated homemaker (<9th grade) households. These
household groups were at the lower end of the income spectrum. The
gains from economies of size via nutritional education tended to increase
the average quantity of food purchased by these household groups. Thus,
nutritional education played a key role via the economy of size phe-
nomenon, increasing the household's food purchasing power and nu-
trient status of low income adolescents in the household.
The incidence of nutrient deficiency was highest among rural adoles-
cents, particularly blacks, where the average income of the household
was the lowest. Rural households also reported low participation in ;
EFNEP, but this was obviously related to the fact that EFNEP was not
operational in the rural area surveyed. These findings suggest, among
other things, that program changes, resulting from administrative and
legislative policy modification, might have been a factor mitigating
against realization of food and nutrition policy objectives among this


61








segment of the low-income population, who are at nutritional risk." One
obvious policy implication of this observation is that family food and
nutrition assistance programs at federal, state, and local levels must be
more effectively targeted if policy objectives are to be realized. These
targeted policy objectives should be evaluated ex post and ex ante in
terms of realized or potentially realizable efficiency and equity distribu-
tion impacts.
r" Federal transfer and food assistance expenditures have expanded
tremendously in the last decade. Expenditures for USDA food and
nutrition program increased from slightly more than $1 billion in 1969 to
More than $10 billion in 1979 [17]. In spite of some evidence that funds
Share increasingly being channeled to those areas with evidence of hunger
"/and malnutrition [13], there is also ample evidence that there is persistent
hunger, undernutrition, and malnutrition among low-income segments
of the U.S. population. Analysis suggested that the low-income popula-
tion covered in this study is one such population experiencing nutritional
problems. Furthermore, the magnitude of the problem is more severe
among rural households, particularly rural black households. It follows,
therefore, that policy must be effectively targeted at these populations if
policy objectives are to be realized.
Rurlt households exhibited the highest incidence of adolescent nutri-
tional deficiency. The household income-food expenditure data helped
to define the boundaries of the money-related problem. In the aggregate,
households with income below 75 percent of the poverty level spent, on
the average, 61 percent of the monthly income for food. Those with 75 to
100 percent of the poverty level spent 38 percent of their monthly income
on food, while those with 100 to 125 percent of the poverty level spent 32
percent of their income for food. In contrast, those households that were
"less poor" (above 125 percent of the poverty level) spent 21 percent of
their monthly income on food. Rural black households registered 56
percent of the households in the below 75 percent of the poverty level
status. In short, rural black households exhibited a significantly higher
level of chronic poverty than any other group.
Given the relatively high incidence of poverty income among black
households (rural black households in particular), an obvious question is
whether there was any income-related association between: (a) house-
hold food expenditure and (b) adolescent nutritional status within these
households. A review of the regression results and descriptive data
provide some insight into the answer to this question. The regression
results indicated a positive and significant relationship between house-

11. Persons considered to be at nutrition risk are those with nutrient require-
ment greater than that of the population at large. Adolescents, as a result of their
rapid growth spurt, are vulnerable to nutritional stress.


62








hold income and monthly food expenditure for the aggregate sample.
The estimated aggregate income elasticity for food expenditure was 0.32.
This result suggests that for every 1 percent increase in monthly income,
food expenditure increased by 0.32 percent, when all other variables
were held constant. The estimated income elasticity for food among rural
black households was 0.31. This meant that among these households,
every 1 percent increase in monthly income resulted in a 0.31 percent
increase in food expenditure.
The estimated aggregate MPS for food was 0.094. The MPS for food
ranged from a low of $0.06 for urban households to a high of $0.17 for
hispanic households. There were no significant differences between the
MPS's for black and white households (0.90 and 0.092, respectively).
However, the MPS for rural households (0.117) was considerably larger
than that of urban households (0.064). For every additional dollar in-
crease in monthly income, the rural household spent 11.7 cents compared
to 6.4 cents for urban households.
Given the characteristics of the income elasticity and MPS for food
expenditure, the question posed earlier can be restated in terms of
whether there was any dynamic relationship between these two income-
related characteristics and food expenditure and nutrient status among
chronically poor households. Rural black households had the highest
poverty incidence and the largest household size. The income elasticity
estimate for this household group (0.31) was the lowest among the four
location-related racial groups. As illustrated in food expenditure regres-
sion section, the estimated MPS for rural black households was 0.15.
Comparable MPS's for urban black households, hispanics and rural
white households were 0.086, 0.170, and 0.092, respectively. The high
MPS for rural blacks and hispanics may be due to a tendency by those
household groups to over-report expenditures and under-report house-
hold income. This might also have an effect on the corresponding income
elasticities.
Food Stamp Program participation had a positive impact on rural black
household food expenditure. Descriptive analysis also indicated that 54
percent of rural black households participated in the FSP. The compara-
ble participation rate for all households was only 28 percent. Participa-
tion rates for urban black, hispanic, and rural white households were 33
percent, 36 percent, and 9 percent, respectively. Money income and
household size also had a positive impact on food expenditure among
rural black households. However, nutrition education had a negative \
impact on expenditures. Since rural black households had the highest
participation rate in FSP, and the largest family size, it was interesting to
observe how FSP participation and household size impacted the MPS
estimates. Among FSP participants, the estimated MPS was 0.074, com-
pared to 0.097 among non-participants. This meant that for each addi-


63








tional $1.00, FSP participants food expenditure increased only 7 cents
compared to about 10 cents for non-participants. Given the relatively
high FSP participation rate among rural black households, it appears that
a good portion of their monthly food requirements were being met by
food stamp coupons. Food stamp coupons could have freed up money
income for purchasing of residual food items, hence the lower MPS
among FSP participants.
Black adolescents showed lower mean values than white adolescents
Sfor serum folacin, RBC folacin, hemoglobin, vitamin C, and vitamin B12.
Most adolescents were found to be deficient in folacin, iron, protein, and
zinc. Rural adolescents had a higher incidence of nutrient deficiency than
urban adolescents. However, the highest incidence of nutrient deficiency
for folacin, vitamin C, zinc, and hemoglobin was among rural black
households, the households with the highest incidence of poverty.
A number of policy implications can be drawn from these findings. One
policy implication is that rural black households, as the most economi-
cally and nutritionally disadvantaged group, could have been more effec-
tively targeted for higher levels of income transfers, food assistance,
nutrition education, and employment generating programs. These pro-
grams, while not sufficient to solve problems of underinvestment in
human capital for this group, could have reduced the incidence of poverty
and malnutrition through the interaction of selected socioeconomic vari-
ables discussed previously. For example, given the relatively high posi-
tive interactive impact of FSP participation and nutritional education on
household food expenditure and adolescent nutrient levels among this
group, both the benefits and the coverage of these programs should have
been targeted to this group. Specifically, although the FSP participation
rate was significantly higher among rural black households (54 percent),
the benefits of the program could have been higher if joint policy mecha-
nisms were available that would tie food stamp benefits to participation in
a nutrition education program.
A similar argument holds for an important federal nutrition education
program, the Expanded Food and Nutrition Education Program
(EFNEP). At the time of the survey, the EFNEP was not operational in
rural Sumter County. Thus, the 7 percent of the rural black households,
and 3 percent of the rural white households who were EFNEP partici-
pants, must have received their training in locations other than Sumter
County. Also, given the relatively low level of nutritional education
among rural households, and rural black households in particular (17
percent), the case is strong that an EFNEP in Sumter County would have
interacted positively with a higher level of FSP participation to improve
the nutritional status of rural adolescents, particularly the blacks who
were relatively more disadvantaged.
The implication, in terms of our theoretical framework, is that the

64








quality of life of poor, malnourished households could have been in-
creased by targeting and expanding FSP and EFNEP program benefits
towards these groups. Increased well-being would come through the
interactive impact of both programs, because FSP as an in-kind transfer
program would have freed discretionary income for a wider choice of
consumption items, while the EFNEP would have improved the mix
(characteristics) of food items providing proper nutritional status.
Another policy implication stems from the observed relationships be-
tween unemployment, poverty status, household food expenditure, and
adolescent nutrient level among the chronically poor. The long-run solu-
tion to income-related (poverty) undernutrition and malnutrition is to
find productive employment for those able to work. Rural black house-
holds, with the highest incidence of poverty, had the second highest
percentage of homemakers who were nonparticipants in the labor market
(39 percent). The highest percent of nonworking homemakers was
among hispanics. This relatively high level of unemployment among
poverty-stricken rural black households was obviously associated with
the relatively high participation rate of this group in the FSP. National
studies [17] have found that each 1 percent increase in unemployment
adds an estimated 750,000 to 1 million people to the food stamp rolls.
Many poverty level rural black households with unemployed homemak-
ers apparently had unfulfilled physiological need for food and nutrients.
This need is reflected in the large proportion of money income spent for
food. This unfulfilled food need was also apparent among FSP partici-
pants, since, in spite of in-kind income transfer for food, these low
income households still spent almost 50 percent of their monthly money
income for food. This unfulfilled food need among FSP participating
households could be one explanation why the regression analysis showed
no significant nutrient variation between adolescents from FSP partici-
pating households and those from non-participating households.
Although FSP participation increased the purchasing power of the poor,
it was not enough, nor was it intended to be enough, to meet the
household's physiological need for food and nutrients. In other words, an
effective demand for food and nutrients could have been created by
reducing the income poverty among the group via increased labor force
participation. FSP benefits could then play a supplementary role in
expanding food and nutrient demand. Also, with a higher level of aggre-
gate employment some public revenue expended for FSP could have
been rechanneled into other areas where their opportunity cost was
higher.
The study did not provide answers to all of the questions associated
with nutritional problems of low-income households. The study would
have been more complete if the following aspects had been included in
the analysis: (a) analysis of the 24-hour dietary recall, and statistical

65








comparisons between the dietary nutrient recall parameters and the
biochemical nutrient parameters, (b) a larger sample size, specifically for
urban whites and hispanics, (c) data on individual food items and food
groups, thus making it possible to have a response analysis for each food
group, and (d) nutrition and socioeconomic data series from an earlier
period to facilitate comparative analysis between the pre-program (FSP
and EFNEP) food expenditure and adolescent nutrient status and post-
program status for these levels. Another limitation of the study was that
the survey response for total household income may have included the
value of food stamps for some FSP participant households but not for
others. Some of the households were unwilling to report their food stamp
values.
In spite of aforementioned shortcomings, the empirical findings did
identify important relationships that have national significance for food
and nutrition policy. In addition to the policy recommendations sug-
gested above, the findings provide a basis for the following recommenda-
tions.
1. Inter-disciplinary econutrition research should continue and be in-
tensified. To have an effective food and nutrition data base, interdisci-
plinary teams of scientists from the disciplines of nutrition, economics,
sociology, health and medicine, to name a few, should be encouraged and
supported to undertake comprehensive studies of hunger and malnutri-
tion problems in the United States. Studies should focus on nutrition
surveillance and socioeconomic factors associated with nutritional prob-
lems. Nutrition surveillance should include identification of target
populations who are at nutritional risk levels. Reliable techniques and
methods of measuring nutritional levels should be developed and
adopted. Also, federal nutritional research programs and support should
be broadened to include the socioeconomic, cultural, and political prob-
lems associated with nutritional problems.
2. Federal and state food assistance programs should give emphasis to
nutritional education programs such as the Expanded Food and Nutrition
Program (EFNEP). An important aspect of program upgrading would be
an educational program to increase the awareness of program eligibles to
the relationship between food, nutrition and health. Program eligibility
criteria for a number of welfare programs might have to be amended to
include enrollment in nutrition education programs, if nutrition objec-
tives are to be an integral part of welfare programs.
3. Federal income-transfer and food assistance programs should be
distributed fairly enough to include most of the needy. The findings of
this study suggested that the benefits of these programs have been skewed
somewhat towards urban areas. Given the relatively high incidence of
poverty, unemployment, and malnutrition among the rural population, it
is obvious that antipoverty and nutritional programs were not allocated to
this population in relation to relative need.
66








4. Greater effort is needed to increase the economic base of rural
poverty areas. Such an effort would require the good will, support and
cooperation of the local communities, the business sector, local, state,
and federal government. Two important ingredients for breaking the
vicious circle of poverty and malnutrition in rural areas are educational
and occupational opportunities. This means that priority and emphasis
should be given to reducing the unemployment rate in poverty areas and
to providing higher levels of occupational skills to complement new
occupational opportunities.
5. Further research is needed on the following:
(a) Comparative analysis between the 24-hour dietary recall and the
biochemical procedure.
(b) Evaluation of the nutritional status of low-income households on a
continuous basis so that the impact of the income-transfer programs
could be recorded for further implementation, and evaluation of
food and nutrition policies.
(c) Increased nutritional surveillance and nutritional research at re-
gional and local levels.






























67














APPENDICES


Appendix Table A-1. Poverty income guidelines for all states except Alaska and
Hawaii, 1980.

Nonfarm Farm
Size of family unit family family
---Dollars-
1 3,790 3,250
2 5,010 4,280
3 6,230 5,310
4 7,450 6,340
5 8,670 7,370
6 9,890 8,400
For family units with more than 6 members, add
$1,220 for each additional member in a nonfarm
family and $1,030 for each additional member in a
farm family.

SOURCE: Federal Register [12].


Appendix Table B-1. Average monthly expenditure by type of expenditure and
race, Miami and Sumter County, Florida, 1980.

Urban Rural

Income/Expenditure Black Hispanic Black White

Household income, dollars 881.00 600.00 493.00 1154.00
Food expenditure, dollars 224.00 262.00 237.00 294.00
Percent of income 25.4 43.7 48.1 25.5
Housing expenditure, dollars 210.10 224.10 135.40 238.10
Percent of income 23.8 37.4 27.5 20.6
Clothing expenditure, dollars 84.60 46.70 85.60 57.30
Percent of income 9.6 7.8 17.4 5.0
Medical expenditure, dollars 48.20 97.40 59.30 60.90
Percent of income 5.5 16.2 12.0 5.3
Recreation, dollars 36.30 22.10 27.30 34.10
Percent of income 4.1 3.7 5.5 3.0
Transportation, dollars 53.50 50.00 81.50 137.70
Percent of income 6.1 8.3 1.5 11.9
Alcohol, dollars / 15.70 30.00 12.00 13.50
Percent of income 1.8 5.0 2.4 1.2
Tobacco, dollars 15.70 18.70 13.50 26.60
Percent of income 1.8 3.1 2.7 2.3



69








Appendix Table B-2. Statistical summary of OLS monthly food expenditure equation b,
household category and selected socioeconomic characteristics, Miam
and Sumter County, Florida, 1980.


Household residence FSP participation

Independent Non-
variable Urban Rural Participants participant

Intercept 3.19** 1.98** 3.38** 2.62**
(6.20)' (4.77) (8.41) (6.17)
Household income (I) 0.23** 0.40** 0.15** 0.39**
(2.95) (7.38) (2.64) (6.78)
Household size (HS) 0.59** 0.47** 0.85** 0.41**
(5.39) (4.06) (8.81) (4.10)
Age of homemaker -0.12 0.13* 0.12* 0.01
(1.44) (1.95) (1.70) (0.09)
Ethnicity:
White 0.08 -0.34**
(0.49) (2.30)
Black -0.11 -0.47**
(0.78) (3.33)
Educational level:
<9th grade -0.14 0.23 -0.29 -0.16
(1.09) (1.15) (1.58) (1.11)
9th-12th grade -0.21* 0.11 -0.17 0.01
(1.94) (1.11) (1.04) (0.15)
Nutritional education -0.04 -0.13 -0.10 -0.15**
(0.46) (1.48) (1.36) (2.10)
FSP participation 0.08 0.21**
(0.79) (2.12)
R2 0.3261 0.4101 0.6279 0.3674
F 8.29** 12.81** 13.29** 13.36**

"Food expenditure, household income, and household size are expressed in logarithmic form.
'Numbers in parentheses are values of Student's t test.
"Ethnicity variable is excluded from this sample group because of high multicollinearity between race an(
location.
"*P < 0.10
**P < 0.05














70












Nutritional
education Family size Educational level

Yes No 2-7 5-7 <9th grade 9-12th

2.98** 2.46** 2.91** 3.80** 2.52 3.35**
(5.61) (5.18) (5.41) (7.41) (1.71) (9.36)
0.31** 0.34** 0.40** 0.32** 0.46** 0.28**
(4.25) (5.94) (5.23) (4.82) (3.28) (5.71)
0.58** 0.50** 0.83** 0.48**
(4.88) (4.83) (3.07) (5.46)
-0.01 0.09 -0.04 0.11 -0.03 0.07
(0.14) (1.28) (0.57) (1.34) (0.14) (1.16)

-0.30** 0.00 -0.15 -0.23 -0.26 -0.39**
(2.00) (0.01) (0.73) (1.42) (1.14) (2.13)
-0.50** -0.14 -0.30 -0.41** -0.33** -0.60**
(3.82) (0.55) (1.56) (2.56) (1.80) (3.36)

-0.45** 0.12 -0.20 -0.12
(2.76) (.73) (1.06) (0.73)
-0.16 0.02 -0.05 -0.08
(1.19) (0.29) (0.53) (0.63)
-0.08 -0.22* -0.43** -0.11*
(0.89) (2.49) (2.38) (1.78)
0.18* 0.19** 0.12 0.26** 0.20 0.17**
(1.80) (2.03) (1.11) (2.54) (1.07) (2.27)
0.3713 0.4090 0.3067 0.3055 0.6126 0.3820
7.31** 12.80** 6.63** 6.21"* 5.42** 16.07**




















71









Appendix Table C-1. Functions of, problems associated with deficiency of, and food sources of selected nutrients.

Nutrient Problems associated Major
Nutrient indicator Major function with deficiency food sources

Iron Hemoglobin, Necessary for formation of hemoglo- Weakness and susceptibility to Liver, organ
serum iron bin, the oxygen-carrying pigment of fatigue. Severe deficiency leads meats, leafy
red blood cells, needed for growth, to anemia. Serum iron and vegetables, dried
menstrual losses, hemoglobin are indicators, fruits, and
cereals.

Folacin Serum folacin, Stimulate regeneration of red blood Toxemia of pregnancy, rheuma- Pork liver, car-
red blood cell cells and hemoglobin. Helps in trans- toid arthritis. Severe damage rots, asparagus,
folacin ferring and conversion of one sub- and depletion of maternal re- kidney beans,
stance to another. Involved in blood serves, and round steak.
formation.

Protein Total Essential for normal growth and de- Kwashiorkor-impaired growth; Peanut butter,
protein velopment; for maintenance and re- weak muscles, hair, and skin; di- dried milk, dried
pair old body tissue. Regulates the arrhea; and anemia. Ability to peas, soy pro-
distribution of fluid, prevents accu- combat infection very low. tein, cheese,
mulation of too much acid, stimulates tuna, eggs, liver,
antibodies to help combat infection, ham, and ham-
burger.

Vitamin C Vitamin C Plays a major role in normal resist- Fatigue, weakness, muscle Citrus fruits,
ance to infections. Production of in- cramps, aching bones, shortness broccoli, toma-
tercellular cementing substance; of breath. Decreased physical toes, green pep-
wound healing. Helps in utilization of performance; delayed wound per, raw cab-
iron, calcium, and folic acid. healing, bage, strawber-
ries, and mus-
tard greens.





Appendix Table C.1. (cont.)

Nutrient Problems associated Major
Nutrient indicator Major function with deficiency food sources

Vitamin B12 Vitamin B12 Necessary for normal growth, mainte- Pernicious anemia. Beef liver,
nance of healthy nervous tissue, and clams, lamb,
normal blood formation, eggs, salmon,
and beet greens.

Zinc Hair zinc Part of enzymes involved in energy Failure to maintain pregnancy, Oysters, round
and protein metabolism maintains the fetal malformations; retardation. steak, lamb
integrity of skin and hair. Necessary chops, peanuts,
for mobilization of Vitamin A. Essen- whole wheat,
tial to brain growth; reproduction. oats, and beans.

SOURCE: Adapted from Lancaster [15].







Appendix Table D-1. Statistical summary of OLS nutrient equation by household category
dependent variable, serum folacin (SF); Miami and Sumter County
Florida, 1980.
Sex Race

Independent Male Female White Black Hispanic
variable n=107 n=104 n=81 n=120 n=13

Intercept 6.55 4.09 0.80 10.4** 11.49
(1.35)a (1.36) (0.22) (4.22) (0.52)
Income 0.001 0.0008 0.002* -0.001 -0.01
(0.87) (0.74) (1.79) (0.62) (0.62)
Household size -0.02 0.55** 0.93 -0.03 -0.31
(0.04) (1.67) (1.52) (0.08) (0.14)
Age s 41 -2.80* -1.20 0.38 -2.56** -10.1
(1.69) (1.13) (0.23) (2.13) (1.31)
Sex Female -2.95* -2.39** -3.28
(1.94) (2.08) (0.47)
Race:
White 1.74 -1.84
(0.40) (0.85)
Black 2.99 -1.37
(0.70) (0.69)
Educational level:
9-12th grade -0.16 0.14 -0.07 -0.05 16.7
(0.08) (0.09) (0.03) (0.03) (0.97)
<9th grade 8.00** 0.31 3.82 5.04* 12.3
(2.12) (0.14) (1.00) (1.91) (1.02)
Nut. education-yes 0.14 2.82** 2.67 1.66 2.97
(0.08) (2.47) (1.35) (1.43) (0.26)
Vegetable garden-yes 0.71 0.21 2.23 -1.89 2.66
(0.38) (0.15) (1.33) (1.11) (0.27)
F 1.22 1.67* 2.03** 1.81* 0.98
R2 0.1131 0.1406 0.2025 0.1283 0.594

"Numbers in parentheses are values of Student's t test.
*t-values significant at P < 0.10.
**t-values significant at P < 0.05.















74











Location FSP participation Employed

Urban Rural Yes No Yes No
n=16 n=109 n=160 n= 152 n=152 n=63

10.1** 4.48* 3.84 8.46** 8.20** 1.02
(3.48) (1.66) (0.76) (2.48) (8.29) (0.19)
0.002 -0.002** -0.004 0.0001 0.0007 0.002
(1.47) (2.27) (1.46) (1.30) (0.63) (1.35)
0.49 0.21 1.20** -0.10 0.14 0.49
(1.19) (0.49) (2.76) (0.26) (0.32) (1.06)
-2.87* -0.61 -3.56** -1.94* -1.85 -4.38**
(1.89) (0.50) (2.11) (1.72) (1.64) (2.19)
-2.33* -2.10* -0.84 -2.93** -3.14** -0.42
(1.69) (1.76) (0.48) (2.66) (2.82) (0.19)

1.72 -1.11 0.19 0.10
(0.50) (0.41) (0.06) (0.03)
1.24 -0.07 0.31 -0.21
(0.42) (0.03) (0.10) (0.07)

0.02 -0.14 -2.96 0.89 0.04 2.10
(0.01) (0.08) (0.83) (0.62) (0.03) (0.47)
3.02 2.97 2.02 2.89 3.60 4.60
(1.27) (0.91) (0.50) (1.17) (1.39) (0.95)
1.15 1.55 3.58** 0.88 0.48 4.75**
(0.77) (0.92) (2.06) (0.69) (0.40) (2.12)
- 1.64 1.46 -3.66 1.46 -0.51 2.35
(0.75) (1.12) (1.44) (1.09) (0.37) (0.93)
2.00** 1.88* 3.09** 1.58 1.32 1.59
0.1486 0.1482 0.3871 0.0960 0.0941 0.2548



















75








Appendix Table D-2. Statistical summary of OLS nutrient equation by household category,
dependent variable, serum iron (IR); Miami and Sumter County,
Florida, 1980.

Sex Race

Independent Male Female White Black Hispanic
variable n=107 n= 104 n= 81 n= 120 n=13

Intercept 125.5** 62.2** 120.0** 100.8* 114.2
(5.40)" (2.66) (5.53) (6.08) (1.08)
Income -0.006 0.009 -0.008 0.01 0.04
(0.89) (1.14) (1.36) (1.60) (0.37)
Household size -1.75 -3.82* -5.14 -3.89* -4.79
(0.69) (1.65) (1.42) (1.89) (0.66)
Age R! 41 0.88 -3.54 3.52 -5.39 0.10
(0.11) (0.45) (0.36) (0.77) (0.003)
Sex Female 0.35 14.7** -62.5
(0.04) (2.17) (1.82)
Race:
White -26.83 32.9*
(1.34) (91.76)
Black -18.47 26.42
(0.95) (1.59)
Educational level:
9-12th grade 13.90 17.04 20.1* 12.6 60.6
(1.34) (1.50) (1.69) (1.23) (1.07)
< 9th grade 7.36 16.81 6.56 20.6 13.01
(0.41) (1.02) (0.28) (1.32) (0.25)
Nut. education-yes 10.03 9.21 6.56 11.51** -34.2
(1.14) (0.99) (0.54) (1.69) (0.75)
Vegetable garden-yes 10.92 -12.48 -26.47** 2.97 24.3
(1.21) (1.24) (2.60) (0.31) (0.41)
F 1.47 1.19 1.66 1.84* 1.15
R2 0.1285 0.1135 0.1739 0.1310 0.7753

"Numbers in parentheses are values of Student's t test.
*t-values significant at P < 0.10.
"**t-values significant at P < 0.05.















76












Location FSP participation Employed

Urban Rural Yes No Yes No
n=16 n=109 n=60 n=160 n=144 n=63

109.0** 106.6** 117.4** 99.4** 94.9** 126.4**
(7.61) (5.89) (3.30) (5.04) (4.43) (3.88)
0.009 0.005 0.007 -0.001 0.008 -0.006
(1.14) (0.76) (0.38) (0.29) (1.27) (0.86)
-3.91* -3.29 -1.24 5.34** -5.27** -0.86
(1.96) (1.16) (0.42) (2.43) (2.19) (0.33)
4.29 -2.50 -7.98 2.83 2.02 -20.7*
(0.55) (0.31) (0.68) (0.45) (0.32) (1.73)
-2.83** -4.57 -17.6 -8.87 -9.61 -11.6
(3.12) (0.57) (1.53) (1.45) (1.56) (0.97)

-21.0 16.5 5.16 1.00
(0.84) (0.99) (0.27) (0.04)
-15.1 13.6 6.14 4.14
(0.70) (0.84) (0.33) (0.22)

15.26 16.38 4.85 16.6** 19.9** -8.52
(1.51) (1.44) (0.20) (2.09) (2.49) (0.32)
0.43 23.91 -6.32 7.6 18.15 -17.5
(0.04) (1.08) (0.22) (0.52) (1.20) (0.62)
6.18 -11.82 19.05 11.30* 3.29 17.6
(0.84) (1.01) (1.00) (1.63) (0.49) (1.39)
-10.47 -7.48 19.21 -18.6** -15.2** 0.59
(0.97) (0.86) (1.14) (2.50) (2.01) (0.04)
2.77* 0.83 0.80 2.32** 2.05** 0.76
0.2063 0.0698 0.1382 0.1395 0.1460 0.1363
"Numbers in parentheses are values of Student's t test.
*t-values significant at P 0.10.
"**t-values significant at P < 0.05.















77








Appendix Table D-3. Statistical summary of OLS nutrient equation by household category,
dependent variable, RBC folacin (RF); Miami and Sumter County.
Florida, 1980.

Sex Race

Independent Male Female White Black Hispanic
variable n= 107 n=109 n=78 n=121 n=16

Intercept 256.3** 80.2 158.6** 161.9** 594.8
(4.12)a (1.56) (2.40) (5.6) (1.78)
Income -0.01 -0.001 -0.0002 -0.026* -0.41
(0.61) (0.06) (0.01) (1.78) (1.57)
Household size -5.9 12.48** 6.38 0.78 4.49
(0.89) (2.22) (0.59) (0.19) (0.14)
Age 41 4.48 8.92 35.3 -10.2 -10.2
(0.21) (0.49) (1.20) (0.72) (0.08)
Sex Female -76.6** -11.3 -104.0
(2.84) (0.83) (0.99)
Race:
White -12.71 -15.9
(0.22) (0.43)
Black -83.4 19.4
(1.52) (0.59)
Educational level:
9-12th grade 0.07 23.7 15.3 -0.04 14.5
(0.003) (0.90) (0.41) (0.002) (0.06)
<9th grade -12.5 26.3 5.22 11.7 45.0
(0.26) (0.71) (0.08) (0.38) (0.25)
Nut. education-yes 46.1** 13.2 8.92 44.4** -38.3
(2.02) (0.65) (0.24) (93.25) (0.22)
Vegetable garden-yes 19.7 -4.23 25.2 -4.54 -56.9
(0.83) (0.17) (0.083) (0.23) (0.39)
F 1.85* 1.24 1.23 1.85* 0.79
R2 0.1617 0.1127 0.1398 0.1324 0.540c

"aNumbers in parentheses are values of Student's t test.
"*t-values significant at P < 0.10.
"**t-values significant at P < 0.05.















78












Location FSP participation Employed
Urban Rural Yes No Yes No
n=114 n=104 n=60 n=156 n=149 n=62

194.2** 170.8** 211.6** 209.6** 214.7** 191.8**
(6.82) (3.93) (3.15) (4.16) (4.02) (2.57)
-0.05** 0.01 -0.12** -0.002 -0.003 -0.02
(3.14) (0.73) (3.35) (0.14) (0.16) (1.13)
3.79 -0.67 11.6** -0.94 2.17 1.05
(0.82) (0.09) (2.00) (0.16) (0.34) (0.17)
-13.2 12.18 1.10 2.58 8.08 3.62
(0.77) (0.54) (0.05) (0.16) (0.48) (0.13)
-28.9* -45.5** -16.9 -50.7** -42.4** -31.3
(1.83) (2.05) (0.74) (3.08) (2.52) (1.06)

14.5 -16.2 -32.6 17.7
(0.31) (0.44) (0.73) (0.36)
-11.5 -52.6 -77.3** 17.7
(0.29) (1.36) (1.76) (0.66)

-51.3 13.9 15.4 -10.3
(1.09) (0.66) (0.71) (0.17)
-80.9 33.1 34.2 -52.5
(1.50) (0.92) (0.89) (0.80)
47.5** 1.40 46.3** 25.2 27.7 46.7
(2.85) (0.04) (2.00) (1.34) (1.50) (1.5)
-14.2 30.2 -34.6 19.2 6.86 -0.27
(0.57) (1.24) (1.02) (0.96) (0.33) (0.01)
3.55** 1.08 1.83 2.53** 2.21"* 0.69
0.2145 0.0843 0.2715 0.1485 0.1505 0.1322



















79








Appendix Table D-4. Statistical summary of OLS nutrient equation by household category,
dependent variable, hemoglobin (HGB); Miami and Sumter County,
Florida, 1980.

Sex Race

Independent Male Female White Black Hispanic
variable n=121 n=119 n=92 n= 129 n=18

Intercept 15.64** 14.10 14.98** 14.07** 18.14**
(15.40)a (27.42) (20.15) (34.51) (9.93)
Income 0.0001 0.0001 0.0002 0.0000 -0.002
(0.82) (0.52) (1.39) (0.05) (1.43)
Household size -0.08 0.02 -0.08 -0.02 0.09
(1.20) (0.26) (0.86) (0.29) (0.63)
Age > 41 0.20 -0.03 -0.10 -0.25 -0.18
(0.94) (0.19) (0.43) (1.27) (0.30)
Sex Female -1.02** -1.04** -1.03
(4.72) (5.52) (1.78)
Race:
White -0.70 -0.42
(1.30) (1.10)
Black 1.35** -1.10**
(2.36) (3.16)
Educational level:
9-12th grade -0.06 -0.33 -0.19 -0.22 -1.20
(0.23) (1.26) (0.69) (0.81) (1.08)
< 9th grade -0.31 -0.23 -0.74 -0.02 -0.89
(0.64) (0.61) (1.38) (0.03) (0.96)
Nut. education-yes -0.24 0.01 0.06 -0.11 -1.05
(0.99) (0.06) (0.21) (0.59) (1.20)
Vegetable garden-yes -0.21 -0.02 -0.19 -0.08 0.20
(0.89) (0.08) (0.84) (0.30) (0.24)
F 2.34** 2.55* 3.54** 3.70** 1.95
R2 0.1755 0.1909 0.2796 0.2186 0.6868

"Numbers in parentheses are values of Student's t test.
"*t-values significant at P < 0.10.
"**t-values significant at P < 0.05.















80












Location FSP participation Employed
Urban Rural Yes No Yes No
n=116 n=124 n=64 n=176 n=161 n=73
14.68** 14.53** 14.24** 15.68** 15.61* 14.92**
(36.68) (32.18) (14.43) (35.04) (30.06) (22.00)
-0.0002 0.0001 0.0001 0.0001 0.0001 0.0001
(0.61) (1.52) (0.28) (1.00) (1.07) (0.31)
-0.03 -0.05 0.01 -0.04 -0.02 -0.05
(0.44) (0.76) (0.16) (0.68) (0.34) (0.86)
0.03 -0.11 (0.35) 0.02 0.15 -0.17
(0.12) (0.51) (1.06) (0.16) (0.92) (0.68)
-1.0** -1.2** -0.81** -1.06** -1.19** -0.60**
(5.17) (5.72) (2.49) (7.21) (7.49) (2.33)

0.40 -0.79* -0.74* -0.51
(0.58) (2.18) (1.68) (1.10)
-0.81 -1.48* -1.51* 1.07*
(1.38) (4.26) (3.47) (2.63)

-0.58** 0.09 -0.08 -0.19 -0.16 0.06
(2.09) (0.31) (0.11) (1.03) (0.77) (0.12)
-0.02 0.21 -0.51 -0.26 -0.08 -0.32
(0.06) (0.37) (0.63) (0.78) (0.21) (0.57)
0.07 -0.35 0.12 -0.14 -0.19 0.20
(0.35) (1.19) (0.36) (0.83) (1.06) (0.72)
-0.57* -0.57 0.67 0.18 -0.40** 0.63**
(1.90) (0.64) (1.40) (1.03) (2.06) (2.03)
3.70** 5.23** 2.11** 9.50** 9.04** 2.34**
0.2389 0.2921 0.2844 0.3654 0.4002 0.2969



















81








Appendix Table D-5. Statistical summary of OLS nutrient equation by household category,
dependent variable, protein (PRTN); Miami and Sumter County,
Florida, 1980.

Sex Race

Independent Male Female White Black Hispanic
variable n=86 n=91 n=50 n= 108 n=18

Intercept 58.34** 64.44** 57.39** 67.53** 97.24**
(5.54)' (8.16) (5.87) (10.13) (2.83)
Income -0.0006 0.0004* 0.0002 0.006** -0.04
(0.19) (1.74) (0.11) (2.07) (1.63)
Household size 0.34 1.28 -0.66 0.98 3.74
(0.28) (1.41) (0.52) (0.96) (1.35)
Age > 41 0.12 -4.95* -6.93* -4.30 26.88**
(0.03) (1.77) (1.80) (1.31) (2.40)
Sex Female 4.25 -1.30 -13.63
(1.21) (0.41) (1.25)
Race:
White -1.82 -7.93
(0.20) (1.43)
Black 9.91 2.95
(1.13) (0.60)
Educational level:
9-12th grade 1.01 -8.40* 6.33 -5.59 -1.67
(0.17) (1.84) (0.90) (1.21) (0.08)
< 9th grade 5.13 -2.52 7.19 1.78 4.35
(0.55) (0.41) (0.73) (0.24) (0.25)
Nut. education-yes 8.35* 1.50 5.63 3.49 -10.19
(1.95) (0.49) (1.37) (1.11) (0.61)
Vegetable garden-yes -2.70 4.54 -2.74 1.39 -25.59
(0.54) (1.01) (0.65) (0.28) (1.68)
F 1.82* 3.07** 1.11 1.87* 1.34
R2 0.1952 0.2771 0.1992 0.1464 0.6012

"Numbers in parentheses are values of Student's t test.
*t-values significant at P < 0.10.
"**t-values significant at P < 0.05.















82












Location FSP participation Employed
Urban Rural Yes No Yes No
n=112 n=65 n=53 n=124 n=115 n=56

67.38** 59.85** 39.9** 71.0** 63.65** 60.16**
(9.90) (8.44) (2.62) (8.82) (7.06) (5.19)
0.005 -0.0001 -0.0000 0.003 .006** -0.0003
(1.54) (0.08) (0.003) (1.27) (2.17) (0.08)
1.09 -0.14 2.02* 0.12 0.50 1.18
(1.10) (0.16) (1.78) (0.11) (0.43) (1.15)
1.06 -4.57* -2.06 -2.46 -3.67 -4.35
(0.30) (1.66) (0.45) (0.84) (1.21) (0.96)
-1.92 1.89 0.65 0.95 -3.45 0.43
(0.59) (0.68) (0.13) (0.33) (1.14) (0.09)

8.79 -10.19 -5.17 -5.44
(0.96) (1.61) (0.70) (0.63)
15.49* 0.36 6.77 3.66
(1.95) (0.06) (0.96) (0.54)

-2.26 2.45 1.71 -5.41 -1.86 -3.30
(0.48) (0.44) (0.15) (1.38) (0.43) (0.35)
-2.10 -4.70 5.98 -1.94 6.53 -3.17
(0.37) (0.56) (0.48) (0.31) (0.93) (0.30)
0.88 2.87 2.91 7.09** 1.84 8.71
(0.25) (0.84) (0.58) (2.31) (0.61) (0.44)
1.26 -2.47 5.17 -1.90 -6.27 7.12
(0.25) (0.70) (0.70) (0.48) (1.44) (1.01)
0.99 0.63 1.13 3.16** 2.19** 1.54
0.0802 0.0941 0.2127 0.2186 0.1896 0.2777



















83








Appendix Table D-6. Statistical summary of OLS nutrient equation by household category,
dependent variable, Vitamin C (VITC); Miami and Sumter County,
Florida, 1980.

Sex Race

Independent Male Female White Black Hispanic
variable n=79 n=112 n=78 n=114 n=18

Intercept 0.92** 1.10** 1.26** 2.25** 2.25**
(0.48)a (4.28) (4.84) (2.33) (2.33)
Income 0.00005 0.00004 0.00007 -0.0015** -0.0015**
(0.64) (0.48) (1.03) (2.13) (2.13)
Household size -0.03 -0.02 -0.08* 0.01 0.01
(1.04) (0.54) (1.84) (0.15) (0.15)
Age 40 -0.11 -0.09 -0.17 0.25 0.25
(1.09) (0.98) (1.44) (0.79) (0.79)
Sex Female -0.17 -0.32 -0.32
(1.63) (1.06) (1.06)
Race:
White 0.04 -0.10
(0.19) (0.52)
Black 0.23 -0.02
(1.04) (0.09)
Educational level:
9-12th grade 0.15 -0.13 0.14 0.12 0.09
(1.24) (0.90) (0.98) (0.20) (0.76)
< 9th grade 0.38* -0.01 0.30 0.42 0.07
(1.81) (0.05) (1.23) (0.86) (0.46)
Nut. education-yes -0.04 0.18* 0.02 -0.52 -0.07
(0.39) (1.81) (0.12) (1.12) (0.83)
Vegetable garden-yes 0.12 -0.13 -0.12 -0.26 0.24*
(1.12) (1.08) (1.07) (0.61) (1.86)
F 0.77 0.94 1.09 1.03 1.14
R2 0.0806 0.0855 0.1259 0.5623 0.0429

"Numbers in parentheses are values of Student's t test.
*t-values significant at P < 0.10.
"**t-values significant at P < 0.05.















84












Location FSP participation Employed

Urban Rural Yes No Yes No
n=116 n=95 n= 54 n= 158 n= 141 n= 64

0.99* 1.26** 1.28** 1.03** 0.89** 1.70**
(5.75) (5.91) (2.80) (4.82) (3.49) (6.35)
.0000 .00006 .00001 .00007 .00008 .00005
(0.12) (0.88) (0.46) (1.24) (1.12) (0.64)
-0.01 0.07** -0.01 -0.04 -0.03 -0.01
(0.50) (2.08) (0.31) (1.59) (1.11) (0.56)
-0.02 -0.17* -0.04 -0.11 -0.10 -0.16
(0.24) (1.72) (0.23) (1.52) (1.24) (0.49)
-0.01 -0.24** -0.08 0.13" -0.13 -0.04
(0.07) (2.52) (0.53) (1.78) (1.52) (0.40)

-0.19 0.04 0.06 -0.05
(0.64) (0.24) (0.28) (0.28)
-0.30 0.20 0.26 -0.28
(1.26) (1.22) (1.23) (1.84)

0.10 0.10 -0.08 0.07 0.11 -0.60*
(0.70) (0.70) (0.23) (0.74) (1.03) (2.80)
0.33 0.33 0.08 0.12 0.18 -0.40*
(1.39) (1.39) (0.20) (0.74) (0.96) (1.68)
0.08 0.08 0.03 0.08 0.06 -0.06
(0.60) (0.60) (0.22) (1.03) (0.64) (0.49)
-0.07 0.08 0.30 -0.04 -0.04 0.10
(0.70) (0.60) (1.46) (0.41) (0.39) (0.83)
0.53 1.84* 0.89 1.59 1.10 2.05**
0.1630 0.1630 0.1719 0.0984 0.0860 0.3021



















85








Appendix Table D-7. Statistical summary of OLS nutrient equation by household category,
dependent variable, Vitamin B12 (VITB12); Miami and Sumter County,
Florida, 1980.
Sex Race

Independent Male Female White Black Hispanic
variable n=92 n=85 n=76 n=85 n=-15

Intercept 369.1** 406.8* 350.5** 482.5** 299.7*
(5.86)a (4.55) (5.76) (7.02) (2.56)
Income 0.01 -0.04 -0.01 0.009 0.14
(0.72) (1.12) (0.65) (0.23) (1.40)
Household size -1.05 0.21 7.81 -7.67 -24.3*
(0.14) (0.02) (0.80) (0.82) (2.28)
Age 5 40 -35.9 -0.34 -4.47 -41.2 -37.7
(0.59) (0.12) (0.17) (1.41) (1.07)
Sex Female 35.2 -4.82 80.1*
(1.47) (0.17) (2.08)
Race:
White 0.28 26.3
(0.01) (0.44)
Black 116.2** 70.2
(2.24) (1.21)
Educational level:
9-12th grade 1.80 15.9 5.62 30.7 -5.83
(0.06) (0.32) (0.16) (0.68) (0.09)
< 9th grade 77.0 63.0 13.4 168.3** -26.3
(1.56) (0.88) (0.22) (2.29) (0.41)
Nut. education-yes -48.9* -28.7 -30.0 -41.8 29.8
(1.98) (0.88) (1.00) (1.40) (0.56)
Vegetable garden-yes 15.2 32.6 17.6 27.7 166.2*
(0.59) (0.86) (0.66) (0.68) (2.13)
F 2.58** 0.63 1.13 1.00 1.97
R2 0.2418 0.0789 0.1333 0.1072 0.7795

"Numbers in parentheses are values of Student's t test.
*t-values significant at P < 0.10.
"**t-values significant at P < 0.05.















86












Location FSP participation Employed
Urban Rural Yes No Yes No
n=74 n=103 n=47 n=130 n=118 n=57

"462.5** 376.1** 407.9** 369.7** 389.4** 250.7*
(2.56) (6.96) (3.11) (6.40) (5.65) (2.54)
-0.01 -0.01 -0.10 -0.001 -0.007 0.01
(0.25) (0.82) (1.15) (0.08) (0.31) (0.43)
-1.90 7.09 -9.60 0.08 -0.64 1.58
(0.20) (0.89) (0.79) (0.01) (0.08) (0.18)
0.15 4.70 6.52 -27.2 -14.2 -3.3
(0.004) (0.21) (0.17) (1.34) (0.65) (0.10)
-34.20 53.9** -29.3 23.1 15.9 21.4
(1.16) (2.38) (0.70) (1.17) (0.75) (0.61)

-25.8 14.1 7.8 29.5
(0.29) (0.32) (0.14) (0.47)
145.1* 87.4** 62.5 133,0
(1.78) (2.03) (1.13) (0.47)

56.2 28.9 -19.7 14.0 9.2 53.7
(1.25) (0.52) (0.21) (0.51) (0.31) (0.67)
80.8 40.8 104.9 46.8 32.1 138.8
(1.51) (0.66) (0.90) (1.04) (0.60) (1.50)
-77.6** -4.24 11.14 -42.6* -34.1 -13.8
(2.41) (0.14) (0.24) (1.95) (1.47) (0.35)
6.40 17.17 54.65 20.7 26.9 2.94
(0.14) (0.69) (0.89) (0.88) (1.04) (0.27)
1.01 0.99 1.49 1.84* 1.00 1.25
0.1245 0.0877 0.2927 0.1339 0.0856 0.2134



















87








Appendix Table D-8. Statistical summary of OLS nutrient equation by household category,
dependent variable, hair zinc (HAIRZN); Miami and Sumter County,
Florida, 1980.

Sex Race

Independent Male Female White Black Hispanic
variable n= 122 n = 127 n=99 n = 131 n = 18

Intercept 149.5** 164.7** 159.0 152.1** 108.9
(3.55) (4.63) (5.01) (5.11) (0.46)
Income -0.003 -0.002 -0.007 -0.002 0.10
(0.31) (0.22) (0.79) (0.11) (1.80)
Household size -7.31 3.62 -0.16 -3.98 17.2**
(1.61) (0.92) (0.03) (0.96) (2.87)
Age 41 0.05 8.33 1.21 -5.03 43.2
(0.004) (0.68) (0.09) (0.35) (1.78)
Sex Female 19.6 8.56 59.8**
(1.48) (0.64) (2.58)
Race:
White 19.9 -6.52
(0.55) (0.25)
Black 29.1 -22.9
(0.83) (0.95)
Educational level:
9-12th grade 13.9 -11.1 -1.97 10.7 -74.1
(0.72) (0.62) (0.11) (0.53) (1.64)
< 9th grade 56.9* -8.3 -0.96 27.1 5.86
(1.75) (0.32) (0.03) (0.87) (0.15)
Nut. education-yes -29.4* -15.7 -40.8** -14.0 -6.7**
(1.84) (1.21) (2.37) (1.04) (2.99)
Vegetable garden-yes 9.28 1.43 13.9 4.00 -99.4**
(0.57) (0.09) (0.98) (0.21) (3.02)
F 0.93 -0.55 1.16 0.37 5.29
R2 .0769 .0450 .1046 .0270 .8561

"Numbers in parentheses are values of Student's t test.
"*t-values significant at P < 0.10.
"**t-values significant at P < 0.05.
















88












Location FSP participation Employed

Urban Rural Yes No Yes No
7= 118 n = 131 n= 68 n= 181 n= 168 n= 80

46.5** 168.1** 99.8* 158.3* 168.3** 160.2**
(5.03) (6.04) (1.83) (4.61) (4.23) (4.06)
-0.003 -0.004 0.02 -0.006 0.001 -0.004
(0.26) (0.42) (0.65) (0.72) (0.10) (0.48)
-1.80 -2.97 -6.98* 2.48 -4.64 -1.95
(0.43) (0.69) (1.84) (0.59) (1.02) (0.59)
9.32 -5.96 20.9 -1.00 -2.01 7.64
(0.61) (0.49) (1.35) (0.09) (0.17) (0.55)
1.92 26.7** 14.0 11.6 12.6 24.8*
(0.14) (2.27) (0.94) (1.05) (1.08) (1.77)

-10.1 -6.92 -8.7 29.7
(0.31) (0.25) (0.28) (1.11)
-7.24 -15.3 -14.3 7.5
(0.26) (0.51) (0.43) (0.32)

4.73 -1.65 62.8 -1.24 11.0 -29.8
(0.24) (0.09) (1.52) (0.09) (0.69) (1.1)
21.1 5.39 77.8* 1.58 3.2 6.84
(0.86) (0.16) (1.78) (0.06) (0.11) (0.22)
-8.58 -49.5** -7.68 -25.7** -26.2** -7.9
(0.57) (3.03) (0.48) (2.06) (2.04) (0.49)
-4.56 10.5 37.8* 1.09 12.8 -20.9
(0.21) (0.81) (1.72) (0.08) (0.88) (1.18)
0.29 1.89* 1.32 0.88 0.97 1.17
.023 .1231 .1880 .0494 .0580 .01448



















89














REFERENCES
[1] Adrian, John, and Raymond Daniel. 1976. "Impact of Socioeconomic Factors
on Consumption of Selected Food Nutrients in the United States." Amer-
ican Journal of Agricultural Economics 58(1):31-38.
[2] Allen, R. G. D., and A. L. Bowley. 1955. Family Expenditure. Staples Press,
London.
[3] Bagali, Jagadev N. 1975. "Nutritional and Nonnutritional Components of
Demand for Food." Ph.D. dissertation, University of Florida, Gaines-
ville.
[4] Becker, G. S. 1976. The Economic Approach to Human Behavior. The
University of Chicago Press, Chicago, Illinois.
[5] Blanciforti, L., R. Green, and S. Lane. 1981. "Income and Expenditure for
Relatively More Versus Relatively Less Nutritious Food over the Life
Cycle." American Journal of Agricultural Economics 63(2):225-260.
[6] Boehm, William T. 1979. "Farm Structure and a Changing Food Policy
Environment." Pages 263-268 in Structure Issues of American Agricul-
ture. Agricultural Report 438, USDA, ESCS.
[7] Bureau of Economic and Business Research, College of Business Administra-
tion, University of Florida. 1980. Florida Statistical Abstract. The Uni-
versity Presses of Florida, Gainesville.
[8] Bureau of Nutritional Sciences and Health Protection Branch, Department of
National Health and Welfare, Canada. 1973. Report on the Relationship
Between Income and Nutrition Based on Analysis of Nutrition, Canada
Data. Ottawa.
[9] Buse, R. C., and L. E. Salathe. 1979. "Household Expenditure in the United
States, 1960-61: The Last Word." Agricultural Economics Staff Paper,
Series No. 168, University of Wisconsin, Madison.
[10] Chanarin, I. 1979. "Folate in Blood, CSE and Tissues." In The Megalobastic
Anemias. Blackwell Science Publications, Oxford.
[11] Davis, C. G., and P. H. Neenan. 1979. "Impact of Food Stamp and Nutrition
Education Programs on Food Group Expenditure and Nutrient Intake of
Low-Income Households." Southern Journal of Agricultural Economics
11(2):121-129.
[12] Federal Register. 1980. Vol. 45, No. 78. U. S. Government Printing Office,
Washington, D.C.
[13] Gallo, Anthony E., and William T. Boehm. 1978. "Food Expenditures by
Income Group." Pages 21-26 in National Food Review. USDA, ESCS,
NFR-3.
[14] Guthrie, H. A. 1975. Introductory Nutrition. 3rd ed. The C. V. Mosby
Company, St. Louis.
[15] Lancaster, K. L. 1966. "A New Approach to Consumer Theory." Journal of
Political Economy 74:132-157.
[16] Lane, Sylvia. 1978. "Food Distribution and Food Stamp Program Effects on
Food Consumption and Nutritional Achievement of Low Income Persons
in Kern County, California." American Journal of Agricultural Econom-
ics 60(1):108-116.
[17] Longen, K., and J. Allen. 1981 "Domestic Food Programs." Page 2 in
National Food Review. USDA, ESCS, NFR-13.
[18] Madden, J. P., and M. Yoder. 1972. Program Evaluation: Food Stamp and
Commodity Distribution in the Rural Areas of Central Pennsylvania.


91








Bulletin 780, Agricultural Experiment Station, The Pennsylvania State
University, University Park.
[19] Madden, J. P., Jane Goodman, and Helen Guthrie. 1976. "Validity of the
24-hour Recall." Journal of the American Dietetic Association 68(1):143-
147.
[20] National Center for Health Statistics. 1979. Food Consumption Profiles of
White and Black Persons Aged 1-74 Years. Vital and Health Statistics
Series 11-No. 210. DHEW Publication No. (PHS) 79-1658.
[21] Neenan, P. H., and C. G. Davis. 1977. "Impacts of the Food Stamp Program
on Low-Income Household Food Consumption in Rural Florida." South-
ern Journal of Agricultural Economics 9(2):89-97.
[22] Phlips, L. 1974. Applied Consumption Analysis. North Holland Publishing
Co., Amsterdam-Oxford.
[23] Prais, S. J., and H. S. Houthakker. 1955. The Analysis of Family Budgets.
Cambridge University Press, Cambridge.
[24] Prochaska, Fred J., and R. A. Schrimpet. 1973. "Opportunity Cost of Time
and Other Socioeconomic Effects of Away-From-Home Food Consump-
tion." American Journal of Agricultural Economics 55(4):595-603.
[25] Sauberlich, H. E., R. P. Dowdy, and J. H. Skala. 1979. Laboratory Test for
the Assessment of Nutritional Status. 3rd ed. CRC Press, Inc., Cleveland.
Pages 59, 60.
[26) Scearce, W. Keith, and Robert B. Jensen. 1979. "Food Stamp Program
Effects on Availability of Food Nutrients for Low-Income Families in the
Southern Region of the U. S." Southern Journal of Agricultural Econom-
ics 11(2):113-120.
[27] Salathe, Larry E., and Reuben C. Buse. 1979. Household Food Consumption
Patterns in the United States. Technical Bulletin No. 1587, USDA, ESCS,
National Economic Analysis Division.
[28] Schuh, G. Edward. 1977. "Neoclassical Economic Theory, Poverty and-
Income Distribution." Pages 87-110 in Rural Poverty and the Policy
Crisis, edited by R. O. Coppedge and C. G. Davis. Iowa State University
Press, Ames.
[29] Smallwood, D., and J. Blaylock. 1981. Impact of Household Size and Income
on Food Spending Patterns. Technical Bulletin 1650, USDA, ESCS.
[30] Strain, W. H., W. J. Poriers, A. Flynn, and O. A. Hill. 1972. "Trace Element
Nutriture and Metabolism through Hair Analysis." Page 383 in Trace
Substances in Environmental Health-V, edited by D. D. Hemphill. Uni-
versity of Missouri, Columbia.
[31] U. S. Department of Agriculture. 1978. Agricultural-Food Policy Review.
ESCS. ARPR-2
[32] U. S. Department of Health, Education, and Welfare, NCHS. 1974. Pre-
liminary Findings of the First Health and Nutritional Examination Survey,
United States 1971-1972. DHEW Publication No. (HRS) 74-1279-1.
[33] U.S. Department of Health, Education, and Welfare, Health Service Mental
Health Administration. 1972. Ten State Nutrition Survey, 1968-70.
DHEW Publication No. (HSM) 72-8134, Atlanta, Georgia.
[34] Wagner, P. A., L. B. Bailey, G. J. Christakis, C. G. Davis, H. Appledorf,
and J. S. Dinning. 1981. "Zinc Status of Adolescents and Elderly from
Low Income Households." Paper presented at XII International Con-
gress of Nutrition, San Diego, California.
[35] West, Donald, and David W. Price. 1976. "The Effects of Income, Assets,
Food Programs and Household Size on Food Consumption." American
Journal of Agricultural Economics 58(4):725-730.

92





University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs