Citation
A comparison of rates of social interaction between aged widowed and aged married individuals /

Material Information

Title:
A comparison of rates of social interaction between aged widowed and aged married individuals /
Creator:
Petrowsky, Marc Lee, 1948-
Publication Date:
Copyright Date:
1975
Language:
English
Physical Description:
vii, 119 leaves : ; 28 cm.

Subjects

Subjects / Keywords:
Economic research ( jstor )
Educational research ( jstor )
Educational resources ( jstor )
Formal organization ( jstor )
Married status ( jstor )
Older adults ( jstor )
Religious organizations ( jstor )
Sex education ( jstor )
Social interaction ( jstor )
Widowers ( jstor )
Dissertations, Academic -- Sociology -- UF ( lcsh )
Married people ( lcsh )
Older people ( lcsh )
Sociology thesis Ph. D ( lcsh )
Widowers ( lcsh )
Widows ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis--University of Florida.
Bibliography:
Bibliography: leaves 112-118.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Marc Petrowsky.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
026031938 ( AlephBibNum )
03652911 ( OCLC )
AAW1800 ( NOTIS )

Downloads

This item has the following downloads:


Full Text















A COMPARISON OF RATES OF SOCIAL INTERACTION
BETWEEN AGED WIDOWED AND AGED MARRIED INDIVIDUALS












BY

MARC PETROWSKY


A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA IN PARTIAL
FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY


UNIVERSITY OF FLORIDA
1975





































To Iris
















ACKNOWLEDT.nMENTS


This study represents the culmination of my formal

sociology training. Therefore it seems only fitting that

I should recognize all individuals who played a role in

this education. To this end, I would like to first thank

my parents for providing me the incentive and support neces-

sary to further my college education at both the graduate

and undergraduate levels. Furthermore, I want to recognize

the instrumental role Dr. Richard Larson played in my early

days as a graduate student when he served as both counselor

and friend. Finally, I extend deep appreciation to Drs.

Agresti, Gorman, Leslie, Warheit and especially Dr. Berardo,

my chairman, for the intellectual stimulation and guidance

they provided throughout my doctoral program. Their assis-

tance in helping me conceptualize and implement this pro-

ject was crucial and more than generous, but I thank them

even more for trying to cultivate in me the sociological

imagination. A final note of thanks is in order. Data upon

which this research is based came from a large scale sample

survey entitled Evaluating Southern Mental Health Needs and

Services, MH15900. Dr. Warheit graciously provided me access

to these data.
















TABLE OF CONTENTS


Page

ACKNOWLEDGMENTS. ... . . . . . . . iii

LIST OF TABLES . . .. ..... . . .. v

ABSTRACT . . . . . . . .. . . vi

CHAPTER I INTRODUCTION . ... . . . . 1

CHAPTER II REVIEW OF LITERATURE . . ... .10

CHAPTER III METHODOLOGY . . . . . ... 31

CHAPTER IV FINDINGS . . . ...... . . . 58

CHAPTER V SUMMARY AND CONCLUSIONS . . .. 88

BlBLIOGRAPHY . . . . . . . . . . 112

BIOGRAPHICAL SKETCH . . . . . ... . .119















LIST OF TABLES


Page

TABLE 1. Frequency and Percent of Respondents by
Levels of Marital Status, Sex, Age,
Race, Education, Health Status, and Income 53

TABLE 2. Comparisons Between Married and Widowed
Individuals in Terms of Frequency of
Interaction in Nearby Kin Groups, Far
Away Kin Groups, Friend Groups, Religious
Organizations, and Formal Organizations . 61

TABLE 3. Measure of Association Between, and
Frequency and Percent of Married and
Widowed Respondents Engaging in Interaction
with Nearby Kin by Sex, Age, Race, Education,
Health Status, and Income ..... ... 67

TABLE 4. Measure of Association Between, and
Frequency and Percent of Married and
Widowed Respondents Engaging in Interaction
with Far Away Kin by Sex, Age, Race,
Education, Health Status, and Income .. 71

TABLE 5. Measure of Association Between, and
Frequency and Percent of Married and
Widowed Respondents Engaging in Interaction
with Friends by Sex, Age, Race, Education,
Health Status, and Income .. . . . .76

TABLE 6. Measure of Association Between, and
Frequency and Percent of Married and
Widowed Respondents Engaging in Interaction
in Religious Organizations by Sex, Age,
Race, Education, Health Status, and Income .. 80

TABLE 7. Measure of Association Between, and
Frequency and Percent of Married and
Widowed Respondents Engaging in Interaction
in Formal Organizations by Sex, Age, Race,
Education, Health Status, and Income .. .. 84













Abstract of Dissertation Presented to the Graduate Council
in Partial Fulfillment of the Requirements for the Degree
of Doctor of Philosophy


A COMPARISON OF RATES OF SOCIAL INTERACTION
BETWEEN AGED WIDOWED AND AGED MARRIED INDIVIDUALS


By

Marc Petrowsky

June, 1975

Chairman: Dr. Felix M. Berardo
Major Department: Sociology


This research examines the social interaction of the

aged from the perspective of the exchange conceptual frame-

work. Interaction is measured in terms of frequency of

participation in kin groups, friend groups, formal organiza-

tions, and religious organizations. Rates of social inter-

action are compared between aged widowed and aged married

individuals in each of the areas of social interaction in-

vestigated both before and after controlling for resources

which are considered, from the exchange conceptual framework,

to be necessary to maintain social interaction. Resources

examined in this research include age, sex, race, education,

income, and health status.

A systematic sample was drawn from a sample frame

developed from electrical utility company residential house-

hold listings in Alachua County, Florida. Structured inter-

views yielded data on 128 aged widowed and 145 aged married

individuals sixty years old or older.

vi









The findings indicate no significant differences in

rates of interaction in kin and friend groups between

aged widowed and aged married persons. Additionally, the

aged widowed interact in religious and formal organiza-

tions more than aged married individuals. Controls for

resources provide virtually no support for an interpre-

tation of social interaction of the aged from the exchange

conceptual framework.
















CHAPTER I

INTRODUCTION


This research examines the social interaction of the

aged in our society. Interaction is measured in terms of

frequency of participation in kin groups, friend groups,

formal organizations, and religious organizations, a sub-

group of formal organizations but one which has been shown

to be important in assessing the interaction of the aged

(Berardo, 1967; and Townsend, 1957). Rates of social inter-

action are compared between aged widowed and aged married

individuals. The interaction rates are subsequently com-

pared again; however the second comparisons are made con-

trolling for resources which are considered, from the ex-

change conceptual framework, to be both relevant and necessary

to maintain social interaction. Resources consist of both

material and nonmaterial goods (Homans, 1958:597) or activ-

ities and sentiments (Homans, 1961:34) which are considered

to be valued exchangeable commodities. For this study resources

include education, income, and health status. Additionally,

age, sex, and race are also controlled because they are as-

sumed to be indirectly indicative of valued resources.










Justification for the Research

The area of investigation was chosen primarily for

two reasons. First, the number of the aged is rapidly

increasing both in absolute numbers and as a proportion of

the entire population of American society. Because of this

growth the aged as a group, and the widowed as an important

sub-category of that group, have become an important area

for sociological investigation (Berardo, 1968).

Second, research of the aged is extremely limited.

This has two interrelated and unfortunate consequences.

First, pure scientific knowledge of the area is limited and

relatively unorganized. This inadequacy, representing a gap

in knowledge of later stages of the family life-cycle, is

regrettable in and of itself. Second, because of the insuf-

ficiencies of scientific knowledge in this area there is a

pitifully inadequate pool of research findings available

which can be drawn on to guide private or governmental agencies

which might attempt to promote legislation, provide funds,

or in other ways attempt to alleviate the unfortunate posi-

tions in which many of the aged find themselves. By orga-

nizing research findings extant in this area and analyzing

available data on aged widowed and married individuals, this

research will hopefully contribute to the literature in the

area. Further elaboration of each of these reasons follows.

For 150 years the population of the United States was

an aging population, as measured in terms of median age,

until the rising birth rates of the post-World War IT decades

reversed this trend. This reversal notwithstanding, the









absolute number of the aged as well as the proportion of

the aged in our society has increased steadily if not

rapidly. There were just over 3.1 million aged persons

in the United States at the turn of the century, by 1950

there were 12.3 million and the 1970 census revealed over

twenty million citizens 65 or older (United States Census,

1970).

The proportionate growth of the aged population paral-

lels their increase in absolute numbers. In 1900 4.1 per-

cent of the population was 65 years of age or older. In

fifty years the proportion of the aged in our society had

climbed to 8.1 percent and in 1970 the elderly comprised

almost 10 percent of the total population (Statistical Abstract

of the United States, 1972:30).

This increase in the number of the aged has come about

through a complex variety of factors such as the "number of

births in appropriate earlier periods, declining mortality,

and immigration" (Sheldon, 1960:40). These factors, as

dependent variables, resulted from the matrix of advancing

science and technologism which grew at an unprecedented rate

from the turn of the century until the present.

Moreover, among the aged, widowhood, according to

Berardo (1968:191), is "rapidly becoming a major phenomenon

of American society." In 1970 the aged population included

six million widows and 1.5 million widowers, approximately

40 percent of the total aged population (McKain, 1972:61).

These figures reveal that, among the aged widowed, there










are over four times as many women as men, and an analysis

of the trends presented by census data for the past sixty

years suggests a continued increase in the gap between

widows and widowers. For example, in 1910 there were 1.47

million widowers who comprised 4.4 percent of the entire

population. Forty years later there were 2.3 million

widowers who constituted 4.2 percent of the population.

By 1971 the number of widowers has decreased to two million

representing only 3.1 percent of the population. The number

of widows, in contrast, grew steadily from 3.18 million in

1910 to 9.78 million in 1971. They also registered a steady

proportionate increase from 10.3 percent in 1910 to 13.8

percent in 1971 (Statistical Abstract of the United States,

1972:30). The main factors contributing to the excess of

widows over widowers include the higher mortality rates

among males, the higher remarriage rates for men, and the

fact that men usually marry women younger than themselves

(Jacobson, 1959:25-27).

The second justification for this study is founded upon

the paucity of research in an area which has pressing need

for scientific investigation to both illuminate the sub-

stantive area and provide findings which can be used to

direct social action in attempts to alleviate some of the

disadvantaged conditions confronting many of the aged

widowed. For example, research suggests that the widowed

are likely to be economically impoverished (Lopata, 1969;

Marris, 1958; Statistical Abstract of the United States, 1969).










have higher death rates than married individuals (Cove, 1972;

and Townsend, 1957), exhibit higher suicide rates than

married individuals (Berardo, 1967; Bock, 1972; and Gove,

1972), have a higher rate of mental disorders than married

individuals (Berardo, 1967), experience deterioration of

health status (Berardo, 1967), and indicate that they arc

lonely (Berardo, 1967; and Lopata, 1969).

These findings indicate that widowhood, although in-

evitable for a significant proportion of the entire popu-

lation, offers something less than an enviable status for

many of the aged in our society. This study will hopefully

shed some light on the subject matter at hand by systemat-

ically organizing research findings pertaining to the aged

widowed and married individuals in addition to examining

statistical data on 273 aged widowed and married persons.


History of Scientific Interest in Aging

Systematic studies of the aged began only recently

and were first initiated by biologists who were interested

in time-related changes of cells, tissues, and physiological

mechanisms. Interest in the biological aspects of aging in

the United States was signaled by the establishment of the

American Research Club on Aging in 1939. In 1945 the

Gerontological Society, Inc., was founded. Both provided

funds as well as a focal point for generating interest which

further stimulated research in the area.


IThis section draws heavily from Tibbitts (1960).










Psychological research on aging followed a similar

pattern. In the early 1930's W. R. Miles conducted the

Stanford Later Maturity Research Project which was the

first systematic psychological study of aging. The further-

ance of psychological research on aging mushroomed after the

American Psychological Association gave formal recognition

to research in this area in 1946.

Research on the social and economic aspects of aging

began in the 1920's and was primarily concerned with pro-

viding solutions to practical problems facing the aged.

This interest was precipitated primarily by three factors:

increased visibility of the aged, separation of the elderly

from the workforce, and the American value emphasizing in-

dividual happiness and well-being (Pollak, 1948).

The increased visibility of the aged resulted from

technologism and industrialization which had a tremendous

impact on fertility, mortality, and migration rates. One

of the consequences was the doubling in size of the aged

population from 1900 to 1930 and again before mid-century

(Tibbitts, 1960). Additionally, with the economic emphasis

shifting from agrarianism to industrialism, the work and

family roles of the elderly underwent dramatic change.

Their usefulness as repositories of traditional farming

wisdom and as owners of the land was undermined by the

industrial revolution which created a viable alternative to

a land-based economy. The end result was the generation of









several social problems related to the financial, social,

and physical adaptation of the aged to an industrialized,

urbanized milieu. Research in the 1930's and 1940's

documented the adjustment problems of older citizens. In

1943, under the leadership of Ernest W. Burgess, The

Committee on Social Adjustment in Old Age was established.

During the decades of the 1940's and 1950's various univer-

sities followed the lead of the University of Chicago and

the University of Michigan in stimulating research and

interest in the social ramifications of maturation and old

age.

In 1950 a significant portion of the program of the

National Conference on Aging was devoted to social and

economic aspects of aging. The Gerontological Society,

in 1952, distinguished a separate Division of Psychology

and Social Science. In 1956 the Inter-University Training

Institute in Social Gerontology was established and supported

by the National Institutes of Health. Additionally, the

International Association of Gerontology established a

social science research division in that year.

During this same period family sociologists, following

the path paved earlier by Sorokin, Zimmerman, and Galpin

(1931), were also drawing more attention to the aged in

family settings by focusing on stages in the family life

cycle (Duvall, 1957; Click, 1957; Rodgers, 1964; and Rodgers

and Hill, 1964). It is hoped that this research will con-

tribute to the rapidly expanding knowledge in this area.









Rationale of the Present Study

This study is organized into two interrelated but

analytically distinct parts. The initial portion of the

work integrates findings gleaned from research pertaining

to isolation of the aged widowed and married individuals.

The mechanism for integrating, as presented in Zetterberg

(1965), is an axiomatic approach to theory building.

Research findings are organized into empirical or ordinary

propositions. Empirical propositions with similar linkages

are combined and abstracted to higher levels of explanation.

These then become derived propositions which are expanded to

form theoretical propositions which have high information

value. This task of theory construction is conducted with-

in the exchange conceptual framework, thus linking the theory

to a more general theoretical model.

The second portion of the study provides for an empirical

test of the constructed theory. Hypotheses compare differ-

ences between the interaction rates of the aged widowed and

the aged married in kinship groups, friend groups, formal

organizations, and religious organizations controlling for

age, sex, marital status, education, income and health status.

The research examines the possibility of social interaction

among the aged being contingent upon control over valued

resources.

This study is organized into four additional chapters.

The second chapter reviews the literature in the area, both

the theoretical literature as well as research findings. The







9


third chapter presents the hypotheses and research design.

The fourth chapter discusses the research findings,while

the summary and conclusions from the study are presented

in the final chapter.















CHAPTER IT

REVIEW OF LITERATURE


The review of literature for this study is divided into

two sections. Literature relevant to the exchange conceptual

framework is presented in the first part. A survey and organ-

ization of research findings pertaining to the social isola-

tion of the aged married and the aged widowed follows.


Review of the Theoretical Literature

This section on the theoretical literature accomplishes

two purposes. It first discusses the history and applica-

bility of theoretical conceptual frameworks. Secondly, it

provides an analysis of the exchange conceptual framework.


Conceptual Framework

Zetterberg (1965) suggests that sociology contains

humanistic as well as scientistic underpinnings. The scien-

tific tradition encompasses two types of what is commonly

labeled theory. The first type, taxonomies, consists of

descriptive statements which do not offer any explanations.

The second type, referred to by Zetterberg as theoretical

sociology, consists of systematically interrelated proposi-

tions which are derived from and inspire research.

The latter provides the bulwark of the axiomatic approach

to theory building, the approach utilized in this research.

10










The former is identified by Hill and Hansen (1960) and Nye

and Berardo (1966) as a major ingredient of conceptual

frameworks.

Hill and Hansen (1960), noting the paucity of family

theory, challenged family sociologists to interrelate re-

search findings into sets of research propositions guided

by what Zetterberg referred to as taxonomies. These taxon-

omies, or frames of reference, play a pivotal role in build-

ing sociological theory, posit Hill and Hansen, by providing

researchers with groups of "interrelated but not necessarily

intordefined concepts generally applicable to the arena of

marriage and the family" (Hill and Hansen, 1960:300). Addi-

tionally, conceptual frameworks, for the purposes of this

research, are considered to consist of time and space dimen-

sions (Hill and Hansen, 1960) as well as specification of

assumptions underlying the framework (Nye and Berardo, 1966).

The Hill and Hansen article was followed by several

articles which suggested a multiplicity of new and existing



IDuring the course of the presentation of this research
any reference to the word "theory" is explicitly assumed to
mean conceptual framework unless otherwise indicated. Although
the two concepts are used interchangeably, an important ana-
lytical distinction can easily be made between the two. A
theory is a set of interrelated propositions which purports
to explain some phenomenon. A comparison of this definition
with the components of conceptual frameworks articulated
above reveals that theories can be understood to be much
more specific explanations of the empirical world while con-
ceptual frameworks provide broad guidelines within which
theories are developed.










frameworks to guide family research. In 1969 Edwards sug-

gested that the exchange conceptual framework would provide

a powerful theoretical model for analyzing family behavior.


Exchange Conceptual Framework

Exchange theory is the conceptual framework used in

this research to guide the interpretation of the analysis

of rates of interaction of aged widowed and married individ-

uals. Review of the theoretical literature pertaining to

the exchange framework follows.

Although it has been most succinctly articulated only

recently, many of the underlying assumptions and concepts

of exchange theory can be found in the classical literature.

Selected works from Marx, Comte, and Simmel reveal attempts

by these early masters to utilize, either directly or in-

directly, aspects of exchange theory to explain social pro-

cesses. Simmel, perhaps one of the most forthright pro-

ponents of exchange as a characteristic of human action,

wrote that "exchange is the objectification of human inter-

action" (Simmel, 1950:.388).

More recently aspects of the exchange framework have

been applied by social psychologists, social philosophers,

and especially anthropologists, to guide research activity

and to offer explanations of various facets of social phenom

ena. Anthropologists including Malinowski (1932), Mauss

(1954), and Thurnwald (1932) have all documented the process

of exchange in their observations of preliterate tribes and

societies. Additionally, the importance of the norm of










reciprocity, an integral concept in exchange theory, has

been portrayed by many contemporary social theorists such

as Becker (1956), Gouldner (1960), and Levi-Strauss (1957).

Perhaps the widest application of exchange theory,

albeit circuitous in nature, has been by sociologists in

family research. Dating and courtship behavior (McCall,

1966; Rogers and Havens, 1960; and Waller, 1937) as well

as the area of marital power (Blood and Hamblin, 1958;

Blood and Wolfe, 1960; Heer, 1958; and lleer, 1963) has been

approached from this perspective.

The preceding theorists and researchers have dealt

with exchange theory in primarily an unsystematic fashion.

However, during a five-year period beginning in 1959, three

monographs were published which fully set forth the concepts,

underlying assumptions, and in one case, propositions of

exchange theory (Blau, 1964; llomans, 1961; and Thibaut and

Kelley, 1959). The discussion of the exchange conceptual

framework closely follows these three books.

The underpinnings of exchange theory are found in both

behavioral psychology and elementary economics. Combined,

central ideas from these two disciplines yield a theory

which attempts to analyze social interaction as occurring

through a process of exchange of resources by actors occupy-

ing positions in the social structure (Homans, 1958:597).

Two assumptions provide the foundation for this exchange

process. First, behavior is goal oriented and achieved

through interaction with others. Second, actors, in pur-

suit of various goals, attempt to adapt means to further










the achievement of these ends (Blau, 1964:5). Consequences

of this goal oriented behavior include rewards as well as

costs.


Rewards

Rewards, or resources as they are sometimes called,

consist of both material and nonmaterial goods (Homans,

1958:597) or activities and sentiments (Homans, 1961:34).

There are two basic types of rewards, intrinsic and extrin-

sic. Intrinsic rewards result from types of social inter-

action where the social relationship itself is considered

to be a reward. Social situations which yield extrinsic

rewards for actors are those in which the social relation-

ship is a mechanism through which these rewards are realized

(Blau, 1964:58-59). The more highly valued a reward, the

more often individuals will extend effort to obtain that

resource. The value of resources, be they objective (activi

ties or material goods) or subjective (sentiments or non-

material goods) in nature, is determined by four variables:

group values, unique individual experiences, deprivation,

and satiation.

Values are a social product of group existence. There-

fore, individuals who belong to the same group or the same

kinds of groups are more likely to value the same resources

or types of resources (Homans, 1961:46). However, this

similarity of group experience is circumscribed by unique

individual experiences which are likely to channel the










development of values so that they do not always exactly

reflect other group members' values.

Additionally, the individual's state of deprivation

and satiation affects the value of resources (Homans,

1961:47-49). Resources highly valued but in short supply

become even more highly valued while resources constantly

supplied decline in value. The variability and subjectivity

of the determination of values leads to problems of measure-

ment and analysis. Homans suggests that "our knowledge

of values will always be imperfect, and the predictions

we make from it will be gross and statistical, based on a

few obvious similarities and differences, bound to go wrong

in detail" (1961:46-47).


Costs

Costs refer to "[actors that operate to inhibit or

deter the performance of a sequence of behavior" (Thibaut

and Kelley, 1959:12). Additionally, lomans maintains that

the concept of cost, or aversive conditions, should include

recognition of alternative rewards forgone by investing

resources in pursuit of one particular goal rather than a

potential alternative ([omans, 1961:58).


Profit

Profit is defined as "the difference between the value

of the reward a man gets by emitting a particular unit-activ

ity, forgone in emitting the first" (lomans, 1961:63). It

follows that the more valuable the reward stemming from a









particular activity, the more often an actor will emit it,

while the more costly it proves to be, the less he will

emit it. The preceding reduces to the simple formula of

rewards minus costs equals profit. It is assumed that

continued exchange is contingent upon both actors realizing

a profit. Indeed, "the open secret of human exchange is to

give the other man behavior that is more valuable to him

than it is costly to you and to get from him behavior that

is more valuable to you than it is costly to him" (Homans,

1961:62).

The actor's perception of his profit is important be-

cause individuals will typically terminate social relation-

ships which do not appear profitable. Thibaut and Kelley

have articulated two standards upon which these outcomes

of interaction are evaluated. The comparison level is

"the standard against which the member evaluates the attrac-

tiveness of the relationship or how satisfactory it is."

The second, the comparison level of alternatives, refers to

"the standard the member uses in deciding whether to remain

in or to leave the relationship" (Thibaut and Kelley, 1959:21).

The former is important because it determines for actors

whether profits gleaned from a relationship are worth the

costs. If an actor feels the relationship is unrewarding

he has the option of seeking alternative relationships or dis-

solving the relationship and going without the association

with its concomitant rewards and costs. The comparison

level of alternatives is important because profits are not










only evaluated in terms of rewards minus costs but "common

norms develop in societies that stipulate fair rates of

exchange between these benefits" (Blau, 1964:155). Homans's

concept of "distributive justice" also expresses this

notion.


Propositions

The preceding analysis of the concepts and assumptions

underlying the exchange framework have been succinctly or-

ganized by Homans (1961:53-55) into the four following pro-

positions.


Proposition 1 If in the past the occurrence of a
particular stimulus situation has
been the occasion on which a man's
activity has been rewarded, then the
more similar the present stimulus
situation is to the past one, the
more likely he is to emit the activity,
or some similar activity, now.

Proposition 2 The more often within a given period
of time a man's activity rewards the
activity of another, the more often
the other will emit the activity.

Proposition 3 The more valuable to a man a unit of
the activity another gives him, the
more often he will emit activity re-
warded by the activity of the other.

Proposition 4 The more often a man has in the recent
past received a rewarding activity from
another, the less valuable any further
unit of that activity becomes to him.

Exchange theory makes no assumptions of explicit

rational behavior. It is only concerned with the immediate

alternatives open to the actor, as he sees them, and his

resulting choice. As Blau suggests "the only assumption

made is that human beings choose between alternative potential









associates or courses of action by evaluating the experiences

or expected experiences with each in terms of a preference

ranking and then selecting the best alternative" (1964:18).

A totally rational situation, as found in game theory,

does not exist because individuals do not have complete in-

formation on all conceivable alternatives, their preferences

do not remain constant, and the actors do not pursue only

one goal, to the exclusion of all others (Blau, 1964:18).

The foregoing analysis examines the general concepts,

assumptions, and propositions underlying exchange theory.

Edwards (1969) has suggested some minor modifications in

this approach to make it more amenable to the special con-

ditions typical of interaction among family members. Be-

cause this research focuses specifically on interaction of

aged widowed and married individuals these modifications

are given consideration.

Edwards states that in an analysis of family inter-

action it is difficult to conceptualize what is exchanged.

This problem stems in part from the informality of family

associations plus the isolation of familial interaction,

both of which tend to obfuscate the resources exchanged

(Edwards, 1969:520). Additionally, exchange in family

situations can be based upon either intrinsic or extrinsic

rewards. This research does not attempt to measure intrinsic

rewards. Moreover, it is difficult to fully specify exchange

or alternatives for extrinsic rewards. However, with all

its problems of conceptualization and measurement, exchange









theory is assumed to be a particularly valuable approach

for explaining behavioral outcomes of the aged widowed.


Review of the Research Literature

The review of the research literature is divided into

three sections. The first part presents a series of empir-

ical generalizations abstracted from research relevant to

the social isolation of aged widowed and aged married

individuals. The second section presents derived proposi-

tions obtained by abstracting or combining empirical pro-

positions. The final part presents a theoretical proposi-

tion which is acquired by abstracting and combining derived

propositions to form a more powerful information statement.


Empirical Propositions

The first step in relating variates, according to

Zetterberg's axiomatic approach to theory building, consists

of the formulation of empirical propositions. These proposi-

tions depict relationships existing between two or more

variables. The variables are designated as either deter-

minants or results. Determinant or independent variables

cause results or dependent variables. Within each proposi-

tion the determinants are related to the results by a variety

of linkages.

Zetterherg (1965) posits that the linkages in a causal

relationship can be described in terms of five pairs of

descriptive attributes. A relationship may be either rever-

sible (if X, then Y; and if Y, then X) or irreversible (if










X, then Y; but if Y, then no conclusion about X), deter-

ministic (if X, then always Y) or stochastic (if X, then

probably Y), sequential (if X, then later Y) or coextensive

(if X, then also Y), sufficient (if X, then Y, regardless

of anything else) or contingent (if X, then Y, but only if

Z), and necessary (if X, and only if X, then Y) or sub-

stitutable (if X, then Y; but if Z, then also Y). Later

in the research empirical propositions with similar linkages

are combined and further abstracted to yield derived pro-

positions.

The following empirical propositions were abstracted

from research relevant to the social interaction of aged

widowed and aged married individuals.


(1) The widowed, particularly males, are less
likely to interact in formal organizations
than are married individuals.

Berardo (1967), Bock (1972), and Bock and Webber (1972)

all posit that, in terms of participation in formal organi-

zations, widowers are dramatically less likely to interact

than either widows or married individuals while the dif-

ferences between the latter two are minimal. Furthermore,

Pihlblad and Adams (1972) depict no differences in parti-

cipation in formal organizations between married and widowed

women but noticeable differences between women and men,

especially widowed men. However, considered collectively,

the widowed according to Harvey and Bahr (1974) are less

likely to interact in formal organizations than aged married









individuals. Moreover, Harvey and Bahr proceed to sug-

gest that differences in interaction between widowed and

married aged individuals are a function of age and income

variables and not widowhood status.


(2) Widows are more likely to interact in
religious organizations than are either
widowers or married individuals.

Berardo (1967) and Townsend (1957) support this pro-

position. Pihlblad and Adams (1972) suggest that widowhood

decreases participation in religious organizations for men

but not for women. Berardo states that the discrepancy

between widows and widowers results from the fact that

women have traditionally maintained the liaison between the

family and the religious institution. Moreover, this role

is not significantly diminished with the death of the

husband. Additionally, widows can interact in religious

institutions on the same level as married individuals be-

cause they are not discriminated against because of their

single status.


(3) The widowed are less likely to interact with
friends than are married individuals.

Berardo (1967) reveals that the widowed are more iso-

lated in terms of friendship interaction than married indivi-

duals. Moreover, Booth (1972) holds that married individuals

express more close friend relationships than the widowed.

Furthermore, they both note that the discrepancy between

aged widowers and aged married individuals, in terms of

absolute number as well as in frequency of interaction with










friends, is particularly great. The impact of sex in

circumscribing friend interaction is provided additional

support by Pihlblad and Adams (1972) who suggest that

widowhood decreases friend interaction for men but not

women. However, Townsend (1957) indicates that widows and

widowers visit their neighbors more than married individuals.

Because visiting neighbors can be construed as an index

of Friendship it would appear that Townsend and Berardo sug-

gest conflicting findings. However, this discrepancy might

be accounted for by the finding of Berardo (1967) that

"widowhood creates greater strains for rural residents."

To elaborate, Berardo's study focuses on aged people

in a rural county in Washington whereas Townsend's study

focuses on 200 aged people from Bethnal Green, London. Blau

(1961) suggests that the status of isolation depends on the

structural context of older people. That is to say, if an

individual's marital status is married and all of his peers

are widowed then he will perceive himself as isolated because

he is deviant vis-a-vis the marital status structure of the

community. To return to the original problem, the widowed

in Townsend's study might be less isolated than married in-

dividuals because they live in an urban area with greater

potential interaction with those of the same marital status.

On the other hand, the married individuals in Berardo's study

could be less isolated because of the diminished potential

for interaction with others of similar status that exists

for the widowed in rural areas.









(4) The widowed are less likely to interact
with their children than are married
individuals.

Berardo (1967) and Townsend (1957) support this find-

ing. Interaction subsumes visits (Townsend), companionship,

transportation, advice, money, and extension of gifts

(Berardo). All these indices of interaction have been used

by sociologists to examine kinship interaction (Sussman and

Burchinal, 1962a). Although Berardo posits that widows

have high rates of contact with their children he proceeds

to explain this in terms of higher incidences of widows

living with their children. It is assumed that the opportunity

to live with one's children is diminished for married individ-

u:ils, thereby distorting the compared rates of interaction

between aged survivors and married individuals. Therefore,

the widowed, considered collectively, have less interaction

with their children than married individuals have with theirs.

However, findings from Adams' (1968) middle class sample

yield some contradictory evidence. He determined that widows

sec their daughters more frequently than married women. This

contradiction suggests that social class may play an important

role in social interaction of the aged. This relationship

is examined in Empirical Proposition 12.


(5) The widowed are less likely to interact with
kin than are married individuals.

Berardo (1967), Bock and Webber (1972), and Townsend

(1957) suggest that kin interaction declines with the death

of a spouse. This relationship is also demonstrated for









elderly men but not elderly women in research by Pihlblad

and Adams (1972).


(6) The widowed are more likely to commit
suicide than are married individuals.

Berardo (1968), Bock (1972), Bock and Webber (1972),

Dublin (1963), and Durkheim (1951) all document the in-

creased likelihood of suicide among the widowed.


(7) The widowed are more likely to be isolated
than married individuals.

Research findings from Berardo (1967), Bock and Webber

(1972), Harvey (1973), and Townsend (1957) all support this

proposition.


(8) For widows, the higher the educational
status, the greater the participation in
formal organizations.

Lopata (1973) suggests this proposition.


(9) For widows, the higher the financial status,
the greater the participation in formal
organizations.

Lopata (1973) supports this relationship. Moreover,

Harvey and Bahr (1974) suggest that organizational affilia-

tion for the widowed is circumscribed by socioeconomic

status more than widowhood status.


(L0) For widows, the higher the financial status,
the greater the friendship satisfaction.

Lopata (1973) indicates that widows who develop satis-

factory friendship relationships have higher education and a

comfortable income. Blau (1969) and Griffiths et al.(1971) lend









further support to this proposition by suggesting a posi-

tive relationship between socioeconomic status and friend-

ship interaction among the aged. Additionally, Rosow (1967)

suggests that the middle class elderly indicate a dispro-

portionate number of friends when compared with elderly

working class individuals.


(11) For widows, a direct relationship exists
between educational status and friendship
satis faction.

Again Lopata (1973) states this proposition.


(12) The greater the socioeconomic status of
the aged, the less likely is social
isolation.

Griffiths et al. (1971), Lowenthal (1964), and Townsend

(1957) posit that a disproportionate number of aged isolates

tend to be from lower socioeconomic strata. It should be

noted that this empirical proposition includes, but is not

limited to, the widowed. This is the only proposition to

examine the relationship between two variables for the aged

in general. The inclusion of this proposition is necessary

however, to depict the relationship between isolation and

one aspect of available resources, socioeconomic status.

Later, in forming the theoretical proposition, the role of

tis proposition in establishing the relationship between

resources and isolation will become evident.


Derived Propositions

Derived propositions, which have greater information

value than empirical propositions, can he obtained by










abstracting or combining empirical propositions. Similar

causal linkages are necessary in order to subsume two or

more empirical propositions under a more powerful derived

statement. The preceding empirical propositions were de

rived from research findings. These propositions will now

be combined to yield more general explanatory statements

referred to as derived propositions.


(A) The widowed are less likely to interact
in formal organizations than are married
individuals.

Empirical Propositions I and 2 combine to form

Derived Proposition A. This combination is possible be-

cause the causal linkages of both propositions are irrever-

sible, stochastic, coextensive, contingent, and substitut-

able. The second empirical proposition is subsumed under

the first because churches are considered formal organiza-

tions (l.eslie et al., 1973:307). Moreover, membership

and participation constitute indices of interaction in these

organizations.

Although it appears that the two contributing proposi-

tions are inconsistent, it must be noted that churches, re-

presenting only one of the many types of formal organiza-

tions, provide a social milieu wherein the widowed are not

discriminated against because of their single status (Berardo,

1967). Moreover, because of their traditional link with the

church, widows have the opportunity to use it as an arena

in which they can exchange resources for social interaction.










Resources of widows, such as cooking, cleaning, and

sewing abilities, are highly valued in many church activities

like bazaars, covered dish dinners, and sewing circles.

Moreover, it is presumed widows have more time to invest

in church activities than do married individuals (Berardo,

1967).

Because of the aforementioned reasons, it is assumed

that the church, in relation to other formal organizations,

offers unusual opportunities to the aged widow for engaging

in social interaction. However, considering all types of

formal organizations collectively, available evidence

(Berardo, 1967; Bock, 1972; Bock and Webber, 1972; Harvey

and Bahr, 1974; Pihlblad and Adams, 1972; and Townsend,

1957) holds that the widowed as a group are more isolated

than married individuals in these organizations.


(B) The widowed are less likely to interact
with friends than are married individuals.

Empirical Proposition 3, in its present form, yields

Derived Proposition B.


(C) The widowed are less likely to interact
with kin than are married individuals.

Empirical Propositions 4 and 5 combine to yield Derived

Proposition C. This combination is possible because the

causal linkages of both propositions are irreversible,

stochastic, coextensive, and substitutable. Because children

are members of the kin group it is logical to subsume Empirical

Proposition 4 under ,mpirical Proposition 5.









(D) The widowed are more likely to be isolated
than married individuals.

Empirical Propositions 6 and 7 combine to yield Derived

Proposition D. This combination is possible because the

causal linkages of both propositions are irreversible,

stochastic, coextensive, contingent, and substitutable.

Bock and Webber (1972) state that high suicide rates of the

aged widowed are functions of greater social isolation. It

therefore seems logical that suicide rates can serve as an

index of isolation, yielding greater support for the derived

proposition.


(E) For widows, the greater the resources,
the more likely the participation in formal
organizations.

Empirical Propositions 8 and 9 combine to yield Derived

Proposition E. This combination is possible because the

causal linkages of both propositions are irreversible,

stochastic, coextensive, contingent, and substitutable.

Edwards (1969) states that, although it depends on the

social setting, an individual's knowledge is a well defined

resource. For this research it is assumed that knowledge

and education are positively related. Moreover, McCall

(1966) and Streib (1972) hold that money, which is equated

with financial status, also constitutes a resource.


(F) For widows, the greater the resources,
the more likely friendship satisfaction
will result.

Empirical Propositions 10 and 11 combine to yield

Derived Proposition F. This combination is possible because


~










the causal linkages of both propositions are irreversible,

stochastic, coextensive, contingent, and substitutable.

The rationale for dealing with the transformation of

educational and financial status to resources is the same

as in Derived Proposition E.


(G) The greater the resources of the aged,
the less likely is social isolation.

Empirical Proposition 12 is abstracted to yield

Derived Proposition G. Socioeconomic status, based upon

the rationale presented in the two preceding derived pro-

positions, is considered a resource.


Theoretical Propositions

Theoretical propositions are developed by combining

and abstracting derived propositions to form more power-

ful information statements. They represent the final

stage in axiomatic theory construction.


(T) The widowed are more likely to be
socially isolated than married individuals
because isolation is a function of insuf-
ficient resources.

Derived Propositions A, B, C, and D combine to yield

the following statement. The widowed are more likely to

be isolated than married individuals. Derived Propositions

A, B, and C posit that, in terms of interaction in formal

organizations, kin groups, and friend groups, the widowed

are less likely to interact than are married individuals.

These three combined derived propositions can then be









subsumed under Derived Proposition D to yield the first

part of Theoretical Proposition T.

Derived Propositions F, F, and G combine to yield

the following statement. The greater the available

resources, the less likely is social isolation. Derived

Propositions E and F, which pertain to participation in

formal organizations and friendship interaction, are

subsumed under Derived Proposition C to yield the second

part of Theoretical Proposition I.


Conclusion

This concludes the review of all relevant literature.

The hypotheses tested in the present study are generated

from the preceding research, under the guidance of the

exchange conceptual framework. Attention in the third

chapter is focused on the methodology of the research which

specifies the procedure whereby the derived theoretical

proposition is subject to empirical verification.















CHAPTER III

METHODOLOGY


This study examines the social isolation of aged

widowed and aged married individuals. Social isolation

is operationally defined as diminution of social inter-

action. Before comparisons of social interaction rates

between aged widowed and aged married individuals can be

examined, it is necessary to define the terms and concepts

used in the research, to state the various hypotheses, and

to examine their relationship to the previous research.

Data collection procedures are discussed. The statistical

procedures used in describing the data are also presented.

Finally, a summary description of the data is provided.


Concepts

The three concepts which need to be operationally de-

fined are widowhood, social isolation, and social resources.


Widowhood

This study focuses on the social isolation of aged

widowed as compared with aged married individuals. The

aged widowed are defined as those individuals 60 years old

or older who have lost a spouse through death and have not

since remarried. The widowed are Curther distinguished

according to sex. Widows are females 60 years old or older

31










who are not currently married because of the death of a

spouse. Widowers are males 60 years old or older who are

not currently married because of the death of a spouse.


Social Isolation

Studies of social isolation have basically defined it

as attenuation or severance of interpersonal interactions

or relationships (Faris, 1934:155-164; Jaco, 1954:567-577;

Kohn and Clausen, 1955:265-273; Tec and Granick, 1959:226-

232; and Townsend, 1957:169). However, a difficulty arises

when attempting to specify how much attention to social

interaction warrants the application of a label of "social

isolation" to an individual or group (Clausen and Kohn,

1954:140-151).

In the preceding studies scales as well as various

operational definitions of isolation have been used to

demarcate the state of social isolation among individuals.

For this research social isolation refers to diminished

social interaction. To this end, social interaction is

measured in terms of frequency of contact with friends,

kin living nearby, kin living far away, formal organiza-

tions, and religious organizations. Comparisons of rates

of social interaction, in all five areas, between the aged

widowed and aged married individuals, are examined to deter

mine if isolation is a function of diminished control over

social resources. Indices of social interaction include

visits with kin or friends or attendance at group meetings









and functions. Additionally, social interaction with

relatives includes telephone conversations and written

communication.

Social interaction is measured on an ordinal scale.

Therefore, it is assumed that an inverse relationship

exists between rank-ordered categories of social inter-

action and social isolation. A discussion of the ordinal

measures of the five areas of social interaction follows.

Friendship interaction was measured by asking the

respondents if they had any close friends nearby. If so,

they were asked how often they got together with them.

The structured categorical responses included: "all the

time," "often," "sometimes," "seldom," and "never." For

the purposes of this analysis the last two categories were

collapsed,resulting in the following fourfold rank-order

classification of friendship interaction: "all the time,"

"often," "sometimes," and "seldom or never."

There are two measures of kin interaction: one

focuses on interaction with relatives living nearby, the

other assesses the amount of social interaction with rela-

tives living far away. Division of the kin group into one

of these two categories was determined by the respondents'

subjective assessment of which emotionally close kin were

and were not living nearby and far away. If the respondents

indicated that they had any close relatives living either

nearby or far away they were asked how often they got to










see or talk to any of them, excluding relatives living

in the household. Telephone conversations were included

as accepted indices of interaction with all kin regardless

of geographic location while letters were accepted as indi-

cators of interaction only with kin living far away.

Both measures of interaction with relatives had six

fixed answer response-categories: "almost every day,"

"several times a week," "several times a month," "several

times a year," "seldom," and "never." The first two

categories and last two categories were combined for the

analysis to yield the following fourfold rank-order

classification of nearby and far away kin interaction:

"weekly," "monthly," "yearly," and "seldom or never."

Another measure of social interaction was obtained

by asking respondents who were members of formal organiza-

tions how many afternoons and evenings a month they spend

in club activities. The structured response-categories

included: "none," "1-2," "3-5," "6-10," "11-15," and

"over 15." For the analysis of these data the last three

categories are assumed to indicate frequent activity and

are therefore collapsed to form a category called "frequent."

Interaction by respondents participating in clubs three

to five afternoons and evenings a month is labeled "some,"

social interaction one or two afternoons or evenings a

month is called "modest" while no interaction at all is

labeled "none." The preceding process of combining categories










yields a four-fold rank-order classification of social

interaction in formal organizations.

When asked how often they attended the main worship

service of their church or synagogue, respondents claiming

religious affiliation were required to answer from among

the following responses: "every week," "2-3 times a month,"

"once a month," "several times a year," "only once or twice

a year," "less than once a year," and "never." For this

analysis the first response is labeled "weekly," "monthly"

refers to the collapsed second and third categories, "yearly"

includes the responses from the fourth, fifth, and sixth

categories, while the final category is labeled "never."

This procedure results in a four-fold rank-order classifi-

cation of social interaction in religious organizations.

The categories established for the analysis of social

interaction in the five preceding social areas have fre-

quently been constructed by collapsing two or more contiguous

categories. This procedure results in the loss of data.

However, this undesirable consequence is unavoidable because

of the small sample size.


Resources

According to exchange theory, continued social inter-

action is contingent upon the reciprocal exchange of re-

sources. Resources, when expended to consummate social

interaction, become rewards. There are two basic types of

rewards, intrinsic rewards and extrinsic rewards. Intrinsic










rewards result from types of social interaction where the

social relationship itself is considered to be a reward.

Social situations which yield extrinsic rewards for actors

are those in which the social relationship is a mechanism

through which these rewards are realized (Blau, 1964:58-59).

Because of insufficient methodological measuring

techniques this research does not attempt to assess the

impact of the exchange of intrinsic rewards on social

relationships involving aged widowed and aged married individ

uans. This is regrettable because, undoubtably, many forms

of social interaction with and among the elderly depend

upon this type of reward.

However, this research does attempt to determine the

impact of extrinsic rewards upon rates of social interaction

of the elderly. The research design specifies that rates

of social interaction between the aged married and aged

widowed individuals are compared to determine if social

interaction is a function of control over six extrinsic

rewards. Three of the rewards examined include age, sex,

and race, all of which, it is assumed, are more important

as indicators of other resources than as resources them-

selves. For example, Harvey (1973) suggests an inverse

relationship between age and control over resources. The

importance of age as a resource itself is overshadowed by

the impact that age, as an independent variable, has on

control over other resources, health status and income being

two notable examples which are examined in this research.









With regard to the other two resource variables,

Berardo (1967), Bock (1972), Bock and Webber (1972),

Harvey (1973), and Townsend (1957) have all suggested

sex differences among the widowed in terms of rates and

type of social interaction. Additionally, voluminous

research documents the effect of race in circumscribing

social relationships (Simpson and Yinger, 1972:12).

The three remaining variables include education, in

come, and health status. Education was suggested as a

resource by Harvey (1973). Additionally, Edwards (1969)

maintained that knowledge is a resource. Education is

assumed to be positively correlated with knowledge. In

this research the educational resource is measured in

terms of the number of years of education completed for

each of the respondents.

Income, economic resources, or money are mentioned

as viable exchange commodities by Harvey .(1973), McCall

(1966), Streib (1972), and Thibaut and Kelley (1959).

This resource is measured in this study by determining

the income per year per household for each of the respon

dents. Health status is also considered a resource by

Streib (1972), and Thibaut and Kelley (1959). In the

present study health status was determined by asking the

respondents if they had any present physical or health

problems.









Hypotheses

The following ten hypotheses are engineered to examine

the theoretical proposition developed inductively in the

preceding chapter. The first five are designed to test if

significant differences exist between interaction rates of

aged widowed and aged married individuals. Hypotheses 6

through 10 are presented to explore the applicability of

the exchange conceptual framework in providing explanations

for these differences. Unfortunately, direct tests of the

last five hypotheses are impossible. This occurs because

comparisons between the measure of association between

married and widowed individuals in the total sample and the

measure of association between married and widowed individ-

uals for each of the six control variables would involve

tests of statistical significance based on dependent samples.

For this reason no statistical inferences to any conceptual

population are possible. Therefore Hypotheses 6 through 10

will serve only to guide the research effort. Null hypotheses

are unwarranted when no direct inferences to a population

are desired and therefore are not presented for the last

five hypotheses. The subsequent analysis of each proposition

depicts the relationship between the hypothesis and the pre-

vious literature.


Research Hypothesis 1

The aged widowed respondents will interact less fre-

quently with nearby kin than aged married individuals.










Null Hypothesis 1

There will be no difference between aged widowed and

aged married individuals in terms of the frequency of

interaction with nearby kin.

Reviewed literature pertaining to kin interaction

among the elderly suggests that kin interaction declines

with the death of the spouse. Research pertaining to kin

interaction focuses on interaction with children as well

as members of the larger kin network (Adams, 1968; Berardo,

1967; Bock, 1972; Bock and Webber, 1972; Pihiblad and Adams,

1972; and Townsend, 1957). This hypotnesis only examines

interaction with kin considered by the respondents to be

living nearby.


Research Hypothesis 2

The aged widowed respondents will interact less fre-

quently with kin living far away than aged married individ-

uals.


Null hypothesis 2

There will be no difference between aged widowed and

aged married individuals in terms of the frequency of inter-

action with kin living far away.

Research related to this hypothesis includes the same

studies and findings as those presented For the second

hypothesis.









Research Hypothesis 3

The aged widowed respondents will interact less fre-

quently with friends than aged married individuals.


Null Hypothesis 3

There will he no difference between aged widowed and

aged married individuals in terms of the frequency of

friendship interaction.

Previous research pertaining to this hypothesis in-

cludes Berardo (1967), Blau (1961), Booth (1972), Griffiths etal.,

(1971), Lopata (1973), Pihlblad and Adams (1972), Rosow

(1967), and Townsend (1957). With the exception of

Townsend's finding that the widowed visit their neighbors

more than married individuals, the weight of the evidence

from these studies suggests that the widowed are more iso-

lated in terms of friendship interaction than married individ-

uals.


Research llypothesis 4

The aged widowed respondents will interact less fre-

quently in religious organizations than aged married individ-

uals.


Null Hypothesis 4

There will be no difference between aged widowed and

aged married individuals in terms of the frequency of inter-

action in religious organizations.

The literature upon which this hypothesis is based sug-

gests that widows are more likely to interact in religious










organizations than are either widowers or married individuals

(Berardo, 1967; Pihlblad and Adams, 1972; and Townsend, 1957).

However, the hypothesized relationship is still anticipated

because the extremely low rates of interaction in religious

organizations by widowers is expected to bring the total

interaction rate of the widowed below that of the married.


Research Hypothesis 5

The aged widowed respondents will interact less fre-

quently in formal organizations than aged married individ-

uals.


Null Ilypothesis 5

There will be no difference between aged widowed and

aged married individuals in terms of the frequency of

interaction in formal organizations.

Previous research (Berardo, 1967; Bock, 1972; Bock

and Webber, 1972, Harvey and Bahr, 1974; Lopata, 1973;

and Pihlblad and Adams, 1972) suggests that the widowed,

particularly widowers, are less likely to interact in formal

organizations than married individuals.


Research Hypothesis 6

Controlling for age, sex, race, education, health status,

and financial status, the rates of interaction with nearby

kin between the aged widowed and aged married individuals

will be less different than will the rates without controls.

Reviewed literature pertaining to kin interaction among

the elderly conclusively suggests that kin interaction declines









with the death of the spouse. Research pertaining to kin

interaction focuses on interaction with children as well

as members of the larger kin network (Adams, 1968; Berardo,

1967; Bock and Webber, 1972; Pihiblad and Adams, 1972; and

Townsend, 1957). This hypothesis only examines interaction

with kin considered by the respondents to be living nearby.

This hypothesis builds upon previous research by determin-

ing if differences in the rates of social interaction with

nearby kin between the aged widowed and aged married individ-

uals is a function of differential control over resources

examined in this study.


Research Hypothesis 7

Controlling for age, sex, race, education, health sta-

tus, and financial status, rates of interaction with far-

away kin between the aged widowed and aged married individ-

uals will be less different than will the rates without

controls.

Research related to this hypothesis includes the same

studies and findings as those presented for the sixth

hypothesis. This hypothesis builds upon previous research

by determining if differences in rates of social interaction

with kin considered to be living far away between aged

widowed and aged married individuals is a function of dif-

ferential control over social resources examined in this study.


Research Hypothesis 8

Controlling for age, sex, race, education, health sta-

tus, and financial status, the rates of interaction with










friends between the aged widowed and aged married individ-

uals will be less different than will the rates without

controls.

Previous research pertaining to this hypothesis includes

Berardo (1967), Blau (1961), Booth (1972), Griffiths et al.,

(1971), Lopata (1973), Pih]blad and Adams (1972), Rosow

(1967) and Townsend (1957). With the exception of Townsend's

finding that the widowed visit their neighbors more than

married individuals, the weight of the evidence from these

studies suggests that the widowed are more isolated in terms

of friendship interaction than married individuals. Addi-

tionally, among the widowed, the amount of friendship inter-

action is circumscribed by socioeconomic ststus and educa-

tional attainment.

This hypothesis goes beyond the preceding research by

determining if differences in the rates of friendship inter-

action between aged widowed and married individuals is a

function of differential control over resource variables

which include age, race, and health status in addition to

sex, education, and income, variables controlled in earlier

studies.


Research Hypothesis 9

Controlling for age, sex, race, education, health sta-

tus, and financial status, the rates of interaction in

religious organizations between aged widowed and aged married

individuals will be less different than will the rates with-

out controls.










The literature upon which this hypothesis is based

suggests that widows are more likely to interact in

religious organizations than are either widowers or married

individuals (Berardo, 1967; Pillblad and Adams, 1972;

and Townsend, 1957). However, because this research is

examining rates of interaction between aged widowed and

aged married individuals, widows and widowers are grouped

together so that they may be compared with the aged married

individuals. The differences in interaction between widows

and widowers are examined in this hypothesis when statistical

controls are applied for sex. This hypothesis expands upon

research in the area by determining if differences in the

rates of interaction in religious organizations between

aged widowed and aged married individuals is a function of

differential control over resources examined in this study.


Research Hypothesis 10

Controlling for age, sex, race, education, health sta-

tus, and financial status, the rates of interaction in

formal organizations between the aged widowed and aged married

individuals will be less different than will the rates with-

out controls.

Previous research (Berardo, 1967; Bock, 1972; Bock and

Webber, 1972, Harvey and Bahr, 1974; Lopata, 1973, Pihlblad

and Adams, 1972) suggests that the widowed, particularly

widowers, are less likely to interact in formal organizations

than married individuals. Additionally, among widows, a









positive relationship exists between economic status and

education, and rates of interaction in formal organizations.

This hypothesis attempts to extend the results of the pre-

ceding research by determining if differences in the rates

of social interaction between the aged widowed and aged

married individuals is a function of differential control

over resources examined in this study.


Data Collection

Data for this research were obtained from a large scale

sample survey designed to assess the mental health needs of

individuals living in Alachua County, Florida. The popula-

tion consisted of all residential households in that county.

The sampling design specified a systematic random sample

to be selected from a sample frame consisting of electrical

utility company residential household listings. The utility

listings combined claimed a 98 percent coverage of all house-

holds in the area. Because seven companies provide utilities

to the county's inhabitants the sample frame was constructed

to provide proportional representation to each of the utility

lists based upon the number of households they served.

The sample was drawn by initially selecting at random

a household from among the first thirteen households on

the sample frame. Thereafter, every thirteenth household

was included in the sample. Adult respondents (individuals

18 or older) within each household were randomly selected

with the Kish table. If the individual selected refused

to be interviewed the household itself was considered a










refusal and no additional solicitation of interviews

from among the household members was attempted. Potential

respondents were considered refusals if the interviewer

was certain that return visits would prove unsuccessful.

Households where no one was at home were revisited on at

least three separate occasions.

Total sample size included 2315 households. A sub-

sample of 511 respondents was randomly drawn from the

total sample in order to pretest the interview instrument

and polish interviewing techniques. Following completion

of the pilot study, interviewing for the major study,

which became the source of these data, began in June 1970

and was completed by the end of the year. The major study

resulted in 1645 usuable interviews. The total nonresponse

rate for the major study was 16.07 percent.

The nonresponse rate consisted of the following five

categories: "not at homes," 4.6 percent; "refusals," 8.06

percent; "unable to be interviewed," 1.42 percent; "unable

to locate," 1.63 percent; and "sampling problems," .03

percent. Individuals unable to be interviewed were those

suffering from mental or physical impairment to the extent

that they could not be interviewed. Sampling problems

occurred in a very small percentage of the cases when house-

holds moved during the interviewing period to geographic

locations outside the county lines.

Data from this large epidemiological survey yielded

information which is utilized in this research. Socio-

demographic data such as race, sex, age, education, financial










status, and marital status were obtained from the respon-

dents. Additionally, information concerning interaction

with kin and friends in addition to interaction in formal

organizations and religious organizations was collected.

This data bank provided the preceding information on 128

aged widowed and 145 aged married individuals. These 273

individuals are the focal point of this research.


Statistical Procedures

Two statistics are used in this research. They are

the Kolmogorov-Smirnov and the gamma statistics. The

Kolmogorov-Smirnov is the appropriate test for the first

five hypotheses while the gamma is applicable for testing

the final five hypotheses.


Kolmogorov-Siiirnov Statistic

The Kolmogorov-Smirnov statistic is a two-sample non-

parametric test which determines whether "two independent

samples have been drawn from the same population or from

populations with the same distribution" (Siegel, 1956:127).

This statistic is appropriate for testing the first five

hypotheses because the data are collapsed into a limited

number of ordered categories representing social interaction

frequencies. These hypotheses compare aged widowed with

aged married individuals in terms of frequency of social

interaction with friends, kin living both nearby and far

away, and interaction in formal organizations as well as

religious organizations.










Gamma Statistic

This statistic provides a measure of association for

grouped ordinal data (Blalock, 1972:424). It is used in

the last five hypotheses to measure the degree of agreement

between rates of social interaction of aged widowed and

aged married individuals. The research design specifies

that for each of the five hypotheses, rates of social inter-

action of the aged widowed are compared with interaction

rates of aged married individuals in friend groups, nearby

and faraway kin groups, formal organizations, and religious

organizations. Each hypothesis deals with social interaction

in only one of the five specified groups or organizations.

The degree of agreement between aged widowed and aged

married individuals is computed with the gamma statistic for

each area of social interaction. Subsequently, rates of

social interaction in the same areas are measured between

the aged widowed and aged married individuals controlling

for resources considered to influence rates of social inter-

action. The resources which are controlled include age, sex,

race, income, education, and health status.


Statistical Controls

Controls for all six of the resource variables are

accomplished for each of the five hypotheses. Because of

limitations in the data and the insufficient development of

methodological techniques for exerting statistical controls,

only one resource is controlled at a time.

Controls are administered as follows. Each resource is

divided into different levels which are either qualitative










or quantitative in nature depending upon the variable

examined. Measures of association are then computed

between the social interaction rates of aged widowed and

aged married individuals for each level of the control

variable. This process holds constant the impact of the

control variable upon the relationship being investigated.

Then to obtain some feeling for controlling for the

different levels simultaneously, the individual gammas

are averaged to form a partial gamma value. In computing

this average score the gammas for each level are weighted

to provide proportionate representation to the relative

number of respondents in each level of the control variable.

The partial gamma value for each resource is then com-

pared with the gamma calculated between the rates of social

interaction of aged widowed and aged married individuals

before controlling for resources. If the extension of

statistical controls results in the reduction of the gammas

computed for each level of the control variable then that

factor may be assumed to influence the interaction rates

of the elderly. This process is accomplished for each of

the six resources for every hypothesis.

However, before the gammas calculated on each of the

different levels of each of the control variables can be

combined to yield a partial gamma, a test of statistical

interaction has to be computed to determine if the relation-

ships between interaction rates of the aged differ for

different levels of the control variable. If this situation










occurs the individual gammas for each of the levels of the

control variable can not be averaged together for lack of

commonality.

Statistical interaction is tested For significance in

either one of the two following ways depending upon the

nuniher of levels in the control variable. If the control

variable is divided into two levels a Z test statistic is

used to test for a significant difference between the two

measures of association. The standard error of the sampling

distribution is the square root of the summed variances of

the two gammas. If the control variable has three levels,

each of the gammas is weighted and combined to yield a

partial gamma which is then subtracted from each of the

three individual gammas in turn, squared, and divided by

each of the variances of the individual gammas. The results

are summed to provide a score which is tested for significance

on the X2 sampling distribution with c 1 degrees of freedom

where c is the number of levels of the control variable.

These two tests are necessary to determine whether

or not "the sample interactions are sufficiently large that

they could have readily occurred by chance even if there

was no population interaction" (Blalock, 1972:309). If the

interactions are not statistically significant and not large

enough to be substantively meaningful, controls for each re-

source in each hypothesis are administered (Blalock, 1972:309).

However, if there are significant statistical interactions,

different levels of the control variable are acknowledged

as circumscribing relationships between the interaction rates










of aged widowed and aged married individuals. An illus-

tration of the process of exerting statistical controls

in this research follows.

Statistical controls for age for each of the five

areas of social interaction are accomplished by dividing

the aged into three levels: respondents 60 to 65 years old,

66 to 75 years old, and 76 years old and older. Then, for

each level, rates of social interaction are compared between

aged married and aged widowed individuals for each of the

areas of social interaction. Comparisons between the two

aged groups are accomplished with the gamma statistic. The

gammas are weighted and averaged to yield a score representing

comparisons between the aged widowed and aged married con-

trolling for the three levels of age. This score is then

compared with a comparison between rates of social interaction

between the aged widowed and aged married individuals not

controlling for resources.

Each of the remaining resource variables is controlled

in much the same fashion, the only differences among them

resulting from the number and types of levels of control

variables established.

Race, for control purposes, is dichotomized into "blacks"

and "whites." A third racial category, "other," had too few

respondents to be utilized as a control level. Sex is also

used as a control variable. Financial status, which is in-

dicated by yearly household income, is broken down into two

control levels, those with household incomes greater and less










than $4500 per year. This figures was selected, not

because of any particular social, economic, or political

significance, but because it represents the median income

of the aged respondents interviewed.

Health status is separated into two categories, those

who reported that they have health problems, and those who

reported that they do not. Finally, education, for control

purposes, is divided into three levels, those who range

from no formal education to completion of the eighth grade,

individuals who completed some high school or are high

school graduates, and those respondents who have some college,

technical, or trade education.


A Description of the Data

The population to which the findings from these data

are generalizable includes all individuals 60 years old

and older living in Alachua County, Florida. Inferences

can be made from this sample because the assumptions of

probability sampling are met. The sample consists of 273

individuals.

Table 1 summarizes the data by displaying the fre-

quency and percent of respondents according to different

values of the seven variables analyzed in this research:

marital status, sex, age, race, education, health status,

and income. The last six variables are resources which are

controlled in this study in an attempt to investigate the

applicability of the theoretical proposition set forth in

the second chapter which was developed from the perspective

of the exchange conceptual framework.












TABLE 1. Frequency and Percent of Respondents by Levels
of Marital Status, Sex, Age, Race, Education,
Health Status, and Income



Frequency Percent

Married 146 53.1
Marital Status
Widowed 129 46.9

Total 275 100.0


Male 102 37.0
Sex
Female 173 63.0

Total 275 100.0


60 to 65 Years Old 100 36.4

Age 65 to 75 Years Old 120 43.6

75 Years Old and 55 20.0
Older

Total 275 100.0


White 195 70.9
Race
Black 80 29.1

Total 275 100.0


0 to 9th Grade 137 49.8

9th Grade to High 70 25.5
Education School

Some College, Tech- 68 24.8
nical, or Trade
Education


Total 275


100.1












TABLE 1. (continued)


Frequency Percent

Health Problems 197 71.6
Health Status
No Health Problems 78 28.4

Total 275 100.0


Household Income 141 61.6
Less than $4500
Income
Household Income 88 38.4
Equal to or
Greater than $4500

Total 229 100.0









Of the individuals sampled, 53.1 percent are married

and 46.9 percent are widowed. The ratio of married to

widowed individuals in the sample is 1.13 to 1. Using

national census data as a point of reference, the marital

status distribution of the sample includes a dispropor-

tionate number of widowed. The ratio of married to widowed

individuals for United States residents 60 years old and

older is 5.77 to 1.

The sex distribution of this aged sample includes 37

percent males and 63 percent females. The sample, when

compared with the sex distribution of Alachua County,

slightly underrepresents females. Sixty-seven percent of

all individuals 60 or older in the county were women. A

cross tabulation of sex by marital status reveals that

there are over five times as many widows as widowers and

there are one-third again as many married women as men.

The age distribution of the respondents ranges from

60 to 91. The model age of the sample is 65, the mean

age is 70.4, and the median age is 68. For purposes of

statistical analysis, the sample is divided into three

levels. The middle level, which consists of all respondents

65 to 75 years old, is the largest including 43.6 percent of

the interviewees. The 60-to 65-year-old category includes

36.4 percent of the respondents and the remaining level,

consisting of those individuals 75 years old and older, is

smallest encompassing 20 percent of the sample. The age

distribution of Alachua County, the population to which










inferences from the sample are made, closely resembles the

sample. Of all individuals in the county aged 60 or over,

31 percent are 60 to 65, 42.7 percent are 65 to 75, and

24.2 percent are 75 years old and older.

The sample consists of 70.9 percent white and 29.1

percent black individuals. The sampling error in this as

well as the preceding category is fairly small because,

for the county as a whole, the percent of aged whites is

73.3 and 26.5 for blacks. The percentages do not total

to 100 because of a small number (23) of aged nonblack

minority group members living in the county.

Just under one half of all respondents (49.8 percent)

completed less than nine years of formal education. Approx-

imately one fourth (25.5 percent) of the sample completed

at least the ninth grade but did not advance their education

beyond high school graduation. The remaining quarter (24.8

percent) of the interviewees completed some college, tech-

nical, or trade education.

The vast majority of the aged individuals interviewed

reported health problems. Only 28.4 percent perceived

themselves as having no physical or health problems at the

time of the interview.

Interviewees with household incomes less than $4500

comprised 61.6 percent of the sample. The remaining por-

tion of the sample has household incomes equal to or greater

than $4500 per year. The distribution of household income









of the respondents presents some interesting contrasts

when cross tabulated with marital status. For the widowed,

total household incomes range from $300 to $11,400 per year.

However, for married respondents, total household incomes

range from $960 to $45,000. The median income for the two

groups respectively is $2124 and $5000. These particular

figures should be viewed with some skepticism because of

the large number of missing observations. It is impossible

to determine if the missing data are randomly distributed

or if selective factors operated to distort the sample

with regard to reportage of household income.


Summary

This research examines differences between rates of

social interaction for aged widowed and aged married individ-

uals. Social interaction is measured in five different

areas, friend interaction, interaction with kin living

nearby and far away, and interaction in formal organiza-

tions and religious organizations. Comparisons of inter-

action rates between the aged widowed and aged married

individuals are made both before and after controlling for

age, sex, race, education, financial status, and health

status.















CHAPTER IV

FINDINGS


This chapter presents empirical findings gleaned from

this research enterprise. The statistical analyses of the

data are examined separately for each hypothesis.


Hypotheses

Research Hypothesis 1

The aged widowed respondents will interact less fre-

quently with nearby kin than aged married individuals.


Null Hypothesis 1

There will be no difference between aged widowed and

aged married individuals in terms of the frequency of

interaction with nearby kin.

The test for differences between married and widowed

individuals, using the frequency of interaction with nearby

kin, is the Kolmogorov-Smirnov statistic. The differences

between the two groups is not significant (see Table 2).

This result fails to support the contention of findings

from earlier research (Berardo, 1967; Bock and Webber, 1972;

and Townsend, 1957) that aged married individuals interact

with kin more than aged widowed individuals. Insufficient

support for this hypothesis could reflect the finding by









Pihiblad and Adams (1972) that kin interaction declines

upon the death of a spouse for elderly males but not

elderly females indicating that sex, more than marital

status, influences the interaction of the aged with their

children.


Research Hypothesis 2

The aged widowed respondents will interact less fre-

quently with kin living far away than aged married individ-

uals.


Null Hypothesis 2

There will be no difference between aged widowed and

aged married individuals in terms of the frequency of inter-

action with kin living far away.

No significant difference between the two samples of

the aged is found in terms of their rates of interaction

with kin considered to be living far away (see Table 2).

The Kolmogorov-Smirov statistic is used to test for dif-

ferences between the two groups. As in the preceding test,

these results fail to confirm earlier research findings

(Berardo, 1967; Bock and Webber, 1972; and Townsend, 1957).

The inconsistency of the findings from this research with

these earlier studies concerning kin interaction perhaps,

as suggested in the first hypothesis, results more from sex

than widowhood status. If not, few alternative explanations

present themselves. These conflicting results can not

readily be attributed to the small sample size of this survey.










All four groups, that is, aged widowed and married individ-

uals interacting with nearby and far away kin, have at

least 100 respondents. Assuming relatively small sampling

as well as nonsampling errors because of the sample design

and sample size, this inconsistency is perplexing and per-

haps can only be explained by the fact that the studies in

question are making inferences to incomparable populations.

This suggested explanation notwithstanding, further inves-

tigation into this area is most certainly warranted.


Research Hypothesis 3

The aged widowed respondents will interact less fre-

quently with friends than aged married individuals.


Null Hypothesis 3

There will be no difference between aged widowed and

aged married individuals in terms of the frequency of

friendship interaction.

The Kolmogorov-Smirnov statistic is used to test for

differences between the two samples. For aged individuals

a significant difference between widowed and married persons

is found in terms of interacting with friends (see Table 2).

The difference between the two groups is significant at the

.05 level of significance;however, it is not in the hypo-

thesized direction. Therefore the null is not rejected in

favor of the research hypothesis.

This finding lends support to Townsend's (1957) earlier

research on the aged but is inconsistent with research












TABLE 2. Comparisons Between Married and Widowed Individ-
uals in Terms of Frequency of Interaction in
Nearby Kin Groups, Far-Away Kin Groups, Friend
Groups, Religious Organizations and Formal
Organizations


Frequency of
Interaction


Weekly
Monthly*
Yearly
Never


Total


Weekly
Monthly
Yearly"
Never


Marital Status
Cumu- Cumu-
Married lative Widow- lative
(N) percent ed (N) percent


72.0
95.0
98.0
100.0


74.5
90.5
95.2
99.9


100.0 106 99.9+

Kolmogorov-Smirnov=.33


4.0
40.0
83.2
100.0


7.0
46.5
74.6
100.0


100.0 114 100.0

Kolmogorov-Smirnov=1.93


25 20.5


63.1
89.3
100.0


46 38.0


72.7
93.4
100.0


Total


Religious
Organizations


122 100.0 121 100.0

Kolmogorov-Smirnov=7.44p<.05


Weekly*
Monthly
Yearly
Never


36.9
64.6
83.0
100.0


44.9
68.5
89.8
100.0


141 100.0 127 100.0


Kolmogorov-Smirnov=1.71


Group

Nearby
Kin
Groups


Far -
Away
Kin
Groups


Total


Friend
Groups


All the
Time*
Often
Sometimes
Never


Total













TABLE 2. (continued)


Frequency of
Group Interaction


Marital
Cuinm-
Married lative
(N) percent


Status
Cumu -
Widow- lative
ed CN) percent


Fo rmia I
Organizations


Total


67 100.0


49 100.0


Kolmogorov-Smirnov=6.95p<.05


*Cells of maximum cumulative difference.
+
In all tables throughout the dissertation percentages not
totaling to 100 are due to errors in rounding.


Often
Generally
Seldom*
Never


13.4
35.8
74.6
100.0


20.4
46.9
95.9
100.0










findings from Berardo (1967) and Booth (1972). However,

even though the findings are not in the hypothesized direc-

tion, the overall research design developed to investigate

the applicability of the exchange conceptual framework in

explaining the social interaction of the aged, is unaffected.

To this end, this research seeks to determine if the dif-

ference between the two groups is a function of control

over resources.


Research Hypothesis 4

The aged widowed respondents will interact less fre-

quently in religious organizations than aged married

individuals.


Null Hypothesis 4

There will be no difference between aged widowed and

aged married individuals in terms of the frequency of

interaction in religious organizations.

The Kolmogorov-Smirnov statistic is used to test for

differences in frequency of participation in religious

organizations between the two preceding groups of aged

individuals. The null hypothesis is not rejected.

Previous research in this area (Berardo, 1967; Pihlblad

and Adams, 1972; and Townsend, 1957) differentiates by sex

among the aged widowed, suggesting that widows are more

likely to interact in religious organizations than are either

widowers or married individuals. Because this hypothesis

makes no distinction by sex among the aged widowed it can










neither lend support to nor detract from earlier studies.

However, controls for sex are extended in the ninth hypoth-

esis.


Research Hypothesis 5

The aged widowed respondents will interact less fre-

quently in formal organizations than aged married individ-

uals.


Null Hypothesis 5

There will be no difference between aged widowed and

aged married individuals in terms of the frequency of inter-

action in formal organizations.

A significant difference between the two samples of

aged individuals in terms of their interaction in formal

organizations is found (see Table 2). The Kolmogorov-

Smirnov statistic is used to test for differences between

the two groups. The two samples are significantly dif-

ferent at the .05 level of significance. However, the data

are not in the hypothesized direction, suggesting that the

widowed interact with greater frequency in formal organiza-

tions than married individuals. The null hypothesis is not

rejected.

Relevant earlier research on the interaction of the

aged in formal organizations was conducted by Berardo (1967),

Bock and Webber (1972), Harvey and Bahr (1974), and Pihlblad

and Adams (1972). Their analyses were conducted while

implementing controls for sex. They indicate widowers are










less likely to interact in formal organizations than

either married individuals or widows. Moreover, the dif-

ference between the latter two groups is minimal. No

controls for sex are administered in this hypothesis he-

cause it was thought that, when considered together, the

widowed would exhibit smaller rates of social interaction

than married individuals. However, sex is held constant

in the final hypothesis.

Unfortunately, direct tests of the remaining five

hypotheses are impossible. This occurs because comparisons

between the measure of association between widowed and mar-

ried individuals in the total sample and the measure of

association between widowed and married individuals for

each of the six control variables would involve tests of

statistical significance based on dependent samples. For

this reason no statistical inferences to any conceptual

population are possible. Therefore, the succeeding hypoth-

eses will serve only to guide the research effort. Null

hypotheses are unwarranted when no direct inferences to a

population are desired and therefore are not presented for

the last five hypotheses. When no statistical interaction

exists, comparisons between the six partial gammas, repre-

senting the measure of association between interaction

rates of aged widowed and aged married individuals for each

control variable, and the total gamma score, will be de-

scribed for each area of social interaction investigated in

this research.










Research Hypothesis 6

Controlling for sex, age, race, education, health

status, and financial status, the rates of interaction

with nearby kin between the aged widowed and aged mar-

ried individuals will be less different than will the

rates without controls.

Table 3 indicates that the measure of association

between married and widowed respondents engaging in inter-

action with nearby kin is -.03.1 The size of this measure

is negligible. However, it is consistent with the first

hypothesis which found no significant difference between

married and widowed respondents engaging in interaction

with nearby kin. Comparisons of the gamma score with the

partial gammas, which represent controls for the six re-

sources, reveal that none of the variables appear to be

important as social exchange resources.



The formula for gamma is the number of concordant
pairs minus the number of discordant pairs divided by their
sum. All tables in this research compare interaction rates
of widowed persons with interaction rates of married per-
sons. When comparing the two sets of ordered rankings of
social interaction the starting point for determining what
are concordant and discordant pairs is arbitrary. For this
research, negative gamma values indicate that the widowed
interact with greater frequency in the social settings in-
vestigated while positive gamma values mean that married
individuals interact more often in these areas. however,
it should be emphasized that the last five hypotheses are
concerned only with whether controls for variables indicate
that they are likely candidates for exchange resources in
the social context investigated. To this end, the sign
values of the gammas and partial gammas are meaningful only
when observed relative to one another to determine if the
partial gammas are closer to zero than the gammas.











TABLE 3. Measure of Association Between, and Frequency and
Percent of Married and Widowed Respondents
Engaging in Interaction with Nearby Kin by Sex,
Age, Race, Education, Health Status, and Income




Frequency of In- 'Marital Status
Group traction with Married Widowed
Nearby Kin (N) (%) (N) ()


Total Group Weekly 72 72.0 79 74.5
Monthly 23 23.0 17 16.0
Yearly 3 3.0 5 4.7
Never 2 2.0 5 4.7

Total 100 100.0 106 99.9

Gamma=-.03 0 =.15
g

Sex Male Weekly 45 45.0 12 31.3
(Gamma=- 5, Monthly 9 9.0 1 1.0
o =.41) Yearly 1 1.0 0 0.0
g Never 1 1.0 0 0.0

Female Weekly 27 27.0 67 63.2
(Gamma=- .17, Monthly 14 14.0 16 15.1
S=.17) Yearly 2 2.0 5 4.7
gNever 1 1.0 5 4.7

Total 100 100.0 106 100.0

Partial Gamma=-.28 o =.18
g


Age 60 to 65 Weekly 33 33.0 24 22.6
Years Old Monthly 10 10.0 2 1.9
(Gamma=- .19, Yearly 2 2.0 1 1.0
o =.27) Never 0 0.0 2 1.9

65 to 75 Weekly 31 31.0 30 28.3
Years Old Monthly 11 11.0 7 6.6
(Gamma=-.08, Yearly 1 1.0 2 1.9
S=.23) Never 1 1.0 1 1.0
g











TABLE 3. (continued)


Frequency of In- Marital Status
Group teraction with Married Widowed
Nearby Kin () N ) (N) (%)


Age 75 Years Old Weekly 8 8.0 25 23.6
or Older Monthly 2 2.0 8 7.5
(Gamma=. Yearly 0 0.0 2 1.9
o =.35) Never 1 1.0 2 1.9

Total 100 100.0 106 100.1

Partial Gamma=-.08 a =.15
g


Race White Weekly 48 48.0 53 50.0
(Gamma=.06, Monthly 15 15.0 14 13.2
o =.18) Yearly 2 2.0 3 2.8
g Never 1 1.0 4 3.8

Black Weekly 24 24.0 26 24.5
(Gamma=-.23, Monthly 8 8.0 3 2.8
o =.26) Yearly 1 1.0 2 1.9
g Never 1 1.0 1 1.0

Total 100 100.0 106 100.0

Partial Gamma=-.03 o =.14
g


0 to 9th
Grade
(Gamma=-.14,
a =.2)

9th Grade to
High School
Graduate
(Gamma=-.1,
o =.29)

Some College,
Technical or
Trade Education
(Gamma= 31 ,
a =.3)
g


Weekly
Monthly
Yearly
Never

Weekly
Monthly
Yearly
Never


Weekly
Monthly
Yearly
Never

Total


41.4
13.1
1.0
2.0

20.2
5.0
2.0
0.0


10 10.1 11
5 5.0 6
0 0.0 2
0 0.0 1

99 99.8 105


Partial Gamma=-.05


Edu-
ca-
tion


42.8
6.7
2.8
1.9

21.0
3.8
0.0
1.9


10.5
5.7
1.9
1.0

100.0


o =.15
g










TABLE 3. (continued)


Frequency of In-
teraction with
Nearby Kin


Marital Status
Married Widowed
(N) (%) (N) (%)


Health Health Weekly 47 47.0 62 58.5
Status Probl ems Monthly 16 16.0 16 15.1
(Gamma=.01, Yearly 1 1.0 3 2.8
a =.17) Never 2 2.0 4 3.8
g
No Health Weekly 25 25.0 17 16.0
Problems Monthly 7 7.0 1 1.0
(Gamma=-.12, Yearly 2 2.0 2 1.9
a =.31) Never 0 0.0 1 1.0
g
Total 100 100.0 106 100.0

Partial Gamma=-.02 a =.15
g


Income Household Weekly 29 33.7 57 64.8
Income Less Monthly 12 14.0 10 11.4
Than $4500 Yearly 2 2.3 4 4.5
(Gamma=-.24, Never 1 1.2 2 2.3
a =.18)

Household Weekly 33 38.4 8 9.1
Income Equal Monthly 8 9.3 5 5.7
To or Greater Yearly 1 1.2 0 0.0
Than $4500 Never 0 0.0 2 2.3
(Gamma= .53,
o =.22) Total 86 100.1 88 100.1

Partial Gammae=.01 a =.14


*Statistical interaction; z=-2.72; p<.01.


Group










Moreover, the application of controls resulted in an

increase rather than the expected decrease in the measure

of association between rates of interaction of aged mar-

ried and aged widowed individuals for all levels of all

control variables except for the elderly expressing health

problems. This phenomenon, which is found frequently

throughout this research, occurs when the extension of con-

trol levels serves to intensify the measure of association

within the limits established by the control categories.

Furthermore, the attempted control for income among the

individuals sampled resulted in increased gammas for both

high and low income respondents which are sufficiently large

so that the difference between the two levels is statisti-

cally significant. This finding suggests that among the

lower income respondents the widowed interact more fre-

quently with kin living nearby while among the higher in-

come groups the married interact with nearby kin more often.


Research Hypothesis 7

Controlling for age, sex, race, education, health status,

and financial status, the rates of interaction with far away

kin between the aged widowed and aged married individuals

will be less different than will be rates without controls.

The measure of association between married and widowed

individuals engaging in interaction with kin considered by

the respondents to be living far away is -.02 (see Table 4).

For this hypothesis, as well as the preceding one, the










TABLE 4. Measure of Association Between, and Frequency
and Percent of Married and Widowed Respondents
Engaging in Interaction with Far Away Kin by
Sex, Age, Race, Education, Health Status, and
Income


Frequency of In-
Group traction with
Far Away Kin


Total Group Weekly
Monthly
Yearly
Never

Total

Gamma=-.


Sex Male Weekly
(Gamma=.19, Monthly
S=.23) Yearly
g Never

Female Weekly
(Gamma=. 04, Monthly
o =.13) Yearly
g Never

Total

Partial


Age 60 to 65 Weekly
Years Old Monthly
(Gamma=.18, Yearly
o =.18) Never

65 to 75 Weekly
Years Old Monthly
(Gamma=-. 21, Yearly
a =.15) Never
g


Marital Status
Married Widowed
(N) (%) (N) (%)


5 4.0 8 7.0
45 36.0 45 39.5
54 43.2 32 28.1
21 16.8 29 25.4

125 100.0 114 100.0

02 o =.1


4 3.2 0 0.0
18 14.4 6 5.3
31 24.8 3 2.6
14 11.2 7 6.1

1 1.0 8 7.0
27 21.6 39 34.2
23 18.4 29 25.4
7 5.6 22 19.3

125 100.2 114 99.9

Gamma=.09 o =.12


0 0.0 1 1.0
29 23.2 11 9.6
22 17.6 9 7.9
10 8.0 9 7.9

3 2.4 6 5.3
15 12.0 20 17.5
29 23.2 14 12.3
9 7.2 10 8.8










TABLE 4. (continued)


Frequency of In
traction with
Far Away Kin


75 Years
Old and
Older
(Gamma=. 1,
0 =.31)


Weekly
Monthly
Yearly
Never

Total


Marital Status
Married Widowed
(N) (%) (N) (%)


1.6 1
1.0 14
2.4 9
1.6 10


125 100.2


1.0
12.3
7.9
8.8


114 100.3


Partial Gamma=-.01


a =.1
g


Race White Weekly 2 1.6 5 4.4
(Gamma=-.21, Monthly 35 28.0 38 33.3
o =.13) Yearly 38 30.4 19 16.7
Never 14 11.2 13 11.4

Black Weekly 3 2.4 3 2.6
(Gamma=.3, Monthly 10 8.0 7 6.1
o =.17) Yearly 16 12.8 13 11.4
g Never 7 5.6 16 14.0

Total 125 100.0 114 99.9

Partial Gamma*=-.05 o =.11
g

Educa- 0 to 9th Weekly 4 3.2 4 3.6
tion Grade Monthly 14 11.4 21 18.8
(Gamma=-.02, Yearly 31 25.2 17 15.2
o =.14) Never 13 10.6 19 17.0
g
9th Grade to Weekly 1 1.0 1 1.0
High School Monthly 16 13.0 13 11.6
Graduate Yearly 9 7.3 7 6.2
(Gammna=.08, Never 5 4.1 6 5.4
o =.22)

Some College, Weekly 0 0.0 3 2.7
Technical or Monthly 14 11.4 11 9.8
Trade Educa- Yearly 14 11.4 7 6.2
tion Never 2 1.6 3 2.7
(Gamma=-. 21 ,
o =.23) Total 123 100.2 112 100.2
Partial Gamma=-.04 o =.1
g


Group










TABLE 4. (continued)


Frequency of In-
teraction with
Far Away Kin


Marital Status


Married
(N) ( )


Widowed
(N) (%)


Health Health
Status Problems
(Gamma=-.06,
S=.12)
g
No Health
Problems
(Gamma=.14,
o =.2)
g


Weekly
Monthly
Yearly
Never

Weekly
Monthly
Yearly
Never

Total


125 100.0 114 100.1


Partial Gamma= .00 c =.12
g


Income household
Income Less
Than $4500
(Gamma=-.06,
o =.14)

Household
Income Equal
To or Greater
Than $4500
(Gamma=-.08,
o =.23)
g


Weekly
Monthly
Yearly
Never


Weekly
Monthly
Yearly
Never

Total


2.9
11.5
21.2
6.7


1.9
22.1
25.0
8.6


104 99.9 95 100.0


Partial Gamma


.07 a =.12
g


*Statistical interaction; z=-2.42; p<.01.


Group


2.4
24.0
29.6
9.6

1.6
12.0
13.6
7.2


5.3
32.4
23.7
16.7

1.8
7.0
4.4
8.8


4.2
33.7
24.2
18.9


2.1
7.4
5.3
4.2


~









extremely small size of the gamma makes any interpretation

of the impact of statistical controls for variables almost

meaningless. None of the controls extended reveal a diminu-

tion in the strength of the relationship between widowed and

married individuals. Indeed, virtually all controls for the

resources result in increased measures of association be-

tween rates of interaction of aged widowed and aged married

individuals within the control categories. Moreover, the

gammas computed for different levels of race are sufficiently

different to be statistically significant indicating that

among white respondents the widowed are more likely to

interact frequently with kin perceived to be living far

away while among blacks the married tend to interact more

frequently with far away kin.

Earlier research by Pihlblad and Adams (1972) sug-

gests that widowhood decreases kin interaction for men but

not women. These data support this earlier finding. The

control for sex reveals that the measure of association

between married and widowed females is .04 whereas the

comparable gamma computed for males (gamma=.19) indicates

that married males interact more frequently with kin living

far away than widowers.


Research Hypothesis 8

Controlling for age, sex, race, education, health

status, and financial status, the rates of interaction with

friends between the aged widowed and aged married individ-

uals will be less different than will the rates without controls.










A gamma score of -.27 indicates the measure of associa-

tion between rates of interaction with friends for married

and widowed individuals (see Table 5). This finding sug-

gests that the aged widowed tend to interact more frequently

with friends than aged married individuals. Controls for

all six resource variables examined in this hypothesis

indicate no reductions in the measure of association be-

tween married and widowed individuals. The data do not sup-

port the hypothesis.

The research findings from this hypothesis draw atten-

tion to a few noteworthy observations. First, the impact

of income in circumscribing social interaction with friends

is depicted in other research by Blau (1961), Griffiths

(1971), Lopata (1973, and Rosow (1967). These previous

studies suggest a positive relationship between financial

status and both friendship interaction and satisfactory

friendship relationships. This finding is not supported

in the present research.

An additional point of interest in this hypothesis

concerns the extension of educational controls. These con-

trols reveal that the widowed are more likely to interact

with friends than married individuals in each of the three

educational levels. Moreover, it is among the highest

educational level that widowed interact with the greatest

frequency followed by the lowest and then the intermediate

educational levels. This finding is somewhat inconsistent

with Lopata's research which suggests a direct relationship









TABLE 5. Measure of Association Between, and Frequency and
Percent of Married and Widowed Respondents Engaging
in Interaction with Friends by Sex, Age, Race,
Education, Health Status, and Income


Frequency of In
traction with
Friends


Marital Status
Married Widowed
(N) (%) (N) (%)


Total Group All the Time 25 20.5 46 38.0
Often 52 42.6 42 34.7
Sometimes 32 26.2 25 20.7
Never 13 10.6 8 6.6

Total 122 99.9 121 100.0

Gamma=-.27 0 =.1
g


Male
(Gamma=-. 31,
o =.21)


Female
(Gamma=- .18,
o =.13)
g


All the Time
Often
Sometimes
Never

All the Time
Often
Sometimes
Never


122 99.9 121 99.9


Partial Gamma


60 to 65
Years Old
(Gamma=-.13,
0 =.18)

65 to 75
Years Old
(Gamma=-. 39,
o =.13)
g


All the Time
Often
Sometimes
Never

All the Time
Often
Sometimes
Never


.23 =.12


8.2 9
21.3 12
10.6 7
4.9 3


11.5
18.0
13.9
4.1


Groups


9.8
27.0
15.6
7.4

10.6
15.6
10.6
3.3


6.6
4.1
1.6
2.5

31.4
30.6
19.0
4.1


Total


7.4
9.9
5.8
2.5

21.5
11.6
8.3
1.6











TABLE 5. (continued)


Frequency of In
teraction with
Friends


Marital Status
Married Widowed
(N) ( ) (N) (%)


Age 75 Years
Old and
Older
(Gamma=- .37,
o =.25)
g


All the Time
Often
Sometimes
Never


Total


122 100.0 121 100.0


Partial Gamma*=-.07


Race White
(Gamma=-.22,
o =.12)

Black
(Gamma=-.39,
a =.16)
g '


All the Time
Often
Sometimes
Never

All the Time
Often
Sometimes
Never


122 100.0 121 100.1


Partial Gamma=-


.27 g =.09
b


Educa- 0 to 9th
tion Grade
(Gamma =-.28,
0 =.13)

9th Grade to
High School
Graduate
(Gamma=- 09,
o .21)

Some College
Technical or
Trade Educa-
tion
(Gamia=-. 44,
S=.18)
gi


All the Time
Often
Sometimes
Never

All the Time
Often
Sometimes
Never



All the Time
Often
Sometimes
Never

Total
Partial Ganm


13.3
20.0
15.0
4.2

4.2
13.3
6.7
2.5


3.3 12
10.0 12
4. 2 5
3.3 1


23.5
14.3
10.1
3.4

5.0
10.9
5.9
1.7



10.1
10.1
4.2
1.0


120 100.0 119 100.2
ma=-.27 o =.1


Group


1. 0 11
3.3 16
1.6 8
1.6 3


9.1
13.2
6.6
2.5


a =.1
9


13.1
27.9
18 .8
9.0

7.4
14 .8
7.4
1.6


20.7
27.3
15.7
5.0

17.4
7.4
5.0
1.6


Total


~










TABLE 5. (continued)


Frequency of In-
teraction with
Friends


Marital Status
Married Widoew
(N) (%) (N) (%)


Health Health
Status Problems
(Gamma=-. 22,
o =.11)

No Health
Problems
(Gamma=-.36,
S=.18)
g


All the Time
Often
Sometimes
Never

All the Time
Often
Sometimes
Never


15.6
27.0
17.2
6.6

4.9
15.6
9.2
4.2


28.1
28.9
15.7
5.0

9.9
5.8
5.0
1.6


Total 122 100.3 121 100.0

Partial Gamma=-.26 0 =.09
g


Income Ilousehold
lncomei
Less Than
$4500
(Gaiimma=- 09,
a .14)
g
Household
Income
Equal to
Or Great-
er Than
$4500
(Gamma=-.4,
a =.18)
g


All the Time
Often
Sometimes
Never



All the Time
Often
Sometimes
Never

Total


Partial Gamma=


13.3
20.0
10.5
2.8


30.4
28.4
16.7
5.9


9 8.6 7 6.9
22 21.0 7 6.9
16 15.2 4 3.9
9 8.6 1 1.0

105 100.0 102 100.1


.2 a =.11
9


"*Stntistical interaction; X2=6.74; p<.05


Group










between educational status and friend satisfaction. How-

ever, this inconsistency is valid only when the assumption

is made that friend interaction and friend satisfaction

are indicative of one another.


Research hypothesis 9

Controlling for age, sex, race, education, health

status, and financial status, the rates of interaction in

religious organizations between aged widowed and aged mar-

ried individuals will be less different than will the rates

without controls.

The strength of the relationship between interaction

rates in religious organizations for aged widowed and aged

married individuals, as measured by gamma, is -.33 (see

Table 6). This score suggests that the widowed, as a

group, are more likely to interact in religious organiza-

tions with greater frequency than married individuals.

Generally speaking, the hypothesis is not supported by the

data. Controls for none of the variables result in meaning-

ful diminutions in the partial gammas. However, income,

when held constant, demonstrates a very modest reduction in

the measure of association between the two groups. This

indicates a possible interpretation of this factor as an

independent variable which affects social relationships in

the capacity of an exchange resource. However, the extremely

small difference makes any interpretation of this nature

particularly daring.










TABLE 6. Measure of Association Between, and Frequency and
Percent of Married and Widowed Respondents Engaging
in Interaction in Religious Organizations by Sex,
Age, Race, Education, Health Status, and Income



Frequency of In- Marital Status
Group traction in Reli- Married Widowed
gious Organizations (N) (%) (N) (%)


Total Group Weekly 52 36.9 57 44.9
Monthly 39 27.6 30 23.6
Yearly 26 18.4 27 21.2
Never 24 17.0 13 10.2

Total 141 99.9 127 99.9

Gamma=-.13 o =.09


Sex Male Weekly 23 16.3 3 2.4
(Gamma=.28, Monthly 25 17.7 3 2.4
S=.17) Yearly 18 12.8 9 7.1
g Never 14 9.9 3 2.4

Female Weekly 29 20.6 54 42.5
(Gamma= 07, Monthly 14 9.9 27 21.2
o =.13) Yearly 8 5.7 18 14.2
g Never 10 7.1 10 7.9

Total 141 100.0 127 100.1

Partial Gamma=.06 a =.11
g

Age 60 to 65 Weekly 21 14.9 15 11.8
Years Old Monthly 25 17.7 9 7.1
(Camma= -. 27, Yearly 10 7.1 6 4.7
o =.16) Never 10 7.1 1 1.0

65 to 75 Weekly 28 19.8 24 18.9
Years Old Monthly 12 8.5 11 8.7
(Camma=-.06 Yearly 13 9.2 14 11.0
a =.14) Never 11 7.8 5 3.9
0









TABLE 6. (continued)


Frequency of In-
teraction in Reli-
gious Organizations


Marital Status
Married Widowed
(N) (%) (N) (%)


Age 75 Years Old
and Older
(Gamma =- 3,
o =.23)
0


Partial Gamma=-.18


Race White
(Gamma=- .22,
S=.' 11)


Black
(Gaimla 13,
a .18)
g


Partial Gamma=-.12


0 to 9th
Grade
(Gamma=- .03,
a .13)

9th Grade
High School
Graduate
(Gamma=. 00,
o =.18)
if


Some College,
Technical or
Trade Education
(Gamma=- .44 ,
S=.17)
g


Weekly
Monthly
Yearly
Never

Weekly
Monthlly
Yearly
Never


Weekly
Monthly
Yearly
Never

Total


139 100.1 125 100.2


Partial (Gamma=-.12


o =.09
g


Group


Weekly
Monthly
Yearly
Never


2.1 18
1.4 10
2.1 7
2.1 7


Total


14.2
7.9
5.5
5.5


141 99.8


127 100.2


S .1
g


Weckl y
Monthly
Yearly
Never

Weekly
Monthly
Yearly
Never


27.0
14.9
14.2
15.6

9.9
12.8
4.2
1.4


34.6
15.6
12.6
8.7

10.2
8.7
8.7
1.6


Total


141 100.0


Edu-
ca
tion


127 100.1


a =.09
9


17.3
17.3
7.2
6.5

12.9
3.6
6.5
5.8


7.2 19
6.5 4
5.0 7
4.3 1


20.8
13.6
13.6
4.0

8.8
7.2
2.4
4.8


15.2
3.2
5.6
1.0










TABLE 6. (continued)


Group


Frequency of In-
teraction in Reli-
gious Organizations


Health Health
Status Problem
(Camma=-.13,
= .11)

No Health
Problems
(Gamma=-.23,
S=.19)
g


Week ly
Monthly
Yearly
Never

Weekly
Monthly
Yearly
Neve r


Marital Status
Married Widowed
(N) (%) (N) (6)


23.4
16.3
12.8
12.8

13.5
11.3
5.7
4.2


33.1
19.7
16.5
9.4

11.8
3.9
4.7
1.0


Total 141 100.0 127 100.1

Partial Gamma=-.16 o =.09
g


Income Household
Income Less
Than $4500
(Gamma=-.12,
o =.14)
g

Household
Income Equal
To or Greater
Than $4500
(Gamma=-. 11,
a =.2)
g


Weekly
Monthly
Yearly
Never


Weekly
Monthly
Yearly
Never

Total


16.9
13.6
5. 9
7.6


22.0
13.6
11.0
9.3


36.5
20.2
18.3
5.8


118 99.9 104 100.0


Partial Gamma=-.12 o =.12
g










Additionally, controls for sex yield some interesting

findings. Bcrardo (1967), Pihlblad and Adams (1972) and

Townsend (1957) document the impact of sex in influencing

interaction in religious organizations, particularly among

the widowed. In this research controls for sex yield

findings consistent with these earlier reports. Among fe-

males the widowed are more likely to interact in religious

organizations more frequently than married women while

married males tend to participate to a greater degree in

church-related activities than widowers.


Research Hypothesis 10

Controlling for age, sex, race, education, health

status, and financial status, the rates of interaction in

formal organizations between the aged widowed and aged mar-

ried individuals will be less different than will the rates

without controls.

Table 7 indicates that the measure of association

between married and widowed respondents engaging in inter-

action in formal organizations is -.32. This score sug-

gests that the widowed tend to interact with greater fre-

quency in formal organizations than married individuals.

Of all five areas of social interaction investigated, this

has the largest gamma. However, the sample size for formal

organizations is drastically smaller than the other four

areas, suggesting that the total gamma computed for this

area is somewhat suspect. To elaborate, it is impossible

to determine if selective factors operated to systematically










TABLE 7. Measure of Association Between, and Frequency
and Percent of Married and Widowed Respondents
Engaging in Interaction in Formal Organizations
by Sex, Age, Race, Education, Health Status,
and Income


Frequency of In-
teraction in For-
mal Organizations


Total Group


Often
Generally
Seldom
Never


Total 67

Gamma=-. 32


Marital Status
Married Widowed
N) (%) (N) (%)


13.4
22.4
38.8
25.4


20.4
26.5
49.0
4.1


100.0 49 100.0

u =.13
9


Sex Male Often 3 4.5 0 0.0
((;amma=.4, Generally 8 11.9 0 0.0
S=.43) Seldom 13 19.4 1 2.0
g Never 15 22.4 1 2.0

Female Often 6 9.0 10 20.4
(Gamma=-.07, Generally 7 10.4 13 26.5
o =.2) Seldom 13 19.4 23 46.9
g Never 2 3.0 1 2.0

Total 67 100.0 49 99.8

Partial Gammia=.1 =.2
g


60 to
Years
(Gamma=
o
0 =
g

65 to
Years
(Gammaz
o =


65
Old
-.33,
.23)


75
Old
-.54,
.16)


Often
Generally
Seldom
Never


Often
Generally
Seldom
Never


6.0
11.9
22.4
10.4


6.0
7.5
16.4
13.4


6.1
4.1
14.3
0.0


12.2
18.4
20.4
0.0


Group










TABIE 7. (continued)


Frequency of In-
teraction in For-
mal Organizations


Marital Status
Married Widowed
(N) (%) (N) (%)


75 Years Old
and Older
(Gamma=.42,
S=.42)
g


Often
Generally
Seldom
Never


Total


68 100.0 49 100.0


Partial Gamma=-.32 o =.13
g


Race White
(Gamma=-.4,
a =.14)

Black
(Gama=- 01
o =.33)
g


Often
Generally
Seldom
Never
Often
Generally
Seldom
Never


Total


67 100.0 49 99.8


Partial Gamma=-.31


0 to 9th
Grade
(Gamma=-.44,
o =.2)

9th Grade to
High School
Graduate
(Gamma=-.38,
a =.25)

Some College,
Technical or
Trade Educa-
tion
(Camma=- .24,
a =.23)
&nj


Often
Generally
Seldom
Never

Often
Generally
Seldom
Never


Often
Generally
Seldom
Never

Total


Partial Gamma=-.35 a =.13
g


Group


1.5 1
3.0 2
0.0 7
1.5 2


2.0
4.1
14.3
4.1


11.9
16.4
28.4
22.4
1.5
6.0
10.4
3. 0


18.4
20.4
32.6
2.0
2.0
6.1
16.3
2.0


Edu -
ca-
tion


S=.13
g


1.5
9.1
9.1


13.6
1.5
9.1
13.6
6.1


10.6 5
4.5 5
15.2 8
6.1 0

100.0 48


6.2
8.3
18.8
2.1

4.2
8.3
14.6
0.0


10.4
10.4
16.7
0.0

100.0










TABLE 7. (continued)


Frequency of In- Marital Status
Group teraction in For- Married Widowed
mal Organizations (N) (%) (N) (%)


Health Health
Status Problems
(Gamma=-.18,
o =.17)

No Health
Problems
(Gamma= 52,
c =.23)
g


Often
Generally
Seldom
Never

Often
Generally
Seldom
Never

Total


67 100.0 49 99.9


Partial Gamma=-.3


Income Household
Income Less
Than $4500
(Camma=-.38,
o =.22)
g


Often
Generally
Seldom
Never


Household Often
Income Equal Gener
To or Great- Seldo
er Than Never
$4500
(Gamma=- .38, Total
S =.2)
gi


1.8
9.1
10.9
9. 1


10.9
12.7
27.3
18.2


ally
III


11.9
21.2
31.0
4.8



7.1
7.1
16.7
0.0


55 100.0 42 99.8


Partial Gamma--.38


0 =.15
g


10.4
14.9
14.9
16.4

3.0
7.5
23.9
9.0


12.2
22.4
40.8
2.0

8.2
4.1
8.2
2.0


a =.14
g










eliminate from the sample certain groups of respondents.

If this occurred, the total gamma computed for this area

is uncomparable with earlier similar measures.

Generally, controls for none of the six resource

variables result in decreased partial gamma scores. More-

over, comparisons of the gamma score for the entire group

with the gammas computed for each level of each of the

variables considered reveal that the partial gammas are

meaningless because the individual gamma values computed

for each control level are larger than the total gamma.

For this reason none of the factors considered can be

interpreted as viable exchange resources.


Summary

In conclusion, this chapter examines and compares the

interaction rates of aged widowed and aged married indi-

viduals in kin and friend groups as well as in religious

and formal organizations. The first five hypotheses,

designed to test if significant differences exist between

aged widowed and aged married individuals in terms of inter

action in these four broad areas, are not supported. More-

over, the second set of hypotheses, postulated to assess

the impact of variables assumed to act as exchange factors,

reveal that only income could be suggested as an exchange

resource which exerts a modest impact upon religious organic

national interaction. Concluding statements which analyze

and interpret these findings are presented in the following

chapter.
















CHAPTER V

SUMMARY AND CONCLUSIONS


This dissertation has examined the social interaction

of the aged in four broad areas: kin groups, friend groups,

formal organizations, and religious organizations. Inter-

action was measured in terms of frequency of participation

in these areas. Rates of social interaction of the aged

widowed were compared with rates of social interaction of

aged married individuals. The interaction rates were sub-

sequently compared again controlling For resources which

are considered, from the exchange conceptual framework, to

be both relevant and necessary to maintain social interaction.

Exchange resources include education, income, and health

status. Additionally, age, sex, and race were also control-

led because they are assumed to be indirectly indicative of

valued resources. The impact of many of these resources

upon the social realtionships of aged widowed and aged mar-

ried individuals has been suggested or tested in previous

research (Berardo, 1967; Blau, 1961; Bock, 1972; Bock and

Webber, 1972; Griffiths et al., 1971; Harvey and Bahr, 1974;

Lopata, 1973; Lowenthal, 1964; Pihlblad and Adams, 1972;

Rosow, 1967; and Townsend, 1957). However, this study, to

the author's knowledge, is the first application of the









exchange conceptual framework to an analysis of social

interaction rates of the elderly. The research findings

are summarized below.

Each of the first five hypotheses examined the social

interaction of the aged in nearby kin groups, far away kin

groups, friend groups, religious organizations, and formal

organizations. This set of hypotheses was designed to test

whether significant differences exist between aged widowed

and aged married individuals. More specifically, they

attempted to determine if the widowed,as previous research

has suggested, were more isolated than married individuals.

The second five hypotheses sought to determine if dif-

ferences in rates of interaction between the two groups of

aged individuals diminished upon controlling for variables

considered, from the exchange conceptual framework, to cir-

cumscribe social interaction. Any diminution in the

measure of association of the rates of interaction between

the two groups of aged persons may be interpreted as lending

support to the exchange framework. All hypotheses in the

second group focused on interaction in the same five areas

as the preceding group of hypotheses. However, this second

group of hypotheses served only to guide the research effort

without providing direct tests of significance because their

extension of controls involved comparisons between dependent

samples. Attention is now focused on individual analyses of

each of the hypotheses.









Research by Berardo (1967), Bock and Webber (1972),

and Townsend (1957) reports that aged married individuals

interact more with kin than do the aged widowed. Pihlblad

and Adams (1972) demonstrate this relationship for elderly

men but not elderly women. Hypotheses 1 and 2, developed

inductively from these earlier research efforts, posit the

same relationship. However, in order to hold constant the

impact physical distance has on circumscribing the social

interaction of aged persons, kin interaction was divided

into two categories: interaction with kin regarded by the

respondents to be nearby in geographical proximity, and

relatives perceived to be living far away. The first

hypothesis focused on nearby kin interaction. The second

pertained to far away kin. Neither hypothesis was supported.

The Kolmogorov-Smirnov values for the first and second

hypotheses were .33 and 1.93 respectively. This incon-

sistency with earlier research findings raises some ques-

tions which are not easily answered.

It is possible that the separation of kin into two

groups, in effect extending controls for geographic distance,

resulted in the inconsistent research findings. If this

is the case, statistical interaction, resulting from pe-

culiar multiplicative effects from combining nearby with

far away kin groups, could have resulted in significant

differences between aged widowed and aged married persons

in terms of rates of social interaction with kin when no

distinction by physical proximity was made, and no significant









differences between the two groups of elderly when it was.

Other potential explanations include possible problems in

the research designs of the studies involved or perhaps

the samples upon which the studies were based were drawn

from incomparable populations.

Findings from the third hypothesis revealed that the

Kolmogorov-Smirnov value of 7.44, which was significant

at the .05 level, did not support the research hypothesis

but rather indicated that the aged widowed appear to inter-

act with friends more than aged married individuals. This

finding supports Townsend's (1957) earlier research on the

aged but is inconsistent with results from investigations

by Berardo (1967) and Booth (1972). The finding from this

research as well as Townsend's study can perhaps be inter-

preted in the following way. The widowed, conditioned to

hut deprived of interaction with a spouse, are forced to

look outside the nuclear family for the satisfaction of

their social needs while married individuals have less

inclination to do so. Friend interaction appears to be a

viable outlet for meeting these needs. Moreover, assuming

the applicability of the exchange framework to this social

setting, friend interaction is actively sought out and

maintained by those with sufficient resources to make the

relationship rewarding to both. The aged widowed, unen-

cumbered with the obligations associated with married life,

have more free time which enables them to get the greatest

possible mileage from the resources they do possess.









The inconsistency among findings in this as well as

the two preceding hypotheses certainly indicates that fu-

ture research is necessary to clarify the matter. However,

earlier results by Blau (1961) provide information from

which a possible explanation for these contradictory find-

ings is suggested. She found that the isolation of the

elderly is greatly influenced by the social structural con-

text in which older individuals are situated. This is to

say, higher rates of social interaction are associated with

structural situations in which greater opportunities exist

for interaction with individuals of the same marital status.

Unfortunately, no controls for this factor were introduced

in this or earlier research. Thus, lack of comparability

of structural conditions among samples investigated could

obfuscate the relationships among the variables examined,

resulting in the observed inconsistency.

To explain further, samples drawn in Townsend's study

and the present research may have selected individuals from

structural settings composed of greater numbers of widowed

persons thereby providing the aged widowed with greater

opportunities to interact among themselves. On the other

hand, Berardo's (1967) and Booth's (1972) sample may have

consisted of disproportionate numbers of married individ-

uals thereby offering aged married greater possibilities

for interacting with one another. This interpretation is

consistent with the findings from the respective studies.

Berardo's research focused on aged people in a rural county









in Washington whereas Townsend's study examined 200 aged

individuals from Bethnal Green, London. In Berardo's rural

sample the elderly, particularly widowers, situated on iso-

lated farms, perhaps were not afforded the structural op-

portunities for interacting with those individuals in their

same age and sex grade as were the respondents in Townsend's

research who resided in a suburb of London. Furthermore,

this interpretation is supported by Booth who studied non-

institutionalized adults 45 years old and older residing

in Lincoln and Omaha, Nebraska. The sex ratio in the tar-

get population was .89, indicating a large number of single

persons. Assuming a disproportionate number of female

widowed, Booth's finding indicating widowers are much less

likely to have friends than widows lends additional support

to this explanation. Additionally, this problem of uncom-

parability could stem from either the strategy or implementa-

tion of the research designs.

The fourth hypothesis examined the interaction of the

aged in religious organizations. Earlier studies by Berardo

(1967) and Townsend (1957) suggest that widows are more

likely to interact in religious organizations than widowers

or married individuals. Widowers have been shown to be the

most isolated in this respect (Pihlblad and Adams, 1972).

However, in order to be consistent with the earlier hypoth-

eses and the overall design of the study, this hypothesis

did not distinguish by sex among the aged widowed. This

control was extended in the ninth hypothesis.




Full Text

PAGE 1

A COMPARISON OF RATES OF SOCIAL INTERACTION BETWEEN AGED WIDOWED AND AGED MARRIED INDIVIDUALS BY MARC PETROWSKY A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1975

PAGE 2

To Iris

PAGE 3

ACKNOWLEDGMENTS This study represents the culmination of my formal sociology training. Therefore it seems only fitting that I should recognize all individuals who played a role in this education. To this end, I would like to first thank my parents for providing me the incentive and support necessary to further my college education at both the graduate and undergraduate levels. furthermore, I want to recognize the instrumental role Dr. Richard Larson played in my early days as a graduate student when he served as both counselor and friend. Finally, I extend deep appreciation to Drs. Agresti, Gorman, Leslie, Warheit and especially Dr. Berardo, my chairman, for the intellectual stimulation and guidance they provided throughout my doctoral program. Their assistance in helping me conceptualize and implement this project was crucial and more than generous, but I thank them even more for trying to cultivate in me the sociological imagination. A final note of thanks is in order. Data upon which this research is based came from a large scale sample survey entitled Evaluating Southern Mental Health Needs and Services, MH15900. Dr. Warheit graciously provided me access to these data.

PAGE 4

TABLE OF CONTENTS Page ACKNOWLEDGMENTS iii LIST OF TABLES v ABSTRACT vi CHAPTER I INTRODUCTION 1 CHAPTER II REVIEW OF LITERATURE 10 CHAPTER III METHODOLOGY 31 CHAPTER IV FINDINGS 58 CHAPTER V SUMMARY AND CONCLUSIONS 88 BIBLIOGRAPHY 112 BIOGRAPHICAL SKETCH 119 IV

PAGE 5

LIST OF TABLES TABLE 1 TABLE TABLE 3 TABLE 4 TABLE 5 TABLE 6 TABLE 7 Frequency and Percent of Respondents by Levels of Marital Status, Sex, Age, Race, Education, Health Status, and Income Comparisons Between Married and Widowed Individuals in Terms of Frequency of Interaction in Nearby Kin Groups, Far Away Kin Croups, Friend Croups, Religious Organizations, and Formal Organizations Page 53 61 Measure of Association Between, and Frequency and Percent of Married and Widowed Respondents Engaging in Interaction with Nearby Kin by Sex, Age, Race, Education, Health Status, and Income 6 7 Measure of Association Between, and Frequency and Percent of Married and Widowed Respondents Engaging in Interaction with Far Away Kin by Sex, Age, Race, Education, Health Status, and Income . . . Measure of Association Between, and Frequency and Percent of Married and Widowed Respondents Engaging in Interaction with Friends by Sex, Age, Race, Education, Ileal tli Status, and Income 71 76 Measure of Association Between, and Frequency and Percent of Married and Widowed Respondents Engaging in Interaction in Religious Organizations by Sex, Age, Race, Education, Health Status, and Income . . 80 Measure of Association Between, and Frequency and Percent of Married and Widowed Respondents Engaging in Interaction in Formal Organizations by Sex, Age, Race, Education, Health Status, and Income 84

PAGE 6

Abstract of Dissertation Presented to the Graduate Council in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy A COMPARISON OP RATES OF SOCIAL INTERACTION BETWEEN AGED WIDOWED AND AGED MARRIED INDIVIDUALS By Marc Petrowsky June, 197 5 Chairman: Dr. Felix M. Berardo Major Department: Sociology This research examines the social interaction of the aged from the perspective of the exchange conceptual framework. Interaction is measured in terms of frequency of participation in kin groups, friend groups, formal organizations, and religious organizations. Rates of social interaction are compared between aged widowed and aged married individuals in each of the areas of social interaction investigated both before and after controlling for resources which are considered, from the exchange conceptual framework, to be necessary to maintain social interaction. Resources examined in this research include age, sex, race, education, income, and health status. A systematic sample was drawn from a sample frame developed from electrical utility company residential household listings in Alachua County, Florida. Structured interviews yielded data on 128 aged widowed and 145 aged married individuals sixty years old or older. vi

PAGE 7

The findings indicate no significant differences in rates of interaction in kin and friend groups between aged widowed and aged married persons. Additionally, the aged widowed interact in religious and formal organizations more than aged married individuals. Controls for resources provide virtually no support for an interpretation of social interaction of the aged from the exchange conceptual framework.

PAGE 8

CHAPTER I INTRODUCTION This research examines the social interaction of the aged in our society. Interaction is measured in terms of frequency of participation in .kin groups, friend groups, formal organizations, and religious organizations, a subgroup of formal organizations but one which has been shown to be important in assessing the interaction of the aged (Bcrardo, 1967; and Townsend, 1957). Rates of social interaction are compared between aged widowed and aged married individuals. The interaction rates are subsequently compared again; however the second comparisons are made controlling for resources which are considered, from the exchange conceptual framework, to be both relevant and necessary to maintain social interaction. Resources consist of both material and nonmaterial goods (Homans , 1958:597) or activities and sentiments (Homans, 1961:34) which are considered to be valued exchangeable commodities. For this study resources include education, income, and health status. Additionally, age, sex, and race are also controlled because they are assumed to be indirectly indicative of valued resources.

PAGE 9

Just ifi cat ion f or the R esea rch The area of investigation was chosen primarily for two reasons. First, the number of the aged is rapidly increasing both in absolute numbers and as a proportion of the entire population of American society. Because of this growth the aged as a group, and the widowed as an important sub-category of that group, have become an important area for sociological investigation (Berardo, 1968). Second, research of the aged is extremely limited. This lias two interrelated and unfortunate consequences. First, pure scientific knowledge of the area is limited and relatively unorganized. This inadequacy, represent ing a gap in knowledge of later stages of the family life-cycle, is regrettable in and of itself. Second, because of the insufficiencies of scientific knowledge in this area there is a pitifully inadequate pool of research findings available which can be drawn on to guide private or governmental agencies which might attempt to promote legislation, provide funds, or in other ways attempt to alleviate the unfortunate positions in which many of the aged find themselves. By organizing research findings extant in this area and analyzing available data on aged widowed and married individuals, this research will hopefully contribute to the literature in the area. Further elaboration of each of these reasons follows. For 150 years the population of the United States was an aging population, as measured in terms of median age, until the rising birth rates of the post-World War II decades reversed this trend. This reversal notwithstanding, the

PAGE 10

5 absolute number oi : the aged as well as the proportion of the aged in our society has increased steadily if not rapidly. There were just over 3.1 million aged persons in the United States at the turn of the century, by 1950 there were 12.3 million and the 197 census revealed over twenty million citizens 65 or older (United States Census, 19 7 0). The proportionate growth of the aged population parallels their increase in absolute numbers. In 1900 4.1 percent of the population was 65 years of age or older. In fifty years the proportion of the aged in our society had climbed to 8.1 percent and in 1970 the elderly comprised almost 10 percent of the total population (Statistical Abstract of the United States, 1972:30). This increase in the number of the aged has come about through a complex variety of factors such as the "number of births in appropriate earlier periods, declining mortality, and immigration" (Sheldon, 1960:40). These factors, as dependent variables, resulted from the matrix of advancing science and technologism which grew at an unprecedented rate from the turn of the century until the present. Moreover, among the aged, widowhood, according to Berardo (1968:191), is "rapidly becoming a major phenomenon of American society." In 1970 the aged population included six million widows and 1.5 million widowers, approximately 40 percent of the total aged population (McKain, 1972:61). These figures reveal that, among the aged widowed, there

PAGE 11

are over four times as many women as men, and an analysis of the trends presented by census data for the past sixty years suggests a continued increase in the gap between widows and widowers. For example, in 1910 there were 1.47 million widowers who comprised 4.4 percent of the entire population. Forty years later there were 2.3 million widowers who constituted 4.2 percent of the population. By 1971 the number of widowers has decreased to two million representing only 3.1 percent of the population. The number of widows, in contrast, grew steadily from 3.18 million in 1910 to 9.78 million in 1971. They also registered a steady proportionate increase from 10.3 percent in 1910 to 13.8 percent in 1971 (Statistical Abstract of the United States, 1972:30). The main factors contributing to the excess of widows over widowers include the higher mortality rates among males, the higher remarriage rates for men, and the fact that men usually marry women younger than themselves (Jacobson, 1959:25-27). The second justification for this study is founded upon the paucity of research in an area which has pressing need for scientific investigation to both illuminate the substantive area and provide findings which can be used to direct, social action in attempts to alleviate some of the disadvantaged conditions confronting many of the aged widowed. For example, research suggests that the widowed are likely to be economically impoverished (Lopata, 1969; Marris, 1958; Statistical Abstract of the United States, 1969).

PAGE 12

have higher death rates than married individuals (Gove, 1972 and Townsend, 1957), exhibit higher suicide rates than married individuals (Berardo, 1967; Bock, 1972; and Gove, 1972), have a higher rate of mental disorders than married individuals (Berardo, 1967), experience deterioration of health status (Berardo, 1967), and indicate that they are lonely (Berardo, 1967; and Lopata, 1969). These findings indicate that widowhood, although inevitable for a significant proportion of the entire population, offers something less than an enviable status for many of the aged in our society. This study will hopefully shed some light on the subject matter at hand by systematically organizing research findings pertaining to the aged widowed and married individuals in addition to examining statistical data on 273 aged widowed and married persons. History of Scientific Interest in Aging Systematic studies of the aged began only recently and were first initiated by biologists who were interested in time-related changes of cells, tissues, and physiological mechanisms. Interest in the biological aspects of aging in the United States was signaled by the establishment of the American Research Club on Aging in 1939. In 1945 the Gerontological Society, Inc., was founded. Both provided funds as well as a focal point for generating interest which further stimulated research in the area. This section draws heavily from Tibbitts (1960)

PAGE 13

Psychological research on aging followed a similar pattern. In the early 1930's W. R. Miles conducted the Stanford Later Maturity Research Project which was the first systematic psychological study of aging. The furtherance of psychological research on aging mushroomed after the American Psychological Association gave formal recognition to research in this area in 1946. Research on the social and economic aspects of aging began in the 1920 ' s and was primarily concerned with providing solutions to practical problems facing the aged. This interest was precipitated primarily by three factors: increased visibility of the aged, separation of the elderly from the workforce, and the American value emphasizing individual happiness and well-being (Pollak, 1948). The increased visibility of the aged resulted from technologism and industrialization which had a tremendous impact on fertility, mortality, and migration rates. One of the consequences was the doubling in size of the aged population from 1900 to 1930 and again before mid-century (Tibbitts, 1960). Additionally, with the economic emphasis shifting from agrarianism to industrialism, the work and family roles of the elderly underwent dramatic change. Their usefulness as repositories of traditional farming wisdom and as owners of the land was undermined by the industrial revolution which created a viable alternative to a land-based economy. The end result was the generation of

PAGE 14

several social problems related to the financial, social, and physical adaptation of the aged to an industrialized, urbanized milieu. Research in the 1930' s and 1940 's documented the adjustment problems of older citizens. In 1943, under the leadership of Ernest W. Burgess, The Committee on Social Adjustment in Old Age was established. During the decades of the 1940' s and 1950 's various universities followed the lead of the University of Chicago and the University of Michigan in stimulating research and interest in the social ramifications of maturation and old age. In 1950 a significant portion of the program of the National Conference on Aging was devoted to social and economic aspects of aging. The Gerontological Society, in 1952, distinguished a separate Division of Psychology and Social Science. In 1956 the Inter-University Training Institute in Social Gerontology was established and supported by the National Institutes of Health. Additionally, the International Association of Gerontology established a social science research division in that year. During this same period family sociologists, following the path paved earlier by Sorokin, Zimmerman, and Galpin (19.31), were also drawing more attention to the aged in family settings by focusing on stages in the family life cycle (Duvall, 1957; Click, 1957; Rodgers, 1964; and Rodgers and Hill, 1964). It is hoped that this research will contribute to the rapidly expanding knowledge in this area.

PAGE 15

Rationale of the Prese nt Stud y This study is organized into two interrelated but analytically distinct parts. The initial portion of the work integrates findings gleaned from research pertaining to isolation of the aged widowed and married individuals. The mechanism for integrating, as presented in Zetterberg (1965), is an axiomatic approach to theory building. Research findings are organized into empirical or ordinary propositions. Empirical propositions with similar linkages are combined and abstracted to higher levels of explanation. These then become derived propositions which are expanded to form theoretical propositions which have high information value. This task of theory construction is conducted within the exchange conceptual framework, thus linking the theory to a more general theoretical model. The second portion of the study provides for an empirical test of the constructed theory. Hypotheses compare differences between the interaction rates of the aged widowed and the aged married in kinship groups, friend groups, formal organizations, and religious organizations controlling for age, sex, marital status, education, income and health status. The research examines the possibility of social interaction among the aged being contingent upon control over valued resources . This study is organized into four additional chapters. The second chapter reviews the literature in the area, both the theoretical literature as well as research findings. The

PAGE 16

third chapter presents the hypotheses and research design The fourth chapter discusses the research findings, while the summary and conclusions from the study are presented in the final chapter.

PAGE 17

CHAPTER II REVIEW OP LITERATURE The review of literature for this study is divided into two sections. Literature relevant to the exchange conceptual framework is presented in the first part. A survey and organization of research findings pertaining to the social isolation of the aged married and the aged widowed follows. Review of the Theoretical Literat ure This section on the theoretical literature accomplishes two purposes. It first discusses the history and applicability of theoretical conceptual frameworks. Secondly, it provides an analysis of the exchange conceptual framework. Con cept ual Fra me wor k Zetterberg (1965) suggests that sociology contains humanistic as well as scientistic underpinnings. The scientific tradition encompasses two types of what is commonly labeled theory. The first type, taxonomies, consists of descriptive statements which do not offer any explanations. The second type, referred to by Zetterberg as theoretical sociology, consists of systematically interrelated propositions which are derived from and inspire research. The latter provides the bulwark of the axiomatic approach to theory building, the approach utilized in this research. 10

PAGE 18

n The former is identified by Hill and Hansen (1960) and Nye and Berardo (1966) as a major ingredient of conceptual frameworks . Hill and Hansen (1960) , noting the paucity of family theory, challenged family sociologists to interrelate research findings into sets of research propositions guided by what Zetterberg referred to as taxonomies. These taxonomies, or frames of reference, play a pivotal role in building sociological theory, posit Hill and Hansen, by providing researchers with groups of "interrelated but not necessarily interdefined concepts generally applicable to the arena of marriage and the family" (Hill and Hansen, 1960:300). Additionally, conceptual frameworks, for the purposes of this research, are considered to consist of time and space dimensions (Hill and Hansen, 1960) as well as specification of assumptions underlying the framework (Nye and Berardo, 1966) The Hill and Hansen article was followed by several articles which suggested a multiplicity of new and existing During the course of the pre any reference to the word "theory" mean conceptual framework unless o the two concepts are used intercha lytical distinction can easily be theory is a set of interrelated pr to explain some phenomenon. A com with the components of conceptual above reveals that theories can be more specific explanations of the ceptual frameworks provide broad g theories are developed. sentation of this research is explicitly assumed to therwise indicated. Although ngeably, an important anamade between the two. A opositions which purports parison of this definition frameworks articulated understood to be much empirical world while conuidelines within which

PAGE 19

12 frameworks to guide family research. In 1969 Edwards suggested that the exchange conceptual framework would provide a powerful theoretical model for analyzing family behavior. Exchang e Conceptua 1 Framework Exchange theory is the conceptual framework used in this research to guide the interpretation of the analysis of rates of interaction of aged widowed and married individuals. Review of the theoretical literature pertaining to the exchange framework follows. Although it lias been most succinctly articulated only recently, many of the underlying assumptions and concepts of exchange theory can be found in the classical literature. Selected works from Marx, Comte, and Simmel reveal attempts by these early masters to utilize, either directly or indirectly, aspects of exchange theory to explain social processes. Simmel, perhaps one of the most forthright proponents of exchange as a characteristic of human action, wrote that "exchange is the ob j ectif ication of human interaction" (Simmel, 1950:388). More recently aspects of the exchange framework have been applied by social psychologists, social philosophers, and especially anthropologists, to guide research activity and to offer explanations of various facets of social phenomena. Anthropologists including Malinowski (1932), Mauss (1954), and Thurnwald (1932) have all documented the process of exchange in their observations of pre! iterate tribes and societies. Additionally, the importance of the norm of

PAGE 20

13 reciprocity, an integral concept in exchange theory, has been portrayed by many contemporary social theorists such as Becker (1956), Gouldner (1960), and Levi-Strauss (1957). Perhaps the widest application of exchange theory, albeit circuitous in nature, has been by sociologists in family research. Dating and courtship behavior (McCall, 19 66; Rogers and Havens, 1960; and Waller, 1937) as well as the area of marital power (Blood and Hamblin, 1958; Blood and Wolfe, 1960; llecr, 1958; and Ileer, 1963) has been approached from this perspective. The preceding theorists and researchers have dealt with exchange theory in primarily an unsystematic fashion. However, during a five-year period beginning in 1959, three monographs were published which fully set forth the concepts, underlying assumptions, and in one case, propositions of exchange theory (Blau, 1964; Homans, 1961; and Thibaut and Kclley, 1959). The discussion of the exchange conceptual framework closely follows these three books. The underpinnings of exchange theory are found in both behavioral psychology and elementary economics. Combined, central ideas from these two disciplines yield a theory which attempts to analyze social interaction as occurring through a process of exchange of resources by actors occupying positions in the social structure (Homans, 1958:597). Two assumptions provide the foundation for this exchange process. first, behavior is goal oriented and achieved through interaction with others. Second, actors, in pursuit of various goals, attempt to adapt means to further

PAGE 21

14 the achievement of these ends (Blau, 1964:5). Consequences of this goal oriented behavior include rewards as well as costs . Rewards Rewards, or resources as they are sometimes called, consist of both material and nonmaterial goods (Homans , 1958:597) or activities and sentiments (Homans, 1961:34). There are two basic types of rewards, intrinsic and extrinsic Intrinsic rewards result from types of social interaction where the social relationship itself is considered to be a reward. Social situations which yield extrinsic rewards for actors are those in which the social relationship is a mechanism through which these rewards are realized (Blau, 1964:58-59). The more highly valued a reward, the more often individuals will extend effort to obtain that resource. The value of resources, be they objective (activi ties or material goods) or subjective (sentiments or nonmaterial goods) in nature, is determined by four variables: group values, unique individual experiences, deprivation, and satiation. Values are a social product of group existence. Therefore, individuals who belong to the same group or the same kinds of groups are more likely to value the same resources or types of resources (Homans, 1961:46). However, this similarity of group experience is circumscribed by unique individual experiences which are likely to channel the

PAGE 22

15 development of values so that they do not always exactly reflect other group members' values. Additionally, the individual's state of deprivation and satiation affects the value of resources (Homans, 1961:47-49). Resources highly valued but in short supply become even more highly valued while resources constantly supplied decline in value. The variability and subjectivity of the determination of values leads to problems of measurement and analysis. Homans suggests that "our knowledge of values will always be imperfect, and the predictions we make from it will be gross and statistical, based on a few obvious similarities and differences, bound to go wrong in detail" (1961 :46-47) . Costs Costs refer to "factors that operate to inhibit or deter the performance of a sequence of behavior" (Thibaut and Kelley, 1959:12). Additionally, Homans maintains that the concept of cost, or aversive conditions, should include recognition of alternative rewards forgone by investing resources in pursuit of one particular goal rather than a potential alternative (Homans, 1961:58). Profit Profit is defined as "the difference between the value of the reward a man gets by emitting a particular unit-activity, forgone in emitting the first" (Homans, 1961:63). It follows that the more valuable the reward stemming from a

PAGE 23

16 particular activity, the more often an actor will emit it, while the more costly it proves to be, the less he will emit it. The preceding reduces to the simple formula of rewards minus costs, equals profit. It is assumed that continued exchange is contingent upon both actors realizing a profit. Indeed, "the open secret of human exchange is to give the other man behavior that is more valuable to him than it is costly to you and to get from him behavior that is more valuable to you than it is costly to him" (Romans, 1961 :62) . The actor's perception of his profit is important because individuals will typically terminate social relationships which do not appear profitable. Thibaut and Kelley have articulated two standards upon which these outcomes of interaction are evaluated. The comparison level is "the standard against which the member evaluates the attractiveness of the relationship or how satisfactory it is." The second, the comparison level of alternatives, refers to "the standard the member uses in deciding whether to remain in or to leave the relationship" (Thibaut and Kelley, 1959:21). The former is important because it determines for actors whether profits gleaned from a relationship are worth the costs. If an actor feels the relationship is unrewarding he has the option of seeking alternative relationships or dissolving the relationship and going without the association with its concomitant rewards and costs. The comparison level of alternatives is important because profits are not

PAGE 24

17 only evaluated in terms of rewards minus costs but "common norms develop in societies that stipulate fair rates of exchange between these benefits" (Blau, 1964:155). Homans 's concept of "distributive justice" also expresses this not ion . Propositions The preceding analysis of the concepts and assumptions underlying the exchange framework have been succinctly organized by Homans (1961:53-55) into the four following propositions. Proposition 1 If in the past the occurrence of a particular stimulus situation has been the occasion on which a man's activity has been rewarded, then the more similar the present stimulus situation is to the past one, the more likely he is to emit the activity, or some similar activity, now. Proposition 2 The more often within a given period of time a man's activity rewards the activity of another, the more often the other will emit the activity. Proposition 3 The more valuable to a man a unit of the activity another gives him, the more often he will emit activity rewarded by the activity of the other. Proposition 4 The more often a man has in the recent past received a rewarding activity from another, the less valuable any further unit of that activity becomes to him. Hxchange theory makes no assumptions of explicit rational behavior. It is only concerned with the immediate alternatives open to the actor, as he sees them, and his resulting choice. As Blau suggests "the only assumption made is that human beings choose between alternative potential

PAGE 25

associates or courses of action by evaluating the experience's or expected experiences with each in terms of a preference ranking and then selecting the best alternative" (1964:18). A totally rational situation, as found in game theory, does not exist because individuals do not have complete information on all conceivable alternatives, their preferences do not remain constant, and the actors do not pursue only one goal, to the exclusion of all others (Blau, 1964:18). The foregoing analysis examines the general concepts, assumptions, and propositions underlying exchange theory. Edwards (1969) has suggested some minor modifications in this approach to make it more amenable to the special conditions typical of interaction among family members. Because this research focuses specifically on interaction of aged widowed and married individuals these modifications are given consideration. Edwards states that in an analysis of family interaction it is difficult to conceptualize what is exchanged. This problem stems in part from the informality of family associations plus the isolation of familial interaction, both of which tend to obfuscate the resources exchanged (Edwards, 1969:520). Additionally, exchange in family situations can be based upon either intrinsic or extrinsic rewards. This research does not attempt to measure intrinsic rewards. Moreover, it is difficult, to fully specify exchange or alternatives for extrinsic rewards. However, with all its problems of conceptualization and measurement, exchange

PAGE 26

1 9 theory is assumed to be a particularly valuable approach for explaining behavioral outcomes of the aged widowed. Review of the Research Literature The review of the research literature is divided into three sections. The first part presents a series of empirical generalizations abstracted from research relevant to the social isolation of aged widowed and aged married individuals. The second section presents derived propositions obtained by abstracting or combining empirical propositions. The final part presents a theoretical proposition which is acquired by abstracting and combining derived propositions to form a more powerful information statement. Empirical Propositions The first step in relating variates, according to Zetterberg's axiomatic approach to theory building, consists of the formulation of empirical propositions. These propositions depict relationships existing between two or more variables. The variables are designated as either determinants or results. Determinant or independent variables cause results or dependent variables. Within each proposition the determinants are related to the results by a variety of linkages. Zetterberg (1965) posits that the linkages in a causal relationship can be described in terms of five pairs of descriptive attributes. A relationship may be either reversible (if X, then Y; and if Y, then X) or irreversible (if

PAGE 27

20 X, then Y; but if Y, then no conclusion about X) , deterministic (if X, then always Y) or stochastic (if X, then probably Y) , sequential (if X, then later Y) or coextensive (if X, then also Y) , sufficient (if X, then Y, regardless of anything else) or contingent (if X, then Y, but only if Z), and necessary (if X, and only if X, then Y) or substitutable (if X, then Y; but if Z, then also Y) . Later in the research empirical propositions with similar linkages are combined and further abstracted to yield derived propositions . The following empirical propositions were abstracted from research relevant to the social interaction of aged widowed and aged married individuals. (1) The widowed, particularly males, are less likely to interact in formal organizations than are married individuals. Berardo (1967), Bock (1972), and Bock and Webber (1972) all posit that, in terms of participation in formal organizations, widowers are dramatically less likely to interact than either widows or married individuals while the differences between the latter two are minimal. Furthermore, Pihlblad and Adams (1972) depict no differences in participation in formal organizations between married and widowed women but not iceabl e differences between women and men, especially widowed men. However, considered collectively, the widowed according to Harvey and Bahr (1974) are less likely to interact in formal organizations than aged married

PAGE 28

21 individuals. Moreover, Harvey and Bahr proceed to suggest that differences in interaction between widowed and married aged individuals are a function of age and income variables and not widowhood status. (2) Widows are more likely to interact in religious organizations than are either widowers or married individuals. Berardo (1967) and Townsend (1957) support this proposition. Pihlblad and Adams (1972) suggest that widowhood decreases participation in religious organizations for men but not for women. Berardo states that the discrepancy between widows and widowers results from the fact that women have traditionally maintained the liaison between the family and the religious institution. Moreover, this role is not significantly diminished with the death of the husband. Additionally, widows can interact in religious institutions on the same level as married individuals because they are not discriminated against because of their single status. (3) The widowed are less likely to interact with friends than are married individuals. Berardo (1967) reveals that the widowed are more isolated in terms of friendship interaction than married individuals. Moreover, Booth (1972) holds that married individuals express more close friend relationships than the widowed. Furthermore, they both note that the discrepancy between aged widowers and aged married individuals, in terms of absolute number as well as in frequency of interaction with

PAGE 29

22 friends, is particularly great. The impact of sex in circumscribing friend interaction is provided additional support by Pihlblad and Adams (1972) who suggest that widowhood decreases friend interaction for men but not women. However, Townsend (1957) indicates that widows and widowers visit their neighbors more than married individuals. Because visiting neighbors can be construed as an index of friendship it would appear that Townsend and Berardo suggest conflicting findings. However, this discrepancy might be accounted for by the finding of Berardo (1967) that "widowhood creates greater strains for rural residents." To elaborate, Berardo' s study focuses on aged people in a rural county in Washington whereas Townsend ' s study focuses on 200 aged people from Bethnal Green, London. Blau (1961) suggests that the status of isolation depends on the structural context of older people. That is to say, if an individual's marital status is married and all of his peers are widowed then he will perceive himself as isolated because he is deviant vis-a-vis the marital status structure of the community. To return to the original problem, the widowed in Townsend 's study might be less isolated than married individuals because they live in an urban area with greater potential interaction with those of the same marital status. On the other hand, the married individuals in Berardo 's study could be less isolated because of the diminished potential for interaction with others of similar status that exists for the widowed in rural areas.

PAGE 30

2 3 (4) The widowed are less likely to interact with their children than are married individual s . Berardo (1967) and Townsend (1957) support this finding. Interaction subsumes visits (Townsend), companionship, transportation, advice, money, and extension of gifts (Berardo). All these indices of interaction have been used by sociologists to examine kinship interaction (Sussman and Burchinal, 1962a). Although Berardo posits that widows have high rates of contact with their children he proceeds to explain this in terms of higher incidences of widows living with their children. It is assumed that the opportunity to live with one's children is diminished for married individuals, thereby distorting the compared rates of interaction between aged survivors and married individuals. Therefore, the widowed, considered collectively, have less interaction with their children than married individuals have with theirs. However, findings from Adams' (1968) middle class sample yield some contradictory evidence. He determined that widows see their daughters more frequently than married women. This contradiction suggests that social class may play an important role in social interaction of the aged. This relationship is examined in Empirical Proposition 12. (5) The widowed are less likely to interact with kin than are married individuals. Berardo (1967) , Bock and Webber (1972) , and Townsend (1957) suggest that kin interaction declines with the death of a spouse. This relationship is also demonstrated for

PAGE 31

24 elderly men hut not elderly women in research by Pihlblad and Adams (1972) . (6) The widowed are more likely to commit suicide than are married individuals. Berardo (1968), Bock (1972), Bock and Webber (1972), Dublin (1963), and Durkheim (1951) all document the increased likelihood of suicide among the widowed. (7) The widowed are more likely to be isolated than married individuals. Research findings from Berardo (1967) , Bock and Webber (1972), Harvey (1973), and Townsend (1957) all support this proposition. (8) For widows, the higher the educational status, the greater the participation in fo rma 1 organizations . Lopata (1973) suggests this proposition. (9) For widows, the higher the financial status, the greater the participation in formal organizations. Lopata (1973) supports this relationship. Moreover, Harvey and Bahr (1974) suggest that organizational affilia tion for the widowed is circumscribed by socioeconomic status more than widowhood status. (10) For widows, the higher the financial status, the greater the friendship satisfaction. Lopata (1973) indicates that widows who develop satisfactory friendship relationships have higher education and a comfortable income. Blau (1969) and Griffiths et al.(1971) lend

PAGE 32

2 5 further support to this proposition by suggesting a positive relationship between socioeconomic status and friendship interaction among the aged. Additionally, Rosow (1967) suggests that the middle class elderly indicate a disproportionate number of friends when compared with elderly working class individuals. (11) For widows, a direct relationship exists between educational status and friendship sati s fact i on . Again Lopata (197 5) states this proposition. (12) The greater the socioeconomic status of the aged, the less likely is social i sol at ion . Griffiths et al. (1971), Lowenthal (1964), and Townsend (1957) posit that a disproportionate number of aged isolates tend to be from lower socioeconomic strata. It should be noted that this empirical proposition includes, but is not limited to, the widowed. This is the only proposition to examine the relationship between two variables for the aged in general. The inclusion of this proposition is necessary however, to depict the relationship between isolation and one aspect of available resources, socioeconomic status. Later, in forming the theoretical proposition, the role of this proposition in establishing the relationship between resources and isolation will become evident. De rived Pro pos it ions Derived propositions, which have greater information value than empirical propositions, can be obtained by

PAGE 33

2 6 abstracting or combining empirical propositions. Similar causal linkages are necessary in order to subsume two or more empirical propositions under a more powerful derived statement. The preceding empirical propositions were derived from research findings. These propositions will now be combined to yield more general explanatory statements referred to as derived propositions. (A) The widowed are less likely to interact in formal organizations than are married individuals. Empirical Propositions 1 and 2 combine to form Derived Proposition A. This combination is possible because the causal linkages of both propositions are irreversible, stochastic, coextensive, contingent, and substitutable. The second empirical proposition is subsumed under the first because churches are considered formal organizations (Leslie et al . , 1973:307). Moreover, membership and participation constitute indices of interaction in these organ! zations . Although it appears that the two contributing propositions are inconsistent, it must be noted that churches, representing only one of the many types of formal organizations, provide a social milieu wherein the widowed are not discriminated against because of their single status (Berardo, 1967). Moreover, because of their traditional link with the church, widows have the opportunity to use it as an arena in which they can exchange resources for social interaction.

PAGE 34

27 Resources of widows, such as cooking, cleaning, and sewing abilities, are highly valued in many church activities like bazaars, covered dish dinners, and sewing circles. Moreover, it is presumed widows have more time to invest in church activities than do married individuals (Berardo, 1967) . Because o£ the aforementioned reasons, it is assumed that the church, in relation to other formal organizations, offers unusual opportunities to the aged widow for engaging in social interaction. However, considering all types of formal organizations collectively, available evidence (Berardo, 1967; Bock, 1972; Bock and Webber, 1972; Harvey and Bahr, 1974; Pihlblad and Adams, 1972; and Townsend, 1957) holds that the widowed as a group are more isolated than married individuals in these organizations. (B) The widowed are less likely to interact with friends than are married individuals. Empirical Proposition 3, in its present form, yields Derived Proposition B. (C) The widowed are less likely to interact with kin than are married individuals. Empirical Propositions 4 and 5 combine to yield Derived Proposition C. This combination is possible because the causal linkages of both propositions are irreversible, stochastic, coextensive, and substitutable . Because children are members of the kin group it is logical, to subsume Empirical Proposition 4 under Empirical Proposition 5.

PAGE 35

28 (D) The widowed are more likely to be isolated than married individuals. Empirical Propositions 6 and 7 combine to yield Derived Proposition D. This combination is possible because the causal linkages of both propositions are irreversible, stochastic, coextensive, contingent, and substitutable . Bock and Webber (1972) state that high suicide rates of the aged widowed are functions of greater social isolation. It therefore seems logical that suicide rates can serve as an index of isolation, yielding greater support for the derived proposition . (E) For widows, the greater the resources, the more likely the participation in formal organizations. Empirical Propositions 8 and 9 combine to yield Derived Proposition E. This combination is possible because the causal linkages of both propositions are irreversible, stochastic, coextensive, contingent, and substitutable. Edwards (1969) states that, although it depends on the social setting, an individual's knowledge is a well defined resource. lor this research it is assumed that knowledge and education are positively related. Moreover, McCall (1966) and Streib (1972) hold that money, which is equated with financial status, also constitutes a resource. (F) For widows, the greater the resources, the more likely friendship satisfaction will result. Empirical Propositions 10 and 11 combine to yield Derived Proposition F. This combination is possible because

PAGE 36

29 the causal linkages of both propositions are irreversible, stochastic, coextensive, contingent, and subst itutable . The rationale for dealing with the transformation of educational and financial status to resources is the same as in Derived Proposition E. (G) The greater the resources of the aged, the less likely is social isolation. Empirical Proposition 12 is abstracted to yield Derived Proposition G. Socioeconomic status, based upon the rationale presented in the two preceding derived propositions, is considered a resource. Theoret i cal Pro positions Theoretical propositions are developed by combining and abstracting derived propositions to form more powerful information statements. They represent the final stage in axiomatic theory construction. (I) The widowed are more likely to be socially isolated than married individuals because isolation is a function of insufficient resources. Derived Propositions A, B, C, and D combine to yield the following statement. The widowed are more likely to be isolated than married individuals. Derived Propositions A, B, and G posit that, in terms of interaction in formal organizations, kin groups, and friend groups, the widowed are less likely to interact than are married individuals. These three combined derived propositions can then be

PAGE 37

30 subsumed under Derived Proposition D to yield the first part of Theoretical Proposition I. Derived Propositions E, F, and G combine to yield the following statement. The greater the available resources, the less likely is social isolation. Derived Propositions E and F, which pertain to participation in formal organizations and friendship interaction, are subsumed under Derived Proposition G to yield the second part of Theoretical Proposition I. Conclusion This concludes the review of all relevant literature. The hypotheses tested in the present study are generated from the preceding research, under the guidance of the exchange conceptual framework. Attention in the third chapter is focused on the methodology of the research which specifies the procedure whereby the derived theoretical proposition is subject to empirical verification.

PAGE 38

CHAPTER III METHODOLOGY This study examines the social isolation of aged widowed and aged married individuals. Social isolation is operationally defined as diminution of social interaction. Before comparisons of social interaction rates between aged widowed and aged married individuals can be examined, it is necessary to define the terms and concepts used in the research, to state the various hypotheses, and to examine their relationship to the previous research. Data collection procedures are discussed. The statistical procedures used in describing the data are also presented. Finally, a summary description of the data is provided. Conc ep ts The three concepts which need to be operationally defined are widowhood, social isolation, and social resources Widowhood This study focuses on the social isolation of aged widowed as compared with aged married individuals. The aged widowed are defined as those individuals 60 years old or older who have lost a spouse through death and have not since remarried. The widowed are further distinguished according to sex. Widows are females 60 years old or older 31

PAGE 39

32 who are not currently married because of the death of a spouse. Widowers are males 60 years old or older who are not currently married because of the death of a spouse. Social I sol a t ion Studies of social isolation have basically defined it as attenuation or severance of interpersonal interactions or relationships (Paris, 1934:155-164; Jaco, 1954:567-577; Kohn and Clausen, 1955:265-273; Tec and Granick, 1959:226232; and Townsend, 1957:169). However, a difficulty arises when attempting to specify how much attention to social interaction warrants the application of a label of "social isolation" to an individual or group (Clausen and Kohn, 1954:140-151) . In the preceding studies scales as well as various operational definitions of isolation have been used to demarcate the state of social isolation among individuals. For this research social isolation refers to diminished social interaction. To this end, social interaction is measured in terms of frequency of contact with friends, kin living nearby, kin living far away, formal organizations, and religious organizations. Comparisons of rates of social interaction, in all five areas, between the aged widowed and aged married individuals, are examined to deter mine if isolation is a function of diminished control over social resources. Indices of social interaction include visits with kin or friends or attendance at group meetings

PAGE 40

3 3 and functions. Additionally, social interaction with relatives includes telephone conversations and written communicat ion . Social interaction is measured on an ordinal scale. Therefore, it is assumed that an inverse relationship exists between rank-ordered categories of social interaction and social isolation. A discussion of the ordinal measures of the five areas of social interaction follows. Friendship interaction was measured by asking the respondents if they had any close friends nearby. If so, they were asked how often they got together with them. The structured categorical responses included: "all the time," "often," "sometimes," "seldom," and "never." For the purposes of this analysis the last two categories were collapsed, resulting in the following fourfold rank-order classification of friendship interaction: "all the time," "often," "sometimes," and "seldom or never." There are two measures of kin interaction: one focuses on interaction with relatives living nearby, the other assesses the amount of social interaction with relatives living far away. Division of the kin group into one of these two categories was determined by the respondents' subjective assessment of which emotionally close kin were and were not living nearby and far away. If the respondents indicated that they had any close relatives living either nearby or far away they were asked how often they got to

PAGE 41

54 see or talk to any of them, excluding relatives living in the household. Telephone conversations were included as accepted indices of interaction with all kin regardless of geographic location while letters were accepted as indicators of interaction only with kin living far away. Both measures of interaction with relatives had six fixed answer response-categories: "almost every day," "several times a week," "several times a month," "several times a year," "seldom," and "never." The first two categories and last two categories were combined for the analysis to yield the following fourfold rank-order classification of nearby and far away kin interaction: "weekly," "monthly," "yearly," and "seldom or never." Another measure of social interaction was obtained by asking respondents who were members of formal organizations how many afternoons and evenings a month they spend in club activities. The structured response -categories included: "none," "1-2," "3-5," "6-10," "11-15," and "over 15." For the analysis of these data the last three categories are assumed to indicate frequent activity and are therefore collapsed to form a category called "frequent." Interaction by respondents participating in clubs three to five afternoons and evenings a month is labeled "some," social interaction one or two afternoons or evenings a month is called "modest" while no interaction at all is labeled "none." The preceding process of combining categories

PAGE 42

3 5 yields a four-fold rank-order classification of social interaction in formal organizations. When asked how often they attended the main worship service of their church or synagogue, respondents claiming religious affiliation were required to answer from among the following responses: "every week," "2-3 times a month," "once a month," "several times a year," "only once or twice a year," "less than once a year," and "never." For this analysis the first response is labeled "weekly," "monthly" refers to the collapsed second and third categories, "yearly" includes the responses from the fourth, fifth, and sixth categories, while the final category is labeled "never." This procedure results in a four-fold rank-order classification of social interaction in religious organizations. The categories established for the analysis of social interaction in the five preceding social areas have frequently been constructed by collapsing two or more contiguous categories. This procedure results in the loss of data. However, this undesirable consequence is unavoidable because of the small sample size. Resources According to exchange theory, continued social interaction is contingent upon the reciprocal exchange of resources. Resources, when expended to consummate social interaction, become rewards. There are two basic types of rewards, intrinsic rewards and extrinsic rewards. Intrinsic

PAGE 43

3 6 rewards result from types of social interaction where the social relationship itself is considered to be a reward. Social situations which yield extrinsic rewards for actors are those in which the social relationship is a mechanism through which these rewards are realized (Blau, 1964:58-59). Because of insufficient methodological measuring techniques this research does not attempt to assess the impact of the exchange of intrinsic rewards on social relationships involving aged widowed and aged married individuals. This is regrettable because, undoubtably, many forms of social interaction with and among the elderly depend upon this type of reward. However, this research does attempt to determine the impact of extrinsic rewards upon rates of social interaction of the elderly. The research design specifies that rates of social interaction between the aged married and aged widowed individuals are compared to determine if social interaction is a function of control over six extrinsic rewards. Three of the rewards examined include age, sex, and race, all of which, it is assumed, are more important as indicators of other resources than as resources themselves. For example, Harvey (1973) suggests an inverse relationship between age and control over resources. The importance of age as a resource itself is overshadowed by the impact that age, as an independent variable, has on control over other resources, health status and income being two notable examples which are examined in this research.

PAGE 44

37 With regard to the other two resource variables, Berardo (1967), Bock (1972), Bock and Webber (1972), Harvey (1973), and Townsend (1957) have all suggested sex differences among the widowed in terms of rates and type of social interaction. Additionally, voluminous research documents the effect of race in circumscribing social relationships (Simpson and Yinger, 1972:12). The three remaining variables include education, income, and health status. Education was suggested as a resource by Harvey (1973). Additionally, Edwards (1969) maintained that knowledge is a resource. Education is assumed to be positively correlated with knowledge. In this research the educational resource is measured in terms of the number of years of education completed for each of the respondents. Income, economic resources, or money are mentioned as viable exchange commodities by Harvey (1973), McCall (1966), Streib (1972), and Thibaut and Kelley (1959). This resource is measured in this study by determining the income per year per household for each of the respon dents. Health status is also considered a resource by Streib (1972), and Thibaut and Kelley (1959). In the present study health status was determined by asking the respondents if they had any present physical or health problems .

PAGE 45

38 Hypothe ses The following ten hypotheses are engineered to examine the theoretical proposition developed inductively in the preceding chapter. The first five are designed to test if significant differences exist between interaction rates of aged widowed and aged married individuals. Hypotheses 6 through 10 are presented to explore the applicability of the exchange conceptual framework in providing explanations for these differences. Unfortunately, direct tests of the last five hypotheses are impossible. This occurs because comparisons between the measure of association between married and widowed individuals in the total sample and the measure of association between married and widowed individuals for each of the six control variables would involve tests of statistical significance based on dependent samples. For this reason no statistical inferences to any conceptual population are possible. Therefore Hypotheses 6 through 10 will serve only to guide the research effort. Null hypotheses are unwarranted when no direct inferences to a population are desired and therefore are not presented for the last five hypotheses. The subsequent analysis of each proposition depicts the relationship between the hypothesis and the previous literature. Research Hypothesis 1_ The aged widowed respondents will interact less fre quently with nearby kin than aged married individuals.

PAGE 46

39 Null Hypothesis 1 There will be no difference between aged widowed and aged married individuals in terms of the frequency of interaction with nearby kin. Reviewed literature pertaining to kin interaction among the elderly suggests that kin interaction declines with the death of the spouse. Research pertaining to kin interaction focuses on interaction with children as well as members of the larger kin network (Adams, 1968; Berardo, 1967; Bock, 1972; Bock and Webber, 1972; Pihlblad and Adams, 1972; and Townsend, 1957). This hypotnesis only examines interaction with kin considered by the respondents to be living nearby. Rese a rch Hypo th esis 2_ The aged widowed respondents will interact less frequently with kin living far away than aged married individual s . Nul l Hypo t hesis 2_ There will be no difference between aged widowed and aged married individuals in terms of the frequency of interaction with kin living far away. Research related to this hypothesis includes the same studies and findings as those presented for the second hypothes is .

PAGE 47

4 Re sear ch Hypothesi s 3 The aged widowed respondents will interact less frequently with friends than aged married individuals. Null H ypothesi s 3^ There will be no difference between aged widowed and aged married individuals in terms of the frequency of friendship interaction. Previous research pertaining to this hypothesis includes Berardo (1967), Blau (1961), Booth (1972), Griffiths et al (1971), Lopata (1973), Pihlblad and Adams (1972), Rosow (1967), and Townsend (1957). With the exception of Townsend's finding that the widowed visit their neighbors more than married individuals, the weight of the evidence from these studies suggests that the widowed are more isolated in terms of friendship interaction than married individuals. Research Hyp o the si s 4_ The aged widowed respondents will interact less frequently in religious organizations than aged married individu a 1 s . Null Hy po the si s 4 There will be no difference between aged widowed and aged married individuals in terms of the frequency of interaction in religious organizations. The literature upon which this hypothesis is based, suggests that widows are more likely to interact in religious

PAGE 48

41 organizations than are either widowers or married individuals (Berardo, 1967; Pihlblad and Adams, 1972; and Townsend, 1957) However, the hypothesized relationship is still anticipated because the extremely low rates of interaction in religious organizations by widowers is expected to bring the total interaction rate of the widowed below that of the married. Research Hypothesis 5^ The aged widowed respondents will interact less frequently in formal organizations than aged married individuals . Null Hypothesis 5^ There will be no difference between aged widowed and aged married individuals in terms of the frequency of interaction in formal organizations. Previous research (Berardo, 1967; Bock, 1972; Bock and Webber, 1972, Harvey and Bahr, 1974; Lopata, 1973; and Pihlblad and Adams, 1972) suggests that the widowed, particularly widowers, are less likely to interact in formal organizations than married individuals. Research Hypoth e sis i5 Controlling for age, sex, race, education, health status, and financial status, the rates of interaction with nearby kin between the aged widowed and aged married individuals will be less different than will the rates without controls. Reviewed literature pertaining to kin interaction among the elderly conclusively suggests that kin interaction declines

PAGE 49

4 2 with the deatli of the spouse. Research pertaining to kin interaction focuses on interaction with children as well as members of the larger kin network (Adams, 1968; Berardo , 1967; Bock and Webber, 1972; Pihlblad and Adams, 1972; and Townsend, 1957). This hypothesis only examines interaction with kin considered by the respondents to be living nearby. This hypothesis builds upon previous research by determining if differences in the rates of social interaction with nearby kin between the aged widowed and aged married individuals is a function of differential control over resources examined in this study. Research Hypothesis 7_ Controlling for age, sex, race, education, health status, and financial status, rates of interaction with faraway kin between the aged widowed and aged married individuals will be less different than will the rates without controls . Research related to this hypothesis includes the same studies and findings as those presented for the sixth hypothesis. This hypothesis builds upon previous research by determining if differences in rates of social interaction with kin considered to be living far away between aged widowed and aged married individuals is a function of differential control over social resoures examined in this study Research Hypo the s is 8^ Controlling for age, sex, race, education, health status, and financial status, the rates of interaction with

PAGE 50

45 friends between the aged widowed and aged married individuals will be less different than will the rates without controls. Previous research pertaining to this hypothesis includes Berardo (1967), Blau (1961), Booth (1972), Griffiths et al., (1971), Lopata (1973), Pihlblad and Adams (1972), Rosow (1967) and Townsend (1957). With the exception of Townsend's finding that the widowed visit their neighbors more than married individuals, the weight of the evidence from these studies suggests that the widowed are more isolated in terms of friendship interaction than married individuals. Additionally, among the widowed, the amount of friendship interaction is circumscribed by socioeconomic ststus and educational attainment. This hypothesis goes beyond the preceding research by determining if differences in the rates of friendship interaction between aged widowed and married individuals is a function of differential control over resource variables which include age, race, and health status in addition to sex, education, and income, variables controlled in earlier studies . Research Hypothes is 9^ Controlling for age, sex, race, education, health status, and financial status, the rates of interaction in religious organizations between aged widowed and aged married individuals will be less different than will the rates without control s .

PAGE 51

44 The literature upon which this hypothesis is based suggests that widows are more likely to interact in religious organizations than are either widowers or married individuals (Berardo, 1967; Pihlblad and Adams, 1972; and Townsend, 1957). However, because this research is examining rates of interaction between aged widowed and aged married individuals, widows and widowers are grouped together so that they may be compared with the aged married individuals. The differences in interaction between widows and widowers are examined in this hypothesis when statistical controls are applied for sex. This hypothesis expands upon research in the area by determining if differences in the rates of interaction in religious organizations between aged widowed and aged married individuals is a function of differential control over resources examined in this study. Research Hypo t hesis 10 Controlling for age, sex, race, education, health status, and financial status, the rates of interaction in formal organizations between the aged widowed and aged married individuals will be less different than will the rates without controls. Previous research (Rerardo, 1967; Bock, 1972; Bock and Webber, 1972, Harvey and Bahr, 1974; Lopata, 1973, Pihlblad and Adams, 1972) suggests that the widowed, particularly widowers, are less likely to interact in formal organizations than married individuals. Additionally, among widows, a

PAGE 52

45 positive relationship exists between economic status and education, and rates of interaction in formal organizations. This hypothesis attempts to extend the results of the preceding research by determining if differences in the rates of social interaction between the aged widowed and aged married individuals is a function of differential control over resources examined in this study. Data Collection Data for this research were obtained from a large scale sample survey designed to assess the mental health needs of individuals living in Alachua County, Florida. The population consisted of all residential households in that county. The sampling design specified a systematic random sample to be selected from a sample frame consisting of electrical utility company residential household listings. The utility listings combined claimed a 98 percent coverage of all households in the area. Because seven companies provide utilities to the county's inhabitants the sample frame was constructed to provide proportional representation to each of the utility lists based upon the number of households they served. The sample was drawn by initially selecting at random a household from among the first thirteen households on the sample frame. Thereafter, every thirteenth household was included in the sample. Adult respondents (individuals 18 or older) within each household were randomly selected with the Kish table. If the individual selected refused to be interviewed the household itself was considered a

PAGE 53

4 6 refusal and no additional solicitation of interviews from among the household members was attempted. Potential respondents were considered refusals if the interviewer was certain that return visits would prove unsuccessful. Households where no one was at home were revisited on at least three separate occasions. Total sample size included 2315 households. A subsample of 511 respondents was randomly drawn from the total sample in order to pretest the interview instrument and polish interviewing techniques. Following completion of the pilot study, interviewing for the major study, which became the source of these data, began in June 1970 and was completed by the end of the year. The major study resulted in 1645 usuable interviews. The total nonresponse rate for the major study was 16.07 percent. The nonresponse rate consisted of the following five categories: "not at homes," 4.6 percent; "refusals," 8.06 percent; "unable to be interviewed," 1.4 2 percent; "unable to locate," 1.63 percent; and "sampling problems," .03 percent. Individuals unable to be interviewed were those suffering from mental or physical impairment to the extent that they could not be interviewed. Sampling problems occurred in a very small percentage of the cases when households moved during the interviewing period to geographic locations outside the county lines. Data from this large epidemiological survey yielded information which is utilized in this research. Sociodemographic data such as race, sex, age, education, financial

PAGE 54

4 7 status, and marital status were obtained from the respondents. Additionally, information concerning interaction with kin and friends in addition to interaction in formal organizations and religious organizations was collected. This data bank provided the preceding information on 128 aged widowed and 14 5 aged married individuals. These 273 individuals are the focal point of this research. Statistical Procedure s Two statistics are used in this research. They are the Kolmogorov-Smirnov and the gamma statistics. The Kolmogorov-Smirnov is the appropriate test for the first five hypotheses while the gamma is applicable for testing the final five hypotheses. Kolm ogorov-Smirnov Statistic The Kolmogorov-Smirnov statistic is a twosample nonparametric test which determines whether "two independent samples have been drawn from the same population or from populations with the same distribution" (Siegel, 1956:127). This statistic is appropriate for testing the first five hypotheses because the data are collapsed into a limited number of ordered categories representing social interaction frequencies. These hypotheses compare aged widowed with aged married individuals in terms of frequency of social interaction with friends, kin living both nearby and far away, and interaction in formal organizations as well as religious organizations.

PAGE 55

48 Gamma Statistic This statistic provides a measure of association for grouped ordinal data (Blalock, 1972:424). It is used in the last five hypotheses to measure the degree of agreement between rates of social interaction of aged widowed and aged married individuals. The research design specifies that for each of the five hypotheses, rates of social interaction of the aged widowed are compared with interaction rates of aged married individuals in friend groups, nearby and faraway kin groups, formal organizations, and religious organizations. Each hypothesis deals with social interaction in only one of the five specified groups or organizations. The degree of agreement between aged widowed and aged married individuals is computed with the gamma statistic for each area of social interaction. Subsequently, rates of social interaction in the same areas are measured between the aged widowed and aged married individuals controlling for resources considered to influence rates of social interaction. The resources which are controlled include age, sex, race, income, education, and health status. Statistical Controls Controls for all six of the resource variables are accomplished for each of the five hypotheses. Because of limitations in the data and the insufficient development of methodological techniques for exerting statistical controls, only one resource is controlled at a time. Controls are administered as follows. Each resource is divided into different levels which are either qualitative

PAGE 56

4 9 or quantitative in nature depending upon the variable examined. Measures of association are then computed between the social interaction rates of aged widowed and aged married individuals for each level of the control variable. This process holds constant the impact of the control variable upon the relationship being investigated. Then to obtain some feeling for controlling for the different levels simultaneously, the individual gammas are averaged to form a partial gamma value. In computing this average score the gammas for each level are weighted to provide proportionate representation to the relative number of respondents in each level of the control variable. The partial gamma value for each resource is then compared with the gamma calculated between the rates of social interaction of aged widowed and aged married individuals before controlling for resources. If the extension of statistical controls results in the reduction of the gammas computed for each level of the control variable then that factor may be assumed to influence the interaction rates of the elderly. This process is accomplished for each of the six resources for every hypothesis. However, before the gammas calculated on each of the different levels of each of the control variables can be combined to yield a partial gamma, a test of statistical interaction has to be computed to determine if the relationships between interaction rates of the aged differ for different levels of the control variable. If this situation

PAGE 57

50 occurs the individual gammas for each of the levels of the control variable can not be averaged together for lack of commonal ity . Statistical interaction is tested for significance in either one of the two following ways depending upon the number of levels in the control variable. If the control variable is divided into two levels a Z test statistic is used to test for a significant difference between the two measures of association. The standard error of the sampling distribution is the square root of the summed variances of the two gammas. If the control variable has three levels, each of the gammas is weighted and combined to yield a partial gamma which is then subtracted from each of the three individual gammas in turn, squared, and divided by each of the variances of the individual gammas. The results are summed to provide a score which is tested for significance on the X" sampling distribution with c 1 degrees of freedom where c is the number of levels of the control variable. These two tests are necessary to determine whether or not "the sample interactions are sufficiently large that they could have readily occurred by chance even if there was no population interaction" (Blalock, 1972:309). If the interactions are not statistically significant and not large enough to be substantively meaningful, controls for each resource in each hypothesis are administered (Blalock, 1972:509) However, if there are significant statistical interactions, different levels of the control variable are acknowledged as circumscribing relationships between the interaction rates

PAGE 58

51 of aged widowed and aged married individuals. An illustration of the process of exerting statistical controls in this research follows. Statistical controls for age for each of the five areas of social interaction are accomplished by dividing the aged into three levels: respondents 60 to 65 years old, 66 to 7 5 years old, and 76 years old and older. Then, for each level, rates of social interaction are compared between aged married and aged widowed individuals for each of the areas of social interaction. Comparisons between the two aged groups are accomplished with the gamma statistic. The gammas are weighted and averaged to yield a score representing comparisons between the aged widowed and aged married controlling for the three levels of age. This score is then compared with a comparison between rates of social interaction between the aged widowed and aged married individuals not controlling for resources. Each of the remaining resource variables is controlled in much the same fashion, the only differences among them resulting from the number and types of levels of control variables established. Race, for control purposes, is dichotomized into "blacks" and "whites." A third racial category, "other," had too few respondents to be utilized as a control level. Sex is also used as a control variable. Financial status, which is indicated by yearly household income, is broken down into two control levels, those with household incomes greater and less

PAGE 59

5 2 than $4500 per year. This figures was selected, not because of any particular social, economic, or political significance, but because it represents the median income of the aged respondents interviewed. Health status is separated into two categories, those who reported that they have health problems, and those who reported that they do not. Finally, education, for control purposes, is divided into three levels, those who range from no formal education to completion of the eighth grade, individuals who completed some high school or are high school graduates, and those respondents who have some college, technical, or trade education. A Description of the Data The population to which the findings from these data are generalizable includes all individuals 60 years old and older living in Alachua County, Florida. Inferences can be made from this sample because the assumptions of probability sampling are met. The sample consists of 273 individual s . Table 1 summarizes the data by displaying the frequency and percent of respondents according to different values of the seven variables analyzed in this research: marital status, sex, age, race, education, health status, and income. The last six variables are resources which are controlled in this study in an attempt to investigate the applicability of the theoretical proposition set forth in the second chapter which was developed from the perspective of the exchange conceptual framework.

PAGE 60

53 TABLE 1. Frequency and Percent of Respondents by Levels of Marital Status, Sex, Age, Race, Education, Health Status, and Income Frequency Percent Marital Status Sex Age Race Married Widowed Total 146 53.1 12 9 4 6.9 275 100.0 Male Female Total 102 173 27 5 60 to 65 Years Old 100 6 5 to 7 5 Years Old 120 55 7 5 Years Old and Older Total 275 Wh i t e Black Total 3 7 . 63.0 10 0.0 3 6 . 4 43.6 2 0.0 10 0.0 195

PAGE 61

TABLE 1. (continued) 54 Health Status Frequency Percent Health Problems 197 71.6 No Health Problems 78 28.4 Total 275 10 0.0 Income Household Income 141 Less than $4500 Household Income 88 Equal to or Greater than $4500 Total 229 61.6 38.4 100.0

PAGE 62

Of the individuals sampled, 53.1 percent are married and 46.9 percent are widowed. The ratio of married to widowed individuals in the sample is 1.13 to 1. Using national census data as a point of reference, the marital status distribution of the sample includes a disproportionate number of widowed. The ratio of married to widowed individuals for United States residents 60 years old and older is 5.77 to 1. The sex distribution of this aged sample includes 37 percent males and 63 percent females. The sample, when compared with the sex distribution of Alachua County, slightly underrepresents females. Sixty-seven percent of all individuals 60 or older in the county were women. A cross tabulation of sex by marital status reveals that there are over five times as many widows as widowers and there are one-third again as many married women as men. The age distribution of the respondents ranges from 60 to 91. The model age of the sample is 65, the mean age is 70.4, and the median age is 68. for purposes of statistical analysis, the sample is divided into three levels. The middle level, which consists of all respondents 65 to 75 years old, is the largest including 43.6 percent of the interviewees. The 60to 65-year-old category includes 36.4 percent of the respondents and the remaining level, consisting of those individuals 75 years old and older, is smallest encompassing 20 percent of the sample. The age distribution of Alachua County, the population to which

PAGE 63

56 inferences from the sample are made, closely resembles the sample. Of all individuals in the county aged 60 or over, 31 percent are 60 to 65, 42.7 percent are 65 to 75, and 24.2 percent are 75 years old and older. The sample consists of 70.9 percent white and 29.1 percent black individuals. The sampling error in this as well as the preceding category is fairly small because, for the county as a whole, the percent of aged whites is 73.3 and 26.5 for blacks. The percentages do not total to 100 because of a small number (23) of aged nonblack minority group members living in the county. Just under one half of all respondents (49.8 percent) completed less than nine years of formal education. Approximately one fourth (25.5 percent) of the sample completed at least the ninth grade but did not advance their education beyond high school graduation. The remaining quarter (24.8 percent) of the interviewees completed some college, technical, or trade education. The vast majority of the aged individuals interviewed reported health problems. Only 28.4 percent perceived themselves as having no physical or health problems at the time of the interview. interviewees with household incomes less than $4500 comprised 61.6 percent of the sample. The remaining portion of the sample has household incomes equal to or greater than $4500 per year. The distribution of household income

PAGE 64

57 of the respondents presents some interesting contrasts when cross tabulated with marital status. For the widowed, total household incomes range from $300 to $11,400 per year. However, for married respondents, total household incomes range from $960 to $45,000. The median income for the two groups respectively is $2124 and $5000. These particular figures should be viewed with some skepticism because of the large number of missing observations. It is impossible to determine if the missing data are randomly distributed or if selective factors operated to distort the sample with regard to reportage of household income. Summar y This research examines differences between rates of social interaction for aged widowed and aged married individuals. Social interaction is measured in five different areas, friend interaction, interaction with kin living nearby and far away, and interaction in formal organizations and religious organizations. Comparisons of interaction rates between the aged widowed and aged married individuals are made both before and after controlling for age, sex, race, education, financial status, and health status .

PAGE 65

CHAPTER IV FINDINGS This chapter presents empirical findings gleaned from this research enterprise. The statistical analyses of the data are examined separately for each hypothesis. Hypothe ses Researc h Hypothesis 1_ The aged widowed respondents will interact less frequently with nearby kin than aged married individuals. Null Hypo t he s is 1_ There will be no difference between aged widowed and aged married individuals in terms of the frequency of interaction with nearby kin. The test for differences between married and widowed individuals, using the frequency of interaction with nearby kin, is the Kolmogorov-Smirnov statistic. The differences between the two groups is not significant (see Table 2). This result fails to support the contention of findings from earlier research (Berardo, 1967; Bock and Webber, 1972; and Townsend, 1957) that aged married individuals interact with kin more than aged widowed individuals. Insufficient support for this hypothesis could reflect the finding by 58

PAGE 66

59 Pihlblad and Adams (1972) that kin interaction declines upon the death of a spouse for elderly males but not elderly females indicating that sex, more than marital status, influences the interaction of t h e aged with their chi 1 dren . Research Hypo thesis 2_ The aged widowed respondents will interact less frequently with kin living far away than aged married individuals. Nul l Hypothesis 2 There will be no difference between aged widowed and aged married individuals in terms of the frequency of interaction with kin living far away. No significant difference between the two samples of the aged is found in terms of their rates of interaction with kin considered to be living far away (see Table 2). The Ko] mogorov-Smirov statistic is used to test for differences between the two groups. As in the preceding test, these results fail to confirm earlier research findings (Berardo, 1967; Bock and Webber, 1972; and Townsend, 1957). The inconsistency of the findings from this research with these earlier studies concerning kin interaction perhaps, as suggested in the first hypothesis, results more from sex than widowhood status. If not, few alternative explanations present themselves. These conflicting results can not readily be attributed to the small sample size of this survey

PAGE 67

60 All four groups, that is, aged widowed and married individuals interacting with nearby and far away kin, have at least 100 respondents. Assuming relatively small sampling as well as nonsampling errors because of the sample design and sample size, this inconsistency is perplexing and perhaps can only be explained by the fact that the studies in question are making inferences to incomparable populations. This suggested explanation notwithstanding, further investigation into this area is most certainly warranted. Resea rch H ypothesis 3 The aged widowed respondents will interact less frequently with friends than aged married individuals. Nul 1 Hypot h esis 3^ There will be no difference between aged widowed and aged married individuals in terms of the frequency of friendship interaction. The Kolmogorov-Smirnov statistic is used to test for differences between the two samples. For aged individuals a significant difference between widowed and married persons is found in terms of interacting with friends (see Table 2) . The difference between the two groups is significant at the .05 level of significance; however , it is not in the hypothesized direction. Therefore the null is not rejected in favor of the research hypothesis. This finding lends support to Townsend's (1957) earlier research on the aged but is inconsistent with research

PAGE 68

61 TABLE 2. Comparisons Between Married and Widowed Individ uals in Terms of Frequency of Interaction in Nearby Kin Groups, Far-Away Kin Groups, Friend Groups, Religious Organizations and Formal Organi zat ions

PAGE 69

6 2 TABLE (cont inued Group Cumu Frequency of Married lative Interaction (NJ percent ed fN) percent Ma r ita 1 S t a tus Cumu Wi dow1 at i ve Forma 1

PAGE 70

6 3 findings from Berardo (1967) and Booth (1972). However, even though the findings are not in the hypothesized direction, the overall research design developed to investigate the applicability of the exchange conceptual framework in explaining the social interaction of the aged, is unaffected To this end, this research seeks to determine if the difference between the two groups is a function of control over resources. Research Hypothes is 4_ The aged widowed respondents will interact less frequently in religious organizations than aged married individuals . Null Hypothesis 4 There will be no difference between aged widowed and aged married individuals in terms of the frecpaency of interaction in religious organizations. The Kolmogorov-Smi rnov statistic is used to test for differences in frequency of participation in religious organizations between the two preceding groups of aged individuals. The null hypothesis is not rejected. Previous research in this area (Berardo, 1967; Pihlblad and Adams, 1972; and Townsend, 1957) differentiates by sex among the aged widowed, suggesting that widows are more likely to interact in religious organizations than are either widowers or married individuals. Because this hypothesis makes no distinction by sex among the aged widowed it can

PAGE 71

61 neither lend support to nor detract from earlier studies. However, controls for sex are extended in the ninth hypoth Research Hypothesis S_ The aged widowed respondents will interact less frequently in formal organizations than aged married individ u a 1 s . Nul 1 Hypothesis 5^ There will be no difference between aged widowed and aged married individuals in terms of the frequency of interaction in formal organizations. A significant difference between the two samples of aged individuals in terms of their interaction in formal organizations is found (see Table 2). The KolmogorovSmirnov statistic is used to test for differences between the two groups. The two samples are significantly different at the .05 level of significance. However, the data are not in the hypothesized direction, suggesting that the widowed interact with greater frequency in formal organizations than married individuals. The null hypothesis is not rejected. Relevant earlier research on the interaction of the aged in formal organizations was conducted by Berardo (1967) , Bock and Webber (1972), Harvey and Bahr (1974), and Pihlblad and Adams (1972). Their analyses were conducted while implementing controls for sex. They indicate widowers are

PAGE 72

65 less likely to interact in formal organizations than either married individuals or widows. Moreover, the difference between the latter two groups is minimal. No controls for sex are administered in this hypothesis because it was thought that, when considered together, the widowed would exhibit smaller rates of social interaction than married individuals. However, sex is held constant in the final hypothesis. Unfortunately, direct tests of the remaining five hypotheses are impossible. This occurs because comparisons between the measure of association between widowed and married individuals in the total sample and the measure of association between widowed and married individuals for each of the six control variables would involve tests of statistical significance based on dependent samples. For this reason no statistical inferences to any conceptual population are possible. Therefore, the succeeding hypotheses will serve only to guide the research effort. Null hypotheses are unwarranted when no direct inferences to a population are desired and therefore are not presented for the last five hypotheses. When no statistical interaction exists, comparisons between the six partial gammas, representing the measure of association between interaction rates of aged widowed and aged married individuals for each control variable, and the total gamma score, will be described for each area of social interaction investigated in this research.

PAGE 73

66 Research Hypothesis 6 Controlling for sex, age, race, education, health status, and financial status, the rates of interaction with nearby kin between the aged widowed and aged married individuals will be less different than will the rates without controls. Table 3 indicates that the measure of association between married and widowed respondents engaging in interaction with nearby kin is -.03. The size of this measure is negligible. However, it is consistent with the first hypothesis which found no significant difference between married and widowed respondents engaging in interaction wit]] nearby kin. Comparisons of the gamma score with the partial gammas, which represent controls for the six resources, reveal that none of the variables appear to be important as social exchange resources. p a i r s sum. of wi sons, s o c i a are c r e s e a inter vesti indiv it sh conce that the s value when pa rt i 1 The m i n u All dowed Whe 1 int oncor rch , act w g a ted i dual o u 1 d rned they o c i a 1 s o f o b s e r al i>a f ormu s the table p e r s n com eract dant negat i t h g whil s int be em only are 1 cont the g ved r mm as la for g number s in thi ons with paring t ion the and disc ive gamm r eater f e p o s i t i eract mo phasi zed with whe ikely ca e x t i n v e a mm as an elat ive are c 1 o s arama is the number of concordant of discordant pairs divided by their s research compare interaction rates interaction rates of married perhe two sets of ordered rankings of starting point for determining what ordant pairs is arbitrary. For this a values indicate that the widowed requency in the social settings inve gamma values mean that married re often in these areas. However, that the last five hypotheses are ther controls for variables indicate ndidates for exchange resources in stigated. To this end, the sign d partial gammas are meaningful only to one another to determine if the er to zero than the gammas.

PAGE 74

6 7 TABLE 3. Measure of Association Between, and Frequency and Percent of Married and Widowed Respondents Engaging in Interaction with Nearby Kin by Sex, Age, Race, Education, Health Status, and Income Group Frequency of Interaction with Nearby Kin

PAGE 75

6 8 TABLE 5. (continued)

PAGE 76

6 9 TABLE 3. (continued) Frequency of InGroup teraction with Nearby Kin

PAGE 77

7 Moreover, the application of controls resulted in an increase rather than the expected decrease in the measure of association between rates of interaction of aged married and aged widowed individuals for all levels of all control variables except for the elderly expressing health problems. This phenomenon, which is found frequently throughout this research, occurs when the extension of control levels serves to intensify the measure of association within the limits established by the control categories. Furthermore, the attempted control for income among the individuals sampled resulted in increased gammas for both high and low income respondents which are sufficiently large so that the difference between the two levels is statistically significant. This finding suggests that among the lower income respondents the widowed interact more frequently with kin living nearby while among the higher income groups the married interact with nearby kin more often. Research Hypoth e sis 1__ Controlling for age, sex, race, education, health status, and financial status, the rates of interaction with far away kin between the aged widowed and aged married individuals will be less different than will be rates without controls. The measure of association between married and widowed individuals engaging in interaction with kin considered by the respondents to be living far away is -.02 (see Table 4). For tlris hypothesis, as well as the preceding one, the

PAGE 78

71 TABLE 4. Measure of Association Between, and Frequency and Percent of Married and Widowed Respondents Engaging in Interaction with Par Away Kin by Sex, Age, Race, Education, Health Status, and Income Frequency of InGroup teraction with Par Away Kin Total Group Weekly Monthly Yearly Never Maritc

PAGE 79

72 TABLE 4. (continued)

PAGE 80

73 TABLE 4. (continued)

PAGE 81

7 4 extremely small size of the gamma makes any interpretation of the impact of statistical controls for variables almost meaningless. None of the controls extended reveal a diminution in the strength of the relationship between widowed and married individuals. indeed, virtually all controls for the resources result in increased measures of association between rates of interaction of aged widowed and aged married individuals within the control categories. Moreover, the gammas computed for different levels of race are sufficiently different to be statistically significant indicating that among white respondents the widowed are more likely to interact frequently with kin perceived to be living far away while among blacks the married tend to interact more frequently with far away kin. Earlier research by Pihlblad and Adams (1972) suggests that widowhood decreases kin interaction for men but not women. These data support this earlier finding. The control for sex reveals that the measure of association between married and widowed females is .04 whereas the comparable gamma computed for males (gamma=.19) indicates that married males interact more frequently with kin living far away than widowers. R esearch Hypothesis 8_ Controlling for age, sex, race, education, health status, and financial status, the rates of interaction with friends between the aged widowed and aged married individuals will be less different than will the rates without controls

PAGE 82

7 5 A gamma score of -.27 indicates the measure of association between rates of interaction with friends for married and widowed individuals (see Table 5). This finding suggests that the aged widowed tend to interact more frequently with friends than aged married individuals. Controls for all six resource variables examined in this hypothesis indicate no reductions in the measure of association between married and widowed individuals. The data do not support the hypothesis. The research findings from this hypothesis draw attention to a few noteworthy observations. First, the impact of income in circumscribing social interaction with friends is depicted in other research by Blau (1961), Griffiths (1971), Lopata (1973, and Rosow (1967). These previous studies suggest a positive relationship between financial status and both friendship interaction and satisfactory friendship relationships. This finding is not supported in the present research. An additional point of interest in this hypothesis concerns the extension of educational controls. These controls reveal that the widowed are more likely to interact with friends than married individuals in each of the three educational levels. Moreover, it is among the highest educational level that widowed interact with the greatest frequency followed by the lowest and then the intermediate educational levels. This finding is somewhat inconsistent with Lopata ' s research which suggests a direct relationship

PAGE 83

7 6 TABLE 5. Measure of Association Between, and Frequency and Percent of Married and Widowed Respondents Engaging in Interaction with Friends by Sex, Age, Race, Education, Health Status, and Income Groups Frequency of Interaction with Friends Marital Status Married (N) (%) Widowed (N) m Total Group All the Time Often Sometimes Never 25 52 5 2 15 20. 5 42.6 26. 2 10.6 46 38.0 42 34.7 25 20.7 8 6.6 Total 122 9 9.9 121 100.0

PAGE 84

TABLE 5. (continued) 77 Group Frequency of Interaction with Friends Marital Status Married (N) V'o) Widowed :n) m Age 7 5 Years Old and Older (Gamma = . 37 o =.25) g All the Time 1 Often 4 Sometimes 2 Never 2 1.0 11 9.1 3.3 16 13.2 1.6 8 6.6 ] . 6 3 2.5 Total 122 100.0 121 100.0 Partial Gamma*= . 07 a =.1 Race Wh i t e

PAGE 85

TABLE 5. (continued) Group ; requency of Interaction with Friends Marital Status Married (N) (%) Widowed (n) m Health

PAGE 86

79 between educational status and friend satisfaction. However, this inconsistency is valid only when the assumption is made that friend interaction and friend satisfaction are indicative of one another. Research H ypothesis 9 Controlling for age, sex, race, education, health status, and financial status, the rates of interaction in religious organizations between aged widowed and aged married individuals will be less different than will the rates without controls. The strength of the relationship between interaction rates in religious organizations for aged widowed and aged married individuals, as measured by gamma, is -.13 (see Table 6). This score suggests that the widowed, as a group, are more likely to interact in religious organizations with greater frequency than married individuals. Generally speaking, the hypothesis is not supported by the data. Controls for none of the variables result in meaningful diminutions in the partial gammas. However, income, when held constant, demonstrates a very modest reduction in the measure of association between the two groups. This indicates a possible interpretation of this factor as an independent variable which affects social relationships in the capacity of an exchange resource. However, the extremely small difference makes any interpretation of this nature particularly d a r i n g .

PAGE 87

TABLE 6. Measure of Association Between, and Frequency and Percent of Married and Widowed Respondents Engaginj in Interaction in Religious Organizations by Sex, Age, Race, Education, Health Status, and Income Group Marital Status Frequency of Interaction in ReliMarried Widowed gious Organizations (N) {%) (N) (V Total Group Weekly

PAGE 88

TABLE 6. (continued)

PAGE 89

82 TABLE 6.

PAGE 90

8 3 Additionally, controls for sex yield some interesting findings. Berardo (1967), Pihlblad and Adams (1972) and Townsend (1957) document the impact of sex in influencing interaction in religious organizations, particularly among the widowed. In this research controls for sex yield findings consistent with these earlier reports. Among females the widowed are more likely to interact in religious organizations more frequently than married women while married males tend to participate to a greater degree in church-related activities than widowers. Research Hypothesis 10 Controlling for age, sex, race, education, health status, and financial status, the rates of interaction in formal organizations between the aged widowed and aged married individuals will be less different than will the rates without controls. Table 7 indicates that the measure of association between married and widowed respondents engaging in interaction in formal organizations is -.32. This score suggests that the widowed tend to interact with greater frequency in formal organizations than married individuals. Of all five areas of social interaction investigated, this lias the largest gamma. However, the sample size for formal organizations is drastically smaller than the other four areas, suggesting that, the total gamma computed for this area is somewhat suspect. To elaborate, it is impossible to determine if selective factors operated to systematically

PAGE 91

8:1 TABLE 7. Measure of Association Between, and Frequency and Percent of Married and Widowed Respondents Engaging in Interaction in Formal Organizations by Sex, Age, Race, Education, Health Status, and Income Frequency of InMarital Status Group teraction in ForMarried Widowed mal Organizations (N) (%) (N) (I) Total Group Often 9 13.4 10 20.4 Generally 15 22.4 13 26.5 Seldom 26 38.8 24 49.0 Never 17 25.4 2 4.1 Total 67 100.0 49 100.0 Gamma= . 52 a = . 13 Sex

PAGE 92

TABLE 7. (continued) Frequency of InMarital St atus Group teraction in ForMarried Widowed" mal Organizations (N) (!) (N) (!) 75 Years Old

PAGE 93

86 TAB LI: (continued)

PAGE 94

8 7 eliminate from the sample certain groups of respondents. If this occurred, the total gamma computed for this area is uncomparable with earlier similar measures. Generally, controls for none of the six resource variables result in decreased partial gamma scores. Moreover, comparisons of the gamma score for the entire group with the gammas computed for each level of each of the variables considered reveal that the partial gammas are meaningless because the individual gamma values computed for each control level are larger than the total gamma. For this reason none of the factors considered can be interpreted as viable exchange resources. Summary In conclusion, this chapter examines and compares the interaction rates of aged widowed and aged married individuals in kin and friend groups as well as in religious and formal organizations. The first five hypotheses, designed to test if significant differences exist between aged widowed and aged married individuals in terms of inter action in these four broad areas, are not supported. Moreover, the second set of hypotheses, postulated to assess the impact of variables assumed to act as exchange factors, reveal that only income could be suggested as an exchange resource which exerts a modest impact upon religious organizational interaction. Concluding statements which analyze and interpret these findings are presented in the following chapter.

PAGE 95

CHAPTER V SUMMARY AND CONCLUSIONS This dissertation has examined the social interaction of the aged in four broad areas: kin groups, friend groups, formal organizations, and religious organizations. Interaction was measured in terms of frequency of participation in these areas. Rates of social interaction of the aged widowed were compared with rates of social interaction of aged married individuals. The interaction rates were subsequently compared again controlling for resources which are considered, from the exchange conceptual framework, to be both relevant and necessary to maintain social interaction Exchange resources include education, income, and health status. Additionally, age, sex, and race were also controlled because they are assumed to be indirectly indicative of valued resources. The impact of many of these resources upon the social real t ionshi ps of aged widowed and aged married individuals has been suggested or tested in previous research (Berardo, 1967; Blau, 1961; Bock, 1972; Bock and Webber, 1972; Griffiths et al., 1971; Harvey and Bahr, 1974; Lopata, 1973; bowenthal , 1964; Pihlblad and Adams, 1972; Rosow, 1967; and Townsend, 1957). However, this study, to the author's knowledge, is the first application of the

PAGE 96

8 ( .) exchange conceptual framework to an analysis of social interaction rates of the elderly. The research findings are summarized below. Each of the first five hypotheses examined the social interaction of the aged in nearby kin groups, far away kin groups, friend groups, religious organizations, and forma] organizations. This set of hypotheses was designed to test whether significant differences exist between aged widowed and aged married individuals. More specifically, they attempted to determine if the widowed, as previous research has suggested, were more isolated than married individuals. The second five hypotheses sought to determine if differences in rates of interaction between the two groups of aged individuals diminished upon controlling for variables considered, from the exchange conceptual framework, to circumscribe social interaction. Any diminution in the measure of association of the rates of interaction between the two groups of aged persons may be interpreted as lending support to the exchange framework. All hypotheses in the second group focused on interaction in the same five areas as the preceding group of hypotheses. However, this second group of hypotheses served only to guide the research effort without providing direct tests of significance because their extension of controls involved comparisons between dependent samples. Attention is now focused on individual analyses of each of the hypotheses.

PAGE 97

90 Research by Berardo (1967), Bock and Webber (1972), and Townsend (1957) reports that aged married individuals interact more with kin than do the aged widowed. Pihlblad and Adams (1972) demonstrate this relationship for elderly men but not elderly women. Hypotheses 1 and 2, developed inductively from these earlier research efforts, posit the same relationship. However, in order to hold constant the impact physical distance has on circumscribing the social interaction of aged persons, kin interaction was divided into two categories: interaction with kin regarded by the respondents to be nearby in geographical proximity, and relatives perceived to be living far away. The first hypothesis focused on nearby kin interaction. The second pertained to far away kin. Neither hypothesis was supported. The Kolmogorov-Smirnov values for the first and second hypotheses were .33 and 1.93 respectively. This inconsistency with earlier research findings raises some questions which are not easily answered. It is possible that the separation o l~ kin into two groups, in effect extending controls for geographic distance, resulted in the inconsistent research findings. If this is the case, statistical interaction, resulting from peculiar multiplicative effects from combining nearby with far away kin groups, could have resulted in significant differences between aged widowed and aged married persons in terms of rates of social interaction with kin when no distinction by physical proximity was made, and no significant

PAGE 98

91 differences between the two groups of elderly when it was. Other potential explanations include possible problems in the research designs of the studies involved or perhaps the samples upon which the studies were based were drawn from uncomparable populations. Findings from the third hypothesis revealed that the Kolmogorov-Smirnov value of 7.44, which was significant at the .05 level, did not support the research hypothesis but rather indicated that the aged widowed appear to interact with friends more than aged married individuals. This finding supports Townsend's (1957) earlier research on the aged but is inconsistent with results from investigations by Berardo (1967) and Booth (1972). The finding from this research as well as Townsend's study can perhaps be interpreted in the following way. The widowed, conditioned to but deprived of interaction with a spouse, are forced to look outside the nuclear family for the satisfaction of their social needs while married individuals have less inclination to do so. Friend interaction appears to be a viable outlet for meeting these needs. Moreover, assuming the applicability of the exchange framework to this social setting, friend interaction is actively sought out and maintained by those with sufficient resources to make the relationship rewarding to both. The aged widowed, unencumbered with the obligations associated with married life, have more free time which enables them to get the greatest possible mileage from the resources they do possess.

PAGE 99

92 The inconsistency among findings in this as well as the two preceding hypotheses certainly indicates that future research is necessary to clarify the matter. However, earlier results by Blau (1961) provide information from which a possible explanation Tor these contradictory findings is suggested. She found that the isolation of the elderly is greatly influenced by the social structural context in which older individuals are situated. This is to say, higher rates of social interaction are associated with structural situations in which greater opportunities exist for interaction with individuals of the same marital status. Unfortunately, no controls for this factor were introduced in this or earlier research. Thus, lack of comparability of structural conditions among samples investigated could obfuscate the relationships among the variables examined, resulting in the observed inconsistency. To explain further, samples drawn in Townsend's study and the present research may have selected individuals from structural settings composed of greater numbers of widowed persons thereby providing the aged widowed with greater opportunities to interact among themselves. On the other hand, Berardo's (1967) and Booth's (1972) sample may have consisted of disproportionate numbers of married individuals thereby offering aged married greater possibilities for interacting with one another. This interpretation is consistent with the findings from the respective studies. Berardo's research focused on aged people in a rural county

PAGE 100

93 in Washington whereas Townsend's study examined 200 aged individuals from Bethnal Green, London. In Berardo ' s rural sample the elderly, particularly widowers, situated on isolated farms, perhaps were not afforded the structural opportunities for interacting with those individuals in their same age and sex grade as were the respondents in Townsend's research who resided in a suburb of London. Furthermore, this interpretation is supported by Booth who studied noninstitutionalized adults 45 years old and older residing in Lincoln and Omaha, Nebraska. The sex ratio in the target population was .89, indicating a large number of single persons. Assuming a disproportionate number of female widowed, Booth's finding indicating widowers are much less likely to have friends than widows lends additional support to this explanation. Additionally, this problem of uncomparability could stem from either the strategy or implementation of the research designs. The fourth hypothesis examined the interaction of the aged in religious organizations. Earlier studies by Berardo (1967) and Townsend (1957) suggest that widows are more likely to interact in religious organizations than widowers or married individuals. Widowers have been shown to be the most isolated in this respect (Pihlblad and Adams, 1972). However, in order to be consistent with the earlier hypotheses and the overall design of the study, this hypothesis did not distinguish by sex among the aged widowed. This control was extended in the ninth hypothesis.

PAGE 101

91 The aged widowed were hypothesized to interact less frequently in religious organizations than' aged married individuals. No significant differences between the rates of interaction in religious organizations between the two groups of aged was found. The Kolmogorov-Sinirnov value was 1.71. The hypothesis was not supported. It is impossible to determine in this hypothesis if controls for sex among the widowed would have yielded findings consistent with earlier studies. Data in the fifth hypothesis suggested that the aged widowed interact less frequently in formal organizations than aged married individuals. The difference between the two groups was significant at the .05 level but, as in the hypothesis pertaining to friendship interaction, the relationship was not in the hypothesized direction and was therefore not supported. The KolmogorovSmi rnov value was 6.95. This finding is incongruous with earlier research by Berardo (1967), Bock (1972), Bock and Webber (1972), and Harvey and Bahr (1974), who all indicate that the widowed, particularly males, are less likely to interact in formal organizations than married individuals. Problems concerning comparability of samples and research design not withstanding, simple explanations of the differences in research findings are not readily available. Perhaps a plausible explanation might lie that the diminution of sources of interaction for the widowed precipitated by the death of a spouse predisposes the elderly

PAGE 102

95 in this sample to turn to formal organizations to meet social needs formerly satisfied in interaction with the husband or wife. Additionally, the widowed, unfettered with the social obligations which accompany marriage, have more time to devote to such relationships. Therefore the available time stemming from widowhood operates to push the aged widowed into situations where they are seeking social satisfaction as well as pull them into organizational settings where they may meet these social needs. Furthermore, surplus time, in addition to acting as a resource itself, facilitates the exchange of other resources that the aged widowed may possess. A brief summarization of the design and findings of the first five hypotheses is in order. The hypotheses were designed to test if significant differences existed between aged widowed and aged married individuals in terms of their interaction in kin groups, friend groups, religious organizations, and formal organizations. If significant differences were found the second five hypotheses were designed to determine if the differences diminished upon controlling for social resources, thus indirectly putting to test the exchange conceptual framework. However, in three of the five hypothesized relationships, those pertaining to interaction with kin living nearby, kin living far away, and interaction in religious organizations, no significant differences between aged widowed and aged married individuals were found. Moreover,

PAGE 103

96 among the two hypotheses investigating social interaction in friend groups and formal organizations, significant differences between the two groups of elderly were found but the empirical findings were not in the hypothesized direction. This indicates that five hypotheses, developed inductively, were not given empirical support. As suggested in the earlier discussion of the hypotheses, the inconsistent findings could result from many things. Most likely are problems with research and sampling design and comparability of populations. Whatever the source of the problem, the preceding inconsistencies warrant further research . The research design specifies that the second five hypotheses build upon the earlier five by determining if significant differences between aged widowed and aged married individuals are a function of control over resources. Because none of the first five hypotheses were significant the overall expectations surrounding the implementation of the research design have to be modified. However, this fact does not mean that the original design for the final five hypotheses has been completely undermined. It means only that the effect of controlling for resources is less noticeable. Whether the widowed or married interact more in certain social settings is less important than whether differences between the two can be diminished by controlling for variables assumed to be viable exchange resources. With this in mind, the discussion now turns to an analysis and summarization of the final five hypotheses.

PAGE 104

97 The sixth hypothesis compared the measure of association of rates of interaction with nearby kin between aged married and widowed individuals before and after controlling individually for six variables assumed to act as exchange resources. The measure of association of rates of nearby kin interaction between the two groups of aged before controlling for the resource was quite small. The gamma score was -.03. However, this was expected in view of the first hypothesis which found no significant differences between rates of interaction with nearby kin for the two groups of elderly respondents. Comparisons of the total gamma with the partial gammas, which represented controls for the six resources, revealed that none of the variables appeared to be important as social exchange resources. This is to say that controls for the six variables resulted in increased rather than decreased gamma values for all levels of all control factors except for individuals claiming health problems. The value of the measure of association for this control level was .01. However, the standard error of the sampling distribution which includes this gamma was so large that the sample statistic may not reasonably be assumed to reflect the population parameter. The increased gamma values in all other levels of all control variables resulted because the application of controls served to accentuate the measure of association within the established control levels.

PAGE 105

98 Controls for sex suggested that the widowed tend to interact more frequently with nearby kin than married individuals. The gamma values for the two groups are -.5 for males and -.17 for females. This finding conflicts with earlier research by Berardo (1967), Bock and Webber (1972), and Townsend (1957) which suggests that kin interaction declines with the death of a spouse. However, the findings from the present study, particularly for widowers, are suspect because of the small sample size (N=13 widowers) and the large standard error for elderly males (.41). Income was the only other variable controlled which is worthy of extended attention. Controls for income increased the measure of association calculated between widowed and married respondents for both those making more and less than $4500. The difference between the two gammas was large enough to be significant at the .01 level. Moreover, among the wealthier respondents married individuals tended to interact with nearby kin more frequently while among the poorer respondents the widowed interacted more often with nearby kin. Respective gamma values for the two groups were .53 and -.24. The high measure of association computed for wealthy individuals probably resulted from the disproportionate number of poor widowed. Additionally, it should be indicated that there was a tendency for the respondents to cluster in one category, usually the "weekly" class, resulting in high standard errors and as a result, gamma values which need to be interpreted with caution.

PAGE 106

99 The seventh hypothesis compared the measure of association of rates of interaction with far away kin groups between aged widowed and aged married individuals before and after controlling for the six resource variables. As in the preceding hypothesis, the measure of association computed before the resources were held constant was quite small, suggesting the need to discuss with circumspection the implications which might be drawn after examining the impact of statistical controls on this relationship. The gamma value was -.02. However, none of the controls extended revealed a diminution in the strength of the relationship between widowed and married individuals. On the contrary, these values increased for all levels of all control factors. Moreover, controls for race resulted in increased gamma values for both black and white respondents which were so different as to be statistically significant at the .01 level. This statistical interaction necessitates the individ ual interpretation of the gamma values computed for different levels of the race control variable. Among black respondents the married individuals were more likely to interact with far away kin than were widowed respondents while among whites the widowed were more likely to interact with kin living far away. These two relationships are confusing and do not easily lend themselves to explanations emanating from the exchange or any other conceptual framework. The situation becomes even more perplexing when it

PAGE 107

100 is remembered that the opposite relationships were revealed in the first hypothesis for the two race control categories. That is, black widows were more likely to interact with nearby kin than black married individuals and white widows were less likely to interact with kin living nearby than were white married individuals. Future research is needed to clarify the relationships existing among propinquity, kin interaction, race, and marital status. Tn summary, statistical controls for sex, age, education, race, health status, and income failed to provide empirical support in the first two hypotheses for the exchange conceptual, framework. Moreover, these controls generally acted to increase rather than decrease virtually all gammas calculated for different levels of the control variab les . Measures of association between the rates of interaction with friends for aged widowed and aged married individuals were computed in the eighth hypothesis while holding constant the six resource variables. Controls for none of the variables resulted in diminutions in the measure of association between the rates of interaction with friends for the two groups of aged individuals. To this end, earlier theoretical as well as empirical research (Blau, 1961; Griffiths et al., 1971; Lopata, 1973, McCall, 1966; Rosow, 1967; Streib, 1972; and Thibaut and Kelley, 1959) indicating the role of financial assets in friend interaction is inconsistent with findings from this research.

PAGE 108

mi Earlier research posits a direct relationship between educational status and friendship satisfaction (Lopata, 1973). Controls for education revealed no relationship between educational status and friend interaction in this research. The finding suggesting the impact of sex on circumscribing the interaction with friends among the aged was discerned in earlier research by Berardo (1967) and Booth (1972) who indicate that the widowed are more isolated in terms of friendship interaction than married individuals. Moreover, they note that the discrepancy in terms of rates of friend interaction between widowers and married individuals is particularly great. Furthermore, Pihlblad and Adams (1972) suggest that widowhood decreases friend interaction for men but not women. The control for sex in this research revealed that the gamma value is reduced for females but not males. Moreover, in both groups the widowed tend to interact more with friends than married individuals, thus providing support for Townsend's (1957) earlier research. However, the large standard error calculated for males (.21) suggests that the observations in this research must be viewed with caution. It is entirely possible that the particularly disadvantaged position occupied by widowers suggested in research by Berardo (1967), Booth (1972) and especially Pihlblad and Adams (1972) does exist for the population but due to sampling error was not reflected in this sample.

PAGE 109

102 The ninth hypothesis compared the measure of association of rates of interaction in religious organizations between aged widowed and aged married individuals before and after controlling for the six resource variables. Although religious organizations can be subsumed under the broad genre of formal organizations, interaction in religious organizations was singled out because of the attention it received in earlier research and the particular opportunities available for interaction in this area, especially for widows (Berardo, 1967). The strength of the relationship between interaction rates in religious organizations for aged widowed and aged married individuals, as measured by gamma, was -.15. Controls for income yielded a partial gamma score of -.12 which indicates a very modest diminution in the measure of association computed between rates of interaction in religious organizations for elderly married and widowed individuals. This finding might perhaps be interpreted as lending support to the consideration of income as an exchange variable. However, the relatively large standard error of the partial gamma further suggests that this interpretation be viewed with skepticism. The application of controls for the other five potential rewards provided no evidence to support the viability of the exchange conceptual framework in analyzing the interaction of the elderly in religious organizations. However, controls for sex yielded findings consistent with earlier research in the area.

PAGE 110

103 The role of sex in circumscribing interaction of the aged in religious organizations is conclusively documented by earlier empirical investigations (Berardo, 1967; Pihlblad and Adams, 1972; and Townsend, 1957). The present study lends support to these earlier findings. Widows were found to interact more frequently in religious organizations than married women and among males, married individuals were more likely to interact in church-related activities than widowers. Furthermore, the widowed as a group tended to interact more in religious organizations than married Individ u a 1 s . Measures of association of rates of interaction in formal organizations between aged widowed and aged married individuals were accomplished in the final hypothesis while holding constant the six control variables. The gamma measure of association between the two groups of aged individuals was .32, the largest total gamma computed in this research project. However, the strength of this gamma must be viewed with caution because the sample size for formal organizations was much smaller than the three other areas investigated. None of the factors controlled reduced the measure of association between the two groups of elderly. Earlier research by Berardo (1967), Bock (1972) and Harvey and Bahr (1974) suggested that the widowed, particularly males, are less likely to interact in formal organizations than married individuals. Pihlblad and Adams (1972) indicate that widowers are less likely to interact in formal

PAGE 111

104 organizations than married men while there were no differences between married and widowed women. Controls for sex in this researcli provided findings which were largely consistent with this earlier work. Married individuals in the present as well as the earlier studies were found to interact with greater frequency in formal organizations. However, in the current study married women were found to be slightly more predisposed than widows to interacting within formal organizations while among males widowers were discovered to be more active. Precautionary statements are again needed. The sample size for widowers is so small (N=2) and the standard error so great (.43) that the observation pertaining to widowed males is meaningless. Previous research by Lopata (1973) suggests a direct relationship between educational status and participation in formal organizations. Controls for education did not discern this rel at ionship in this research. Perhaps income was even more conspicuous by its absence. Previous research by Lopata (1973) also posits a direct relationship between financial status and participation in formal organizations for widows. Moreover, Harvey and Bahr (1974) suggest that organizational affiliation for the widowed is circumscribed by socioeconomic status more than widowhood status. However, this finding was not borne out in the present research. To briefly summarize, this research was designed to compare the rates of social interaction of aged widowed

PAGE 112

105 individuals with those of aged married persons. This comparison was developed in order to provide a substantive research area in which the applicability of the exchange conceptual framework could be explored. Comparisons of interaction rates between the two groups of elderly were made both before and after controls for six variables were extended. Although modest differences between the two groups of elderly in terms of social interaction were revealed, these controls were still made in an attempt to determine their viability as exchange resources. Empirical support for this theoretical model was not provided in this research. Perhaps other conceptual frameworks would better serve to interpret the social relationships of the aged . Symbolic interact ionist models appear as logical alternative frameworks; however earlier research by Harvey and Bahr (1974) provided no empirical support for either role theory or self theory. Another viable theoretical approach might utilize subcultural differences to explain interaction rates of the two groups of elderly. This mode] would suggest isolation of the elderly to be a function of marital status when that factor operates to create subcultures which provide a social structural setting in which physical as well as emotional needs of the elderly are met. The formation of these subcultures would be contingent upon sufficient numbers of individuals occupying the same relative age, sex, and marital status grades. This theory is

PAGE 113

106 consistent with the earlier observations of the extreme isolation of widowers which contrasts dramatically with the active involvement of widows in kin groups, friend groups, formal organizations, and religious organizations. The small number of widowers found in most social settings precludes the development of subcultures of elderly widowed men while the large number of widows living in close proximity to one another is ideal for the formation of subcultures for elderly widowed females. Weaknesses and Strengths Unfortunately, more can be said about the weaknesses of this research than can be stated concerning its strengths. Perhaps the most glaring deficiency of this study surrounds the inability to operationalize and measure intrinsic resources, variables which probably play a pivotal role when examining kin and friend interaction from an exchange perspective. This problem hinges on several factors, not the least of which are the relatively crude measurings techniques currently available to social scientists. Additionally, the identification of viable exchange resources has not progressed well primarily because of insufficient empirical investigations in this area. Another weakness of this study involves the absence of controls for sex in the first five hypotheses. Earlier research efforts made this distinction which resulted in valuable findings. However, this sin of omission is justified

PAGE 114

107 in the following manner. At the time of the conception of the research design the overriding concern of the investigator was to locate a researchable area in which the viability of the exchange conceptual framework could be examined Earlier studies in the area indicate that the widowed as a group are more isolated in kin groups, friend groups, and formal organizations than married individuals. The relationship between marital status and social isolation was then perceived as an arena in which to test the applicability of exchange theory as a meaningful conceptual framework for interpreting the social isolation of the aged. Therefore, it was decided that married and widowed individuals were to he compared in the first five hypotheses in terms of rates of social interaction to determine if significant differences existed between the two. Following these observations, controls for six variables, sex included, were implemented to determine if controls for these variables resulted in a diminution of these differences thus providing empirical support for the interpretation of these factors as exchange resources. Perhaps another criticism could be leveled at this research because the empirical findings were interpreted only from one conceptual framework, the exchange framework. However, earlier research employed role theory (Berardo, 1967) and role theory and self theory (Harvey, 1973) in examining the social interaction of the aged. Therefore, it was felt

PAGE 115

108 that the exchange framework should be examined in this research context to determine any applicability it might have . The final criticism concerns the controls employed in the research design. Because of the size of the sample and the fact that the research was ex post facto , usual controls applied in an experimental situation were inapplicable. Therefore statistical controls provided the only viable alternative. However, the use of this type of control represents a rude compromise in any research setting. Future researchers attempting to explore the same area would be well advised to select a sample large enough to permit the application of matching techniques. Additionally, for appropriate research settings, and the present research is not a case in point, the random assignment of respondents or subjects to control and experimental groups is an effective method for holding constant confound ing factors. The overriding strength of this study is the fact that it is, to the author's knowledge, the first attempt to interpret previous research findings and examine new research data on the social interaction of the aged from the perspective of the exchange conceptual framework. To this end, this research made extensive use of statistical controls in order to ascertain the impact of certain variables as exchange resources. The notion of holding constant confounding variables when examining the relationship among

PAGE 116

109 two or more factors is exceedingly useful in social research because of the tremendous complexity of social relationships. An earlier criticism leveled at the controls utilized in this research primarily concerns problems with the sophistication of the techniques employed and not with the fact that controls were applied. finally, the area investigated in this research has not been studied enough. This is unfortunate particularly because of the growing size and increased importance of the elderly as a viable political, social, and economic unit in our society. This research will hopefully add to the existing knowledge of the social relationships of the aged. Notes for Future Rese arc h Notes for future research center on three suggestions. First, hypotheses in this study built deductively from earlier research (Berardo, 1967; Blau, 1961; Bock, 1972; Bock and Webber, 1972; Griffiths et al . , 1971; Harvey and Bahr, 1974; Lopata, 1973; Lowenthal, 1964; Pihlblad and Adams, 1972; Rosow, 1967, and Townsend, 1957) were not supported. The inconsistency of research findings between this and earlier studies is certainly a call for more research in this area to determine the actual relationships among the social variables examined. Second, most of the earlier research in this area is not associated with an explicit conceptual framework; Berardo (1967) and Harvey (1973) are

PAGE 117

110 the notable exceptions. Future research needs to make greater use of these models in order to guide the research effort and provide some degree of commonality among studies investigating the same phenomenon. A final observation concerns the application of exchange theory to an analysis of social relationships. As reported earlier, data in this research effort did not provide empirical support for the exchange conceptual framework. However, it must be remembered that variables considered in this research as viable candidates for exchange resources were all extrinsic in nature. intrinsic resources were not considered. This did not result from an oversight but rather from the fact that the available data which were analyzed in this study provided information only on social structural variables such as the age, sex, race, and income distribution of groups of elderly widowed and married individuals. information on intrinsic resources was not available. This is unfortunate, because it seems that if the notion of reciprocity has any validity at all in terms of explaining social relationships involving kin and friends, it would most probably involve the exchange of intrinsic rather than extrinsic resources. Future research efforts which attempt to examine the applicability of the exchange framework, particularly when investigating kin, friend, and intimate interpersonal relationships, should attempt to tap intrinsic resources. To

PAGE 118

Ill this end, attitude scales probably offer the greatest prom ise of available methodological techniques for measuring these rewards.

PAGE 119

BIBLIOGRAPHY Adams, Bert 1968 "The middle -class adult and his widowed or still -married mother." Social Problems 16 (Summer) : 50-59. Aldous , Joan 1970 "Strategies for developing family theory." Journal of Marriage and the family 32 (May): 2 5 0-257. Becker, Howard 1956 Man in Reciprocity. New York: Praeger. Berardo, Pelix M. 1967 "Social adaptation to widowhood among a rural -urban aged population." Washington Agricultural Experiment Station Bulletin 689. December. 1968 "Widowhood status in the United States: Perspective on a neglected aspect o 1 the family life cycle." The Family Coordinator 17 (July): 191-205. 1970 "Survivorship and social isolation: The case of the aged widower." The family Coordinator 19 (January): 11-25. Blalock, Hubert M. 1972 Social Statistics. New York: McGraw-Hill. Blau, Peter M. 1964 Exchange and Power in Social life. New York John Wiley and Sons. Blau, Zena Smith 1961 "Structural constraints on friendship in old age." American Sociological Review 26 (June) : 429-439. Blood, Robert 0. and Robert L. Hamblin 1958 "The effect of the wife's employment on the family power structure." Social forces 26 (May): 347-352. 112

PAGE 120

113 Blood, Robert 0. and Donald M. Wolfe 1960 Husbands and Wives. New York: Free Press. Bock, E. Wilbur 1972 "Aging and suicide: The significance of marital, kinship, and alternative relations." The Family Coordinator 21 (.January): 71-80. Bock, E. Wilbur and Irving L. Webber 197 2 "Suicide among the elderly: Insulating widowhood and mitigating alternatives." Journal of Marriage and the Family 34 [February): 24-31. Booth, Alan 1972 "Sex and social participation." American Sociological Review 3 7 (April): 183-192. Clausen, John and Melvin Kohn 1954 "The ecological approach to social psychiatry American Journal of Sociology 60 (September]: 14 0-151. Dub 1 in , Lou is 1 . 1965 Suicide. New York: Ronald Press. Durkheim, Emile 1951 Suicide. New York: Free Press. Duval 1 , Evelyn 1967 Family Development, 3rd ed . New York: J. B. Lippincott Company. Edwards , John N . 1969 "Familial behavior as social exchange." Journal of Marriage and the Family 31 (August): 518-526. Far is, Robert I: . L. 1954 "Cultural isolation and the schizophrenic personality." American Journal of Sociology 40 (September) : 155-164. G] ick, Paul C. 1957 American Families. New York: John Wiley and Sons. Couldner, Alvin W. 1960 "The norm of reciprocity." American Sociological Review 25 (April): 161-17.

PAGE 121

114 Cove, Walter R. 1972 "Sex, marital status, and mortality." Unpublished paper presented to the Pacific Sociological Association, Portland, Oregon, April 12-15. Griffiths, Kenneth, 0. William Farley, W. Paul Dean, and Lewis L. Boon 1971 "Socioeconomic class and the disadvantaged senior citizen." Aging and Human Development 2 (November) : 288-295. Harvey, Carol D. 1973 Widowhood: Predicting Social Psychological Consequences . Di ssertat ion , Was hi ngton State University. Harvey, Carol D. and Howard M. Bahr 1974 "Widowhood, morale, and affiliation." Journal of Marriage and the family 56 (February J : 97-106. Heer, David M. 19 5 8 "Dominance and the working wife." Social Forces 36 (May): 341-347. 1963 "The measurement and bases of family power: An overview." Marriage and Family Living 25 (May): 133-139. Hill, Reuben and Donald A. Hansen 1960 "The identification of conceptual frameworks utilized in family study." Marriage and Family Living 22 (November): 299-311. Homans , George C . 19 58 "Human behavior as exchange." American Journal of Sociology 65 (May): 597-606. 1961 Social Behavior: its Elementary Forms. New York: Harcourt, Brace, and World. Jaco, E. Cartly 1954 "The social isolation hypothesis and schizophrenia." American Sociological Review 19 (October): 567-577. Jacobson, P a u 1 1959 American Marriage and Divorce. New York Rinehart and Company. Kohn, Melvin and John A. Clausen 1955 "Social isolation and schizophrenia." American Sociological Review 20 (June): 2 6 5 2 7 5 .

PAGE 122

115 Kutner, Bernard, David Fanshel, Alice M. Togo, and Thomas S . L an g e r 1956 Five-Hundred Over Sixty. New York: Russell Sage. Leslie, Gerald R. , Richard F. Larson, and Benjamin L. Gorman 1973 Order and Change. New York: Oxford iln i vers Lty P ress . Lev i St rauss , Claude 19 57 "The principle of reciprocity." In Lewis A. Coser and Bernard Rosenberg, eds., Sociological Theory. New York: MaciniJlan. Linton, Ralph ] 9 5 9 "The natural history of the family." In Ruth N. Anshcn, ed., The Family: Its Function and Destiny. New York: Harpers Litwak, Eugene 19(A) "Occupational mobility and extended family cohesion." American Sociological Review 2 5 (February) : 9-21 . 1960 "Geographic mobility and extended family cohesion." American Sociological Review 25 (June) : 382-594 . Lopata, Helena Z. 1969 "Loneliness: Forms and components." Social Problems 17 (Fall): 248-262. 1973 Widowhood in an American City. Cambridge Mass.: Schenkman Publishing Company. Lowenthal, Marjorie Fiske 1964 "Social isolation and mental illness in old age." American Sociological Review 2 9 (February) : 54-70. Ma 1 i nowsk i , B ron i s 1 aw 1932 ('rime and Custom in Savage Society. London: Paul, Trench, Trubner. Marris, Peter 19 5 8 Widows and Their Families. London: Rout ledge and Kegan Paul. Mauss, Marcel 19 54 Essai sur le don. Translated by I. Cunnison as The Gift: Forms and Functions of Exchange in Archaic Societies. Glencoe, 111.: Free Press.

PAGE 123

1J6 McCal 1 , Michael M. 1966 "Courtship as social exchange: Some historical comparisons. In Bernard Farber, ed. , Kinship and Family Organizations. New York: John Wiley and Sons. McKain, Walter C. 1972 "A new look at older marriages." The Family Coordinator 21 (January): 61-69. Nye, Ivan F. and Felix M. Berardo 1966 Emerging Conceptual Frameworks in Family Analysis. New York: Macmillan Company. Parsons, Talcott and Robert F. Bales 195 5 Family, Socialization and Interaction Process. Clencoe, 111.: Free Press. Pihlblad, C. T. and David L. Adams 1972 "Widowhood, social participation and life satisfaction." Aging and Human Development 3 (November): 323-330. Pollak, Otto 1948 Social Adjustment in Old Age. New York Social Science Research Council. Re iss , Paul J 1962 "The extended kinship system: Correlates of and attitudes on frequency of interaction." Marriage and Family Living 24 (November): 3 3 3 3 3 9 . Robins, Lee N., and Miroda Tomanec 196 2 "Closeness to blood relatives outside the immediate family." Marriage and Family Living 24 (November): 340-346. Rodger s, Roy 11 19 64 Rodgers, Roy 11. 1964 "Toward a theory of family development." Journal of Marriage and the Family 26 (August): 262-270/ and Reuben Hill "The developmental approach." In Harold Christensen, ed., Handbook of Marriage and the family. Chicago: Rand McNally. Rogers, Fvcrett M. and A. Eugene Havens 1960 "Prestige rating and mate selection on a college campus." Marriage and Family Living 22 (February): 55-59. Rosow , Irving 1967 Social Integration of the Aged. New York Free Press .

PAGE 124

117 Sclltiz, Claire, Marie Jahoda, Martin Deutsch, and S. W. Cook 1959 Research Methods in Social Relations. New York: Henry Holt and Company. Sheldon, Henry I). 1960 "The changing demographic profile." in Clark Tibbitts, ed., Handbook of Social Gerontology. Chicago: The University of Chicago Press . Siege 1 , Sidney 1 9 5 6 Nonparamet ric Statistics for the Behavioral Sciences. New York: McGraw-Hill Book Company . Si mm el , Georg 195 The Sociology of Georg Simmel. Translated by Kurt H. Wolff. New York: free Press. Simpson, George and J. Milton Yinger 1972 Racial and Cultural Minorities 4th ed. New York: Harper t\ Row. Sorokin, Pitirim, Carle C. Zimmerman, and C. J. Galpin 1951 A Systematic Source Book in Rural Sociology, 2. Minneapolis: University of Minnesota Press . S t r e i b , Gordon F . 1972 "Older families and their troubles: Familial and social responses." The Family Coordinator 21 (January): 5-19. Sussman, Marvin and fee Burchinal 1962a "Kin family network: Unheralded conceptuali zations of family functioning." Marriage and Family Living 24 (August): 231-240. 1962b "Parental aid to married children: Implications for family functioning." Marriage and Family 24 (November): 320-332. Tec, Nechama and Ruth Granick 19 59 "Social isolation and difficulties in social interaction of residences in a home for the aged." Social Problems 7 (Summer); 226232. Thibaut, John W. and Harold H. Kelley 1959 The Social Psychology of Groups. New York: John Wiley and Sons. Thurnwald, Richard 1932 Economics in Primitive Communities Oxford University Press. L ondon :

PAGE 125

118 Tibbitts, Clark 1960 "Origins, scope, and fields of social gerontology." In Clark Tibbitts, ed., Handbook of Social Gerontology. Chicago University of Chicago Press. Townsend , Peter 1957 The Family Life of Old People Routledge and Kegan Paul. London U. S. Bureau of the Census, Statistical Abstract of the United States: 1969. (90th edition.) Washington D.C. Statistical Abstract of the United States: 1972. (93d edition.) Washington D.C. U. S. Census of Population: 1950. Vol. II, Characteristics of the Population, Part 1, United States Summary. , U. S. Census of Population: 1970. Vol. I, Characteristics of the Population, Part 1, United States Summary, Section 1. Waller, IVIllard 193 7 "The rating and dating complex." American Sociological Review 2 (October) : 728-734. Zetterberg, Hans L. 1965 On Theory and Verification in Sociology, 3rd ed. New York: Bedminster Press.

PAGE 126

BIOGRAPHICAL SKETCH Marc L. Petrowsky was born June 29, 1943 in Topeka, Kansas. In May, 1970, lie received the degree of Bachelor of Arts from Southwestern State College, Weatherford, Oklahoma. In June, 1971, lie was graduated from the University of Florida, Gainesville, with a Master of Arts degree in sociology. He is married to Iris Godley Petrowsky. 119

PAGE 127

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. T elix M. Rerardo, Chairman rofessor of Sociology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. "y Gerald R. Leslie Professor of Sociology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. George J./Wa-rheit Associate Professor of Psychi atry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. / Benjamin L. Gorman Professor of Sociology I certify that 1 have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. .'Man G. Agresti Associate Professor of Statistics

PAGE 128

This dissertation was submitted to the Graduate Faculty of the Department of Sociology in the College of Arts and Sciences and to the Graduate Council, and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. June, 197 5 Dean , Graduate School

PAGE 129

UNIVERSITY OF FLORIDA 3 1262 08666 245 8