Object relations and social networks : correlates of adjustment during college

MISSING IMAGE

Material Information

Title:
Object relations and social networks : correlates of adjustment during college
Physical Description:
viii, 157 leaves : ; 28 cm.
Language:
English
Creator:
Fortney, Robert Peter, 1953-
Publication Date:

Subjects

Subjects / Keywords:
Interpersonal relations   ( lcsh )
Social groups   ( lcsh )
Student adjustment   ( lcsh )
Clinical Psychology thesis Ph. D
Dissertations, Academic -- Clinical Psychology -- UF
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1984.
Bibliography:
Includes bibliographical references (leaves 151-155).
Additional Physical Form:
Also available online.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Robert Peter Fortney.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 030528836
oclc - 11795337
System ID:
AA00025836:00001

Full Text















OBJECT RELATIONS AND SOCIAL NETWORKS:
CORRELATES OF ADJUSTMENT DURING COLLEGE














BY

ROBERT PETER FORTUNE













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 1984































Dedicated

to

that pursuit
which Aldous Huxley termed
a quest for grace














ACKNOWLEDGMENTS

I would like to express my gratitude to my chairman,

Dr. Hugh Davis, whose ironic vision and well-timed comments and support have guided me through this project; the members of my supervisory committee, Dr. Louis Cohen, Dr. Jacque Goldman, Dr. Eileen Fennell, and Dr. Otto Von Mering, who each contributed to the ideas found in this paper and whose object representation will remain with me throughout my professional career; Mark Waugh and Rebecca (Cooley) Behrends, who kept me current with the research on object relations theory; Sandi Jacot and Kathy Toohey for their care and precision in collecting the data; Steve Billingsley, who was willing to barter his computer consultation time; Wesley Seidel, who taught me the rudiments of word processing and who furnished me with the necessary hardware; Sigmund Freud, who cautioned that the discipline he founded should never become a healing cult.and Teresa, who, after providing invaluable assistance in the most tedious phases of this study and enduring a good deal of relative deprivation during the past seven and a half years, still knows what will not fade away.















TABLE OF CONTENTS


PAGE
ACKNOWLEDGEMENTS . iii

LIST OF TABLES . . vi

ABSTRACT . .. x

CHAPTER

ONE INTRODUCTION . 1

Overview *
Theoretical Considerations ....... 4 Object Relations 7 Social Networks 19
Central Thesis: An Interactionist Model 26 Hypotheses 31
1. Interactional Effects .. 33
2. Univariate Comparisons 33 3. Outcome Measures .. .. 34
4. Life Change . 35
5. Qualitative Factors and
ICL Dimensions 35 6. ICL Quadrants . 35

TWO METHOD . .. ..... 36

Subjects o 36
Measures . 36
Category A. Measures of Life Change
and Demographics 36
Category B. Primary Measures o 40
Category C. Measures of Mental Health 46
Procedure 49

THREE RESULTS . 51

Interrater Reliability
and Normative Comparisons ... 51
Interactional Effects ...... 55
Univariate Comparisons 59 Outcome Measures .. 96


iv -









Life Change 97
Qualitative Factors and ICL Dimensions . 98 ICL Quadrants . . . . . 100

FOUR DISCUSSION . . . . . 103

The Hypotheses Reviewed . . . 103 Two Conclusions . . . 108
The Reevaluation of Conceptual Level ..109 The Reevaluation of Network Density . 114 Closing Remarks ....... .118

APPEND ICES

A. NETWORK DENSITY SCORES--AN EXAMPLE ...121

B. CONSENT FORMS . . . . 124

C. DATA COLLECTION INFORMATION . . . 134

Instructions Used in Data Collection 134
Questions Asked by the Subjects and
the Answers Given by the Research
Assistants . . . 138

D. DEMOGRAPHICS QUESTIONNAIRE . . . 139

E. RELIABILITY AND NORMATIVE DATA . . 141

Reliability of the Descriptions of Parents 141
Norms for the Descriptions of Parents
--On Mothers . . 142
Norms for the Descriptions of Parents
--On Fathers . . . . . 143
Norms for Social Networks . . . 144 Norms for the CSRE ... 145
Norms for the POMS 146
Norms for the HSCL 147
Norms for the Love Scale .148
Norms for the ICL . . . . 149

REFERENCES 151

BIOGRAPHICAL SKETCH . . . . . . 156















LIST OF TABLES
TABLE

PAGE

1. Means (M) and Standard Deviations (SD) of
Selected Variables for Three Groups of
Averaged Conceptual Level (CL) of
Object Representations 56

2. Means (M) and Standard Deviations (SD) of
Selected Variables for Three Groups of
Total Network Density 58

3. Summary of the Correlations Between the Outcome
Measures and the Variables of Life Change, Total Network Size, Total Network Density,
and Averaged Conceptual Level 60

4. Correlation Matrix for the Variables of Life
Change, Total Network Size, Total Network
Density, and Averaged Conceptual Level . 61

5. Correlation matrix for the Outcome Measures .. 62

6. Correlations Between the Subscales of the
Outcome Measures and the Conceptual
Level Scores 63

7. Correlations Between the Subscales of the
Outcome Measures and the variables of Life
Change, Total Network Size, and Total
Network Density 65

8. Correlations Between the Subscales of the
Outcome Measures and the Variables of
Family/Relative Network Density, Friendship
Network Density and Nuclear Family-Friendship
(HF-F) Boundary Density 68

9. Summary of the Differences on Nuclear FamilyFriendship (NF-F) Boundary Density for the
Subjects who Lived at Home versus the
Subjects who Lived Away from Home .... 70




vi -









10. Correlations Between the Qualitative Factors
of "Parent as Striving" and "Parent as
Nurturant and the ICL Dimensions
(DOM and LOV) . . . . . 73

11. Correlations Between the Subscales of the
Outcome Measures and the Variables of
"Parent as Striving" and "Parent
as Nurturant". . . . . .* 74

12. Correlations Between the Subscales of the
Outcome Measures and the ICL Dimensions
(DOM and LOV) . . o o . . 76

13. Correlations Between the Variables of "Parent as
Striving" and "Parent as Nurturant" and the
Variables of Life Change, Total Network Size,
Total Network Density, and Averaged Conceptual
Level o o o o 78

14. Correlations Between the ICL Dimensions (DOM and
LOV) and the Variables of Life Change, Total
Network Size, Total Network Density, and
Averaged Conceptual Level . . . 79

15. Correlation Matrix for the Demographic Variables 81

16. Correlations Between the Demographic Variables
and the Description Length Scores o 82
17. Correlations Between the ICL Control Variables
(Number of Items Checked, NIC, and Average
Intensity, AIN) and the Demographic and
Description Length Variables o o o o 84

18. Summary of the Correlations Between the
Demographic Variables and the Variables
of Life Change, Total Network Size, Total
Network Density, and Averaged
Conceptual Level . . . .. 85

19. Correlations Between the Subscales of the
Outcome Measures and the Demographic
Variables--Part A (Age, Class, Sex,
and SES) o 86

20. Correlations Between the Subscales of the
Outcome Measures and the Demographic Variables--Part B (Employed, Credits,
and Domicile) 87




-vii -









21. Correlations Between the Qualitative Variables
of Object Relations and the Demographic
Variables--Part A (Age, Class, Sex,
and SES) 88

22. Correlations Between the Qualitative Factors
of Object Relations and the Demographic
Variables--Part B (Employed, Credits, and
Domicile) 0 a 89

23. Correlations Between the Description
Length Scores and the Variables of
Life Change, Total Network Size, Total
Network Densiy, and Averaged
Conceptual Level . . . . . 90

24. Correlations Between the Subscales of the
Outcome Measures and the Description
Length Scores . . . . . 91

25. Correlations Between the Descriptions Length
Scores and the Qualitative variables of
Object Relations . . . . . 92

26. Correlations Between the ICL Control Variables
(NIC and AIN) and the Variables of Life
Change, Total Network Size, Total Network
Density, and Averaged Conceptual Level . 93

27. Correlations Between the Subscales of the
Outcome Measures and the ICL Control
Variables (NIC and AIN) . . . . 94

28. Correlations Between the ICL Control Variables
(NIC and AIN) and the Qualitative Variables
of Object Relations . . . . 95

29. Reactivity of the Outcome Measures as Seen in
Linear Regression Models with the Independent
Variables of Life Change, Total Network
Density, and Averaged Conceptual Level of
Object Representations . . . . 97

30. Linear Regression Scores for the Variables
of "Parent as Striving" and "Parent as
Nurturant" and the ICL Dimensions
(DOM and LOV) . . . . . 99

31. Relationships Between the Quadrants of the
ICL and Outcome * . . . . 101




viii -















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

OBJECT RELATIONS AND SOCIAL NETWORKS:
CORRELATES OF ADJUSTMENT DURING COLLEGE By

Robert Peter Fortney

August, 1984

Chairman: Hugh Davis, Ph.D. major Department: Department of Clinical Psychology

The interactionist model assumes that psychological adjustment is a function of the relationship between personality and social variables. The present study attempted to demonstrate that people with lower conceptual levels of object representations would function best in higher density networks, while those with higher conceptual levels would be best adjusted in lower density networks. Deviation from the adaptive equilibrium of this inverse relationship was expected to result in greater subjective distress (e.g., poorer mood, more symptoms) and impairment in the areas of love and work.

The study involved 65 female and 52 male undergraduate students. The subjects completed, in succession, a demographics questionnaire, the College Schedule of Recent Events, a 5-minute written description of their mother and









another of their father (which were scored for conceptual level of object representations, verbal fluency, and the qualitative factors of "parent as nurturant" and "parent as striving"), the Interpersonal Checklist for mother and another for father, the Social Network List and the Social Network Map (which were scored for network density), the Profile of Mood States, the Hopkins Symptom Checklist, and the Scale of Feelings. Also, their grade point average (GPA) was collected at the end of the term.

Although conceptual level was reliably rated and most other variables were consistent with the normative data presented in earlier studies, the central hypothesis was not supported. The expected inverse relationship between conceptual level and network density was not found, nor were larger deviations from this inverse relationship associated with poorer scores on the outcome measures. Contrary to expectations, conceptual level was unrelated to outcome, and higher, rather than lower, density networks were correlated with better mood scores.

Two main conclusions are suggested. First, the

conceptual level variable used in the current study does not adequately summarize the aspects of intrapsychic structure which are sensitive to psychopathology. Second, the adaptive value of network density probably varies more in relation to the psychosocial concerns of the individual than had previously been reported in the literature.














CHAPTER ONE
INTRODUCTION


Overview

The present dissertation was a study of how the

interaction between certain personality and social variables was related to the individual's state of psychological adjustment. The following is an overview of this dissertation's organization.

The main body of the introductory chapter begins with the section entitled "Theoretical Considerations" in which some of the theoretical assumptions underlying this study are delineated. The argument is advanced that personality and social variables must be examined in unison so as to fully account for the vicissitudes of psychological development. The section ends with a synopsis of the study's central hypothesis.

The relevant personality and social variables, whose interaction form the crux of the central hypothesis, are presented in the next two sections. In the first of these sections, entitled "Object Relations", the variables that are derived from the subjects' descriptions of their parents are introduced. The most important of these variables is the conceptual level of the subjects' object representations



I -






2


(i.e. the subjects' image of their parents). The conceptual level, as a developmental variable, is thought to denote the maturational level of the subjects' ability to accurately perceive and act on their environment. After this variable is defined, the research that relate conceptual level to the type and severity of psychopathology is presented. The variables that provide a more qualitative assessment of object representations are described at the end of this section.

In the section entitled "Social Networks" the concept of network density and the research relevant to it are introduced. The total density of the subjects' social network, which is conceptualized in this study as a developmental variable, serves as the social variable in the interactional paradigm.

In the next section the central hypothesis is put forth. This section ends with a discussion of the population, which is college students, and their suitability for this study.

In the "Hypotheses" section, a summary is presented of the six hypotheses tested in this study. Although the primary focus of this study is on the central hypothesis, five other hypotheses are proposed. These five auxiliary hypotheses are designed to ascertain whether or not the population behaved in a predictable manner,






3


The method section forms the second chapter of this dissertation. In this chapter a description of the subjects, the measures, and the procedure used in this study is provided. The section entitled "Measures" is divided into three categories. In Category A, the demographic and life change measures are presented. Category B is devoted to those measures which assess the relevant personality and social variables. The outcome measures are introduced in Category C.

In the third chapter the results of the data analyses are summarized. This chapter is divided into two main components. The first component, which encompasses the section entitled "Interrater Reliability and Normative Comparisons", is designed to show that the results in the current study are reliably rated and are consistent with the results reported in the literature. The second component consists of six sections, with each section corresponding to one of the six hypotheses described at the end of Chapter 1. Thus, the findings pertaining to the first hypothesis are summarized in the result section entitled "Interactional Effects", and the findings for the five auxiliary hypotheses are each summarized in one of the five succeeding sections.

The fourth chapter is a discussion of the unexpected and somewhat challenging results of the current study and the implications these results have for future research on object relations and social networks. In the first section,






4


the status of each of the hypotheses is considered in light of the results, with particular attention focused on the central hypothesis. In the next section, the discrepancy between the results of the central hypothesis and the five auxiliary hypotheses are considered. Two main conclusions are drawn. These conclusions are articulated in greater detail in the next two sections. The first section pertains to object representations while the other section pertains to network density. In the last section, the implications of this study are set forth, with suggestions given for future research.

Theoretical Considerations

Most theories of personality ascribe to the principle of structuralization, whereby interpersonal relationships serve as a model and an impetus for intrapsychic organization (Loevinger, 1976). In particular, psychoanalytic theories articulate how patterns of child-parent interactions during the first several years of development serve as the foundation for intrapsychic representations of self and other. In turn, these "object" representations serve as a structure of expectations to be used in organizing and directing subsequent interactions with people. While it seems obvious that these early patterns of interactions form the foundation for and have a fundamental influence on later personality development, one cannot treat the object representations formed after






5

childhood in a serendipitous fashion. Indeed, the work of Sullivan (1953), Erikson (1950), and developmental theorists have lengthened the "formative years" to include the adolescent period and beyond. These theorists, however, have concentrated on the structural characteristics and requirements of the developing individual, without defining the structural qualities of the person's ongoing interpersonal relationships. In a way, the more the internal representations of objects are specified and delineated, the more the structural nature of the person's social environment needs to be articulated. In sum, both internal and external patterns of interpersonal representations are necessary for describing the process of structuralization, a process which continues throughout life.

Most structural theories imply that the developing

individual perceives and processes only those interpersonal expectations which can be organized by his or her current level of object representations. This "selective inattention", to use Sullivan's term, to parts of the objective situation accounts for a certain degree of stability or, at least, slowness of change of the individual's personality structure. For change in object representations to take place, the individual must come into contact with "pacers", which are "Stimulus objects of a level of complexity" (Loevinger, 1976, p. 309) slightly






6

greater than those represented within the individual. Hence, "as the person maintains contact with and thus masters a pacer, his own level of complexity grows and he is ready for a new, more complex pacer" (Loevinger, 1976, p. 309). In other words, no interpersonal development occurs unless the individual has access to the proper interpersonal structure. At the same time, one cannot fully account for the development of psychopathology in an individual unless the structural inadequacies of that individual's interpersonal world are defined. Thomas and Chess, in describing the necessity of an interactionist model to explain the dynamics of development, have used the terms "goodness of fit" and "poorness of fit".

Goodness of fit results when properties of the
environment and its expectations and demands are
in accord with the individual's own capacities.
When this consonance between organism and
environment is present, optimal development in a
progressive direction is possible. Conversely,
poorness of fit involves discrepancies and
dissonances between the individual's capacities and environmental opportunities and demands, so
that distorted development and maladaptive
functioning occur. Goodness of fit is never an
abstraction, but is always formulated in terms of the values, demands, and expectations of a given
culture or socioeconomic group. (Thomas and Chess,
1980, p. 234)

The present study used this concept of fitness as a schema for positing a pattern of interaction between a person's conceptual level of object representations, both along structural and qualitative dimensions, and the structure of that person's social network. The "fitness" of






7


this pattern of interaction was to account for the mental health (i.e., the degree of personal distress and the ability to love and work) of an individual during a particular period in life, which, in this case, was during college. This study was made possible by recent research in the assessment of object representations and social networks. These lines of research, which have developed in relative independence of each other, have been integrated within the present study in a way which seemed germane to describing some of the developmental tasks faced by individuals who were in college, individuals who were essentially in a prolonged adolescence.

Object Relations

One the impediments to research on object

representations has been the lack of a reliable system for assessing these internal structures. Blatt and his associates at Yale have tried within the past several years to develop such a system. A major stimulus for their system, besides, psychoanalytic theory, has come from the cognitive-developmental theories of Werner and Piaget, among others. Werner has proposed "an orthogenetic principle which states that wherever development occurs it proceeds from a state of relative globality and lack of differentiation to a state of increasing differentiation, articulation, and hierarchic integration" (Werner, 1972, p. 47). He noted that with increasing self-other






8

differentiation, the developing individual gradually becomes less dominated by the immediate concrete situation and less impelled by internal affective states. The person is increasingly able to delay action and to understand the needs and motives of other people. This increasing freedom permits more task mastery and social competence. Similarly, Piaget (1954) has described the sequence through which a child passes in the development of a concept of object. Although Piaget's work has largely been based on the study of children's responses to inanimate objects, the process by which the developing individual comes to relate to human objects must be similar. After all, what was of importance in the present study was not to uncover what personality structure is but rather what it does; ego is not a thing but a process (Loevinger, 1976). An important theme derived from Piaget was how the child in each successive stage of cognitive development has the ability to represent the object in increasingly more abstract terms, less bound by the stimulus properties of the object. As the representation of the object becomes more symbolic, it becomes more permanent and stable. Repeated contacts with the real object are no longer necessary to maintain the coherency of the representation. The room for merger between the self and other representations is, with this increase in boundaries, greatly reduced. The person is less likely to confuse his or her own feelings and needs with






9


those of other people. Hence, as the self-system becomes more distinct from its inner objects, it becomes better equipped, more cohesive and more resilient, to handle the strum und drang of everyday life. Finally within Piagetian theory, complete development of object representations does not take place until adolescence, and, according to some data (Rowe and Marcia, 1980), the completion of this development is by no means assured for every individual.

Blatt, borrowing from Werner and Piaget, used the

notion that object representations "proceed from amorphous, global representations, to a somewhat differentiated emphasis on part properties, to representations which are highly articulated and integrated, and closely correspond to reality" (Blatt, 1974, p. 142) as the basis for describing his conceptual levels of object representations. In general, beginning conceptual levels are based on actions of the (human) object that are related to the individual's need gratification, intermediate levels are based on specific perceptual features of the object, and later levels are based on more symbolic and "conceptual" inner characteristics of the object. More specifically, Blatt, Chevron# Quinlan, and Wein (1981) have created a system for assessing five distinct levels, or milestones, in the development of the concept of object representations. They place these five conceptual levels on a scale from one through nine. The protocols to be scored are drawn from the






10


subject's description of his or her mother and father, though the protocols could also be figures from the Thematic Apperception Test (TAT) stories or human responses on the Rorschach. The conceptual levels of object representations are based on the following five categories. Note that four of the nine conceptual level scores are substages between one the five categories. These four transitional levels transcend the criteria for one category but do not quite fulfill the criteria for the next.

1. Sensorimotor-Preoperational (score 1)

The stimulus object is described in terms of his or her actions as these activities and functions relate to the gratification and frustration of the subject. The description focuses on the direct value that the person has for the subject. There is little sense of the person as a distinct or separate individual, but rather that the person is only an agent who makes the subject feel good or bad.

2. Concrete-Perceptual (score 3)

Here, the person is described as a separate entity but only in literal, concrete, and global terms. Hence, the description is often in physical terms, describing what the person looks like. Little emphasis is placed on the part properties or non-physical attributes of the person, but rather the person is represented as a literal, concrete totality.









3. External Iconic (score 5)

The person is described in terms of his or her activities and functions, but, in contrast to sensorimotor-preoperational representations, these activities and functions are uniquely the other person's and are described as having little or no direct reference to the subject's gratification or frustration. External iconic representations appear to be the normative level for college students in the sample reported by Blatt et al. (1981), with a scaled score standard deviation of 1.5.

4. Internal Iconic (score 7)

The person is described in terms of partial attributes or features that are directed toward an inner dimension. What the person does in terms of functions and activities is not important, rather the focus is on what the person thinks, feels, and values.

In both iconic levels the description of the person is mostly one-sided, unidimensional, and unintegrated. There is little or no recognition of complexity, varying levels, subtlety, or development over time.

5. Conceptual (score 9)

The description at this level integrates all the

previous levels in a way which forms a complex synthesis of the varying levels upon which a person can be experienced and understood. There is often an appreciation of the person in time and of the changes and variations which can






12

occur in that person. Inner dimensions are appreciated in their own right and are contrasted with external qualities. Although the description may incorporate concrete and literal terms and reference may be made to the person's need gratifying and frustrating impact on the subject, these aspects are integrated to form a cohesive and multidimensional image of the person, an image wherein apparent contradictions are resolved in the service of a larger synthesis.

Blatt et al. (1981) are not attempting to describe

intellectual or cognitive development (Piaget's goal), but rather they strive to describe the development of the subject's interpersonal perception. These conceptual levels denote the capacity the individual has to organize, experience, and act upon the world of people. Because most psychological disorders develop as a response to interpersonal conflicts, the individual's conceptual level of representing human objects is hypothesized as being associated with the occurrence of psychopathological symptoms. This is not to say that the development of cognitive abilities and interpersonal perception are unrelated. It would be difficult to imagine someone having achieved a "conceptual" level score without having also arrived at the cognitive stage of formal operations. The degree of horizontal decalage (or lag in abilities within a stage) that might exist between cognitive and personality development is for another study to determine.






13

Blatt and his associates have found the conceptual level of object representations to be related to the existence of both schizophrenia and depression. Blatt, Brenneis, Schimek, and Glick (1976) used a precursor of the system described in Blatt et al. (1981) to assess the object representations portrayed in the human figure responses on the Rorschach of normal (the Berkeley longitudinal sample) and schizophrenic adolescents. They found, to nobody's surprise, that when compared to the normal subjects, schizophrenic adolescents had impaired object representations, and that the greater the severity of the disorder, the greater the developmental impairment observed. A more intriguing and unexpected set of results was also noted. For moderately disturbed adolescents in the schizophrenic group, the level of object representations was greater than for normal subjects at the same age when the human figure response on the Rorschach was inaccurately perceived (i.e., the response bore little resemblance to the inkblot stimulus), but they scored at a lower conceptual level of object representations when the human figure was accurately perceived. More severely disturbed members of the schizophrenic group showed uniformly lower conceptual levels, regardless of the form quality of the response. Blatt et al. (1976) speculated that the nature of the schizophrenic person's interaction with the real, interpersonal world had a large influence on that person's






14


ability to access appropriate object relational features. In response to an interpersonally threatening world most schizophrenic individuals prefer to remain in what, for the moderately disturbed, seems to be a highly articulated and well-integrated fantasy world and to relate to the real world on the basis of a more primitive level of object representations. Lidz (1978) has remarked that for many people the onset of schizophrenic symptoms occurs when they enter into the stage of formal operations. The increase in cognitive egocentrism, which accompanies the entrance into every new cognitive stage, is especially great at this stage because formal operations permit one to create hypothetical possibilities, to imagine real outcomes without having to test them out. Lidz suggests that "the egocentric aspects of formal operations are overcome through increasing socialization" (Lidz, 1978, p. 85). Without this socialization (its absence might occur for various reasons) predisposed individuals can develop schizophrenic symptoms. Hence, these individuals may form higher conceptual levels of object representations (based on advanced cognitive ablilities) which can only be manifested in a distorted fashion. If a stable, trusting relationship--the main goal in treating someone who is schizophrenic--can be formed, the patient may be able to use this higher level of object representations in dealing with the real world. Of course, for "poor premorbid" individuals, who never reach the stage






15

of formal operations, one can expect a more primitive form of object representations to be used in relating to both the fantasy and the real world. For many, however, the full expression of any conceptual level of object representations appears to be integrally linked to the nature of the person's interpersonal environment. Incidentally, Blatt et al. (1976) were also able to use their system to trace the progressive development of object representations in normal individuals over time.

Blatt, Wein, Chevron, and Quinlan (1979) investigated the relationship between object representations (scored according to the procedure of Blatt et al., 1981) and the intensity and type of depressive experience in a normal sample of college students. Blatt (1974) has divided depressive episodes into two types: anaclitic and introjective. In anaclitic depression, the person who needs continual contact with other people in order to maintain gratification and a sense of self becomes depressed when a significant object loss is experienced. Introjective depression, on the other hand, is characterized by feelings of being unlovable rather than unloved, of having failed to live up to the high expectations and standards of the introjected (or partially assimilated) other. This depression results not from abandonment or neglect, but is a function of demanding, ambivalent, depreciatory, and hostile child-parent interactions. Blatt (1974) reasoned that






16

introjective depression implies a higher form of object representations than does anaclitic depression, because guilt involves the capacity to be self-reflective, to accept responsibility, and to have some sense of self. Blatt et al. (1979) used the Depressive Experience Questionnaire (DEQ) to measure the type of depression. The DEQ produced three dimensions (dependency, self-criticism, and efficacy) which permitted four classifications (in order of decreasing intensity: anaclitic=high dependency, mixed=high dependency and high self-criticism, introjective=high self-criticism, and nondepressed=high efficacy). Incidentally, Blatt, D'Afflitti, and Quinlan (1976) reported that the DEQ factors are significantly correlated with independent measures of depression (for instance, the self-criticism factor of the DEQ correlates significantly with only those 14 items on the Zung Depression Scale which measure the primary factor of "Loss of Self-Esteem", while the dependency factor correlates with the 5 Zung items that assess the more noncognitive, somatic-vegetative signs of depression, and the efficacy factor correlates with the 4 items which signify hopefulness and optimism). Blatt et al. (1979) found that the conceptual level of object representations, based on the subject's description of mother and father, was in a direct, inverse, relationship with type and severity of depression, such that anaclitically depressed subjects reflected the lowest levels and nondepressed subjects had






17

the highest conceptual levels. Thus, lower conceptual levels appear to be associated with an increased risk of psychopathology.

In addition to the conceptual level of object representations, Blatt et al. (1981) also score the subject's view of the person for twelve qualitative characteristics (e.g., the degree to which the person is seen as nurturant, intellectual, punitive, constructively involved, and so on), as well as the degree of ambivalence the subject has toward the person. These characteristics, when factor analyzed, load on two primary dimensions: "parent as nurturant" (which accounts for 40 percent of the variance) and "Parent as striving" (29 percent of the variance). These two factors bear close resemblance to the affiliation and power dimensions proposed by Leary (1957) as part of his interpersonal system of diagnosis.

Using Sullivan's conceptualization of personality as an interpersonal manifestation, Leary and his associates at the Kaiser Foundation in the 1950's constructed a system that located behavioral interactions between people on a two-dimensional "circumplex". Interpersonal behavior is organized on this circumplex in 8 adjectival categories arranged in a circular array around the orthogonal axes of Love-Hate (LOV) and Dominance-Submission (DOM). Despite Foa's (1961) comment that Leary' s circumplex was first formulated on a more or less intuitive basis (adjectives






18

were collected and arranged in a circular pattern) prior to developing ways of measuring this circumplex, research has tended to support the validity and universality of Leary's conceptualization. Lange (1970), in attempting to validate a measurement device--the Interpersonal Checklist (ICL)--used to assess placement on Leary's circumplex, obtained results which supported the notion that 2 bipolar dimensions (DOM and LOV) are arranged according to Leary's original thesis. Other attempts to develop ways of analyzing interpersonal behavior, such as Foa's facet analysis, Schutz's Firo-B, or Berzins' use of the Personality Research Form (PRF) in the Indiana Matching Study bear a close relationship to Leary's circumplex and can easily be reduced to it (Berzins, 1977). In short, it was thought that the incorporation of the ICL in the present study offered a way of obtaining concurrent validity for the qualitative factors of object representations produced by the system of Blatt et al. (1981). In addition, the ICL had the ability to provide additional descriptive data as well as providing variables which are sensitive to the socially desirable response sets that can be given to test protocols (LaForge, 1976).
Thus, the system devised by Blatt et al. (1981)

appeared to have adequately measured a person's degree of structural intrapsychic organization, and both their system and the ICL seemed able to capture the interpersonal






19


qualities of that organization. what was needed, however, was a procedure for analyzing the structural characteristics of the on-going social network within which the individual lives.

Social Networks

The work on the structural properties of social networks done during the past 25 years by such anthropologists as Barnes, Bott, and Boissevain appears to have supplied the materials for this procedure. More recently, their work has found its way into psychological research through two studies by Hirsch (1979a, 1979b). The ease with which certain facets of network analysis have been transferred across disciplines probably can be attributed to the level of analysis employed by these anthropologists. Namely, their analysis focuses on the individual and the exchanges which take place between individuals, as opposed to focusing on the attributes of the culture as a whole. Thus, the impact of the group on the individual can be studied without losing individual differences.

Many researchers feel that density is the most

important structural component of a network of individuals (Hirsch, 1979a), and density was the key network variable examined in the present study. A social network is defined by Mitchell (1969) as "a specific set of linkages among a defined set of persons, with the . property that the characteristics of the linkages as a whole may be used to






20

interpret [the] social behavior of the person involved" (p. 2). For the purposes of this study, the social network was composed of those people who the subject felt were, at the time of the study, a source of support, encouragement, or guidance. Density is commonly defined as the number of perceived relationships among members of the subject's social network over the total number of possible relationships among those members. Therefore, a lower density network is one where most of the members of an individual's network are viewed as having a relationship with one another. All measurements are based on the subject's perception of whether a relationship exists or not between the members.

Hirsch (1979b) examined college students' satisfaction with the social support they received from their social network as a function of density. Although higher density networks, or "extended families", were thought by many college students to be socially desirable (many students with lower density networks even apologized because their associates did not know one another), lower density networks clearly provided more satisfying support for those students who had them. Goldstein (1978), in studying the social networks of first year dental students, also found, contrary to his expectations, that higher levels of support were associated with lower network density. How can this counter-intuitive finding be explained? Hirsch (1979a)






21

suggested that "in order to minimize the possibility of offending individuals involved in intra-network conflict, high density NSS [natural support system) members tend to provide more ambiguous and superficial feedback, decreasing its supportive value" (p. 4). In more psychodynamic terms, such "feedback" would tend to promote the individual's egocentricity, giving rise to interpersonal distortions, at a time when the individual needs "consentual validation". As a result, the individual would become more deviant in a network (high density) which tolerates little deviation, thus increasing the "poorness of fit".

To further examine the relationship among higher and

lower density networks, Hirsch (1979a) studied two groups of women who were in the process of adapting to a major life change. One group was composed of older women (mean age=37) who were attempting to go back to college, while the other was made up of recently widowed women (mean age=46). Because these women had families, he especially focused on their nuclear family-friendship (NF-F) boundary density, that is the number of relationships reported to exist between the subject's nuclear family members and subject's friends, divided by the total number of possible relationships. The influence of the density rating is increased because shared variance is removed (nearly everyone reports that the members of their nuclear family have relationships with one another). Of note, the NF-F






22


boundary density was not related to the overall amount of life change encountered by these women. Hirsch also examined the multidimensionality (i.e., the diversity and complexity of interpersonal exchange) of the relationships between the subject and the people in their social networks. He found that subjects with higher density networks, as opposed to those with lower density ones, had fewer multidimensional friendships and smaller total networks, with fewer friends in their networks. This set of results implies that higher density, "extended families", are more homogeneous, offer less varied interactions, and are composed of fewer individuals. The "complex", lower density, social network seems to provide more diverse activities and interactions with a greater number of people. Subjects in higher density networks reported significantly lower satisfaction with supports, more psychological and physical symptoms, poorer overall mood, and lower self-esteem than those in less dense networks. These results are intriguing, considering that conventional wisdom would seem to suggest that, in times of stress, a closely knit unit of family and friends ought to promote mental well-being. Why, then, does a lower density network turn out to be more adaptive for this set of women? Hirsch (1979a) points out that these women, in order to cope successfully with this major change in their life (i.e., the loss of their spouse, or re-entering college), were required






23


to reorganize their familial role in such a way as to make the family a less important source of social support. Thus, they needed a support system that could permit intensified involvement outside the nuclear family. The "extended family" provided less opportunity for change because its lower density offered fewer outlets for alternative interpersonal behaviors. Not only were the women more "locked-in", but there was less temporary refuge from the pre-existing conflicts found in commonplace familial roles and relationships. The "complex" network, because of its greater differentiation of friends from family, allowed the woman the chance to see herself in a diversity of social contexts and roles and, thus, to find the interpersonal identity which best worked for her. The reorganization was less costly (in terms of threats to self-esteem which arose when she had to try new interpersonal behaviors) in the complex network because the diversity of structure meant that established commitments to certain activities or functions in a relationship could be changed without the need to establish entirely new relationships. In other words, a person with only unidimensional relationships would have to establish new relationships in order to engage in new functions or activities. In addition, someone with a history of multidimensional relationships also had a greater repertoire of experiences and probably more knowledge that






24


alternative relationships could work and be satisfying. In sum,

the greater diversity of interests and segregation
of different spheres of activity characterizing
low density, multidimensional NSS can be seen to serve as an insurance policy. This policy serves
to protect individuals from having problematic
changes in particular spheres of their lives
become too encompassing, threatening, and
debilitating. The policy provides rewarding alternative social identities and activities,
facilitating a smoother reorganization of one's
life, at less psychic cost. (Hirsch, 1979a, p. 34)

This understanding of denser or more complex social

networks facilitating successful change and coping has also been proposed by Merton (Coser, 1975) and more recently by Pearlin and Schooler (1978). In summarizing the literature on coping (which they defined as "that behavior which protects people from being psychologically harmed by problematic social experience", p. 2), Pearlin and Schooler (1978) concluded that a larger scope and variety of coping behaviors increased the likelihood those behaviors would be effective. They also observed that effective coping behaviors are unequally distributed in American society, so that the affluent, the educated, and the men in this society tend to make greater use of successful coping behaviors. Coser (1975) applied some of Merton's ideas of "role-sets" to the process of individuation. The distinction was made between simple, or restricted, role-sets and complex role-sets. In a simple role-set the members were usually equal in status and were role partners. Most interactions






25

were made with the same people. Examples of simple role-sets could be found in family or kinship groups, or other higher density networks. Status in a complex role-set was unequally distributed throughout the system and the membership of the system was subject to change. Examples of this system could be found in most white-collar job situations. The complex role-set allowed for increased relativity and perspective taking, and the person within it was exposed to more contradictory expectations, more differential interests, more observability by others, and more opportunity to be both an authority figure to some and a subordinate to others. Coser made the argument that "individuality" was enhanced by complex role-sets and differential social structures.

These writers who extolled the virtues of lower density social networks may have been biased in their conclusions because of the populations they worked with. It seemed highly plausible that a lower density network could be effective for middle-class, well-educated, adults living in a post-industrial society. But for others, who are less well-developed in a material, maturational, and psychological sense this may not be true. In traditional farming communities, a higher density social network can be the difference between life and death for many subsistence farmers. Children need an intact, highly integrated family unit on which to base the foundations of development.







26

Sokolovsky, Cohen, Berger, and Geiger (1978) suggested that former schizophrenic patients living in single-room occupancy (SRO) hotels had fewer rehospitalizations if they were integrated into higher density social networks. Hirsch (1979a) also reported that higher density "pressure groups" can be therapeutically successful in bringing about change in certain individuals. He suggested that, "apparently, therefore, the adaptiveness of dense NSS may vary considerably depending on specific person-NSS characteristics" (Hirsch, 1979a, p. 4). Thus, the need existed to understand the interaction between the density of social networks and the level of object representations.

Central Thesis: An Interactionist Model

The central thesis of the present study was that an

inverse relationship between the total density of a person' s social network and his or her conceptual level of object representations would result in a goodness of fit. In other words, the individual at a lower conceptual level would require a higher density network to be sustained at an optimal level of functioning. Similarly, a person with a higher conceptual level would function best in a lower density network. Deviation from the equilibrium of this expected relationship would, therefore, result in greater subjective distress (e.g., poorer mood, more complaints) and impairment in the areas of love and work. The emphasis of this model was not on predicting specific symptoms or






27

categories of psychopathology. Rather, the attempt here was to describe some of the factors that are associated with a more general sense of adaptiveness. The importance for clinicians, however, was that changes in a person's adaptiveness would probably influence his or her help-seeking behavior. Jacobs and his associates (in a series of studies, reported in Minter and Kimball, 1980) have found that a combination of social and personality characteristics (in this case, unresolved role crises, social isolation, life stress, and an angry-defiant copy style) was influential in increasing the likelihood that a person would seek medical attention for the same symptoms (upper respiratory illness and asthma) that another person might ignore. Hence, the people seen by clinicians are probably those who have suffered a negative change in their state of adaptiveness.

The competing hypothesis was that lower density

networks create an optimal, healthy environment, regardless of the level of a person's object representations. Obviously, lower levels of object representations would have contributed to some decrement in functioning, while higher levels would have given rise to better adaptation. But, overall, this alternative hypothesis held that higher density networks would be uniformly restrictive. This hypothesis, however, seemed to have denied any orthogenetic notions of social networks, as well as any qualitative






28


changes that might have taken place in the development of object representations.

Werner's orthogenetic principle can be applied to

social networks, in that a network also "proceeds from a state of relative globality and lack of differentiation to a state of increasing differentiation, articulation, and hierarchic integration" (Werner, 1972, p. 47). Such is the rhythm of living systems. A child begins with a network density close to 1.00, since almost all the significant others in the child's life probably have a relationship with one another (even in cases of divorce, the child would probably perceive his or her mother and father as having a relationship with one another). Higher network density is adaptive for this level of personality development. In later phases of development, as the child enters school or forms a "chumship" with a same-sex peer (Sullivan's pre-adolescent period), network density decreases to about .50 or .40. These figures are still considered "higher" network density in comparison with the norm for later developmental periods. "Adult" density scores hover between .33 and .25 (personal communication, Hirsch, 1982). Thus, optimal density corresponds to the period of development.

So, it seemed reasonable to propose that an individual whose personality development had been arrested at an early stage would have needed a social network with a reciprocal level of density in order to have displayed optimal






29

functioning. A lower density network for a person with a lower conceptual level of object representations would probably have increased the likelihood that he or she would have experienced overwhelming levels of ambiguity and complexity (i.e., receiving a confusing variety of complex responses from people) as well as having an intensified feeling of social alienation (i.e., the feeling that people in the network are not connected in any meaningful way). Internally, the person's fragile sense of self would have been left less cohesive and more fragmented, which would have resulted in a decrement in this person's adaptation to his or her environment. Likewise, following the principle of relative deprivation, a functionally equivalent decrement would have been expected for someone with a higher conceptual level who lived in a higher density social network. Such a combination would most likely have led to regressive behavior in that person.

College students were chosen as the group upon which the interactionist model was to be tested. Besides the pragmatic aspects of availability offered by using this population, the chief advantage to choosing college students was that they were seen as facing a relatively homogeneous set of psychosocial tasks. Also, the viability of this interactionist model should have been particularly highlighted when tested with people who were engaged in tasks that involved making significant interpersonal adaptations.






30

To this end, college students should be seen as being adolescents (White, 1980). Their entry into the employment market and the more formal responsibilities of adulthood have been postponed. Instead, they enter a prolonged transitional period in which the traditional tasks of adolescence, which are emancipation from parents, sexual adjustment, and vocational choice (Blos, 1941), can be further elaborated and worked through. Hence, it is not surprising that people who attend college become more autonomous, flexible, complex, and tolerant of ambiguity, less materialistic and dogmatic, and better able to be aware of and to express their emotions than their peers who did not attend college (Plant, 1965; Chickering and McCormick, 1973). At the same time, most students complete Erikson's (1950) superordinate psychosocial task of adolescence by having achieved a well-established identity by the time they finish college (Waterman and Goldman, 1976). College, therefore, seems to contribute a profound set of experiences which promote the personality development of those who attend. To extend the rite of passage into adulthood by four years or more, however, creates the room for these individuals to be involved in considerable stress and strain. Therefore, this group of adolescents seemed ideally suited for the requirements of the current study.






31

Hypotheses

Six hypotheses, which include the central hypothesis and five auxiliary hypotheses, have been summarized below. The central thesis of the current study, as described in the previous section, was encapsulated in the the first hypothesis.

The second hypothesis had two basic parts. In the

first part, four variables were predicted to be correlated with the outcome measures. The four variables included the two main variables (i.e averaged conceptual level and total network density), plus two other variables (life change and total network size) which had been shown in the literature to possibly be correlated with the outcome variables. (Note that the literature review for the variable of life change appears in the next chapter under the heading "Measures".) In the second part of this hypothesis, predictions were made about the non-influence of certain demographic and control variables. These predictions were made to expose any moderator variables that may have had an influence on the central hypothesis.

The third hypothesis was constructed to show that the outcome measures behaved predictably within this study. Smith and Glass (1977), in their meta-analysis of psychotherapy outcome studies, reported that certain outcome variables were more likely than others to be correlated with the independent variables. Hence, predictions were made






32

about the reactivity of the four outcome measures in the current study.

In the fourth hypothesis, the possible effects of life change on the main variables were considered. Although the prediction was made in the second hypothesis that the two main variables (i.e averaged conceptual level and total network density) would be independent of life change, the effects of life change, or situation stress, are often thought of as being so powerful as to be a main variable in other studies (see the section on life change under the heading "Measures" in the second chapter).

The two final hypotheses concerned the qualitative

factors derived from the written descriptions of parents and the qualitative dimensions (DOM and LOV) from the ICL. The predictions made in these two hypotheses were not directly related to the central hypothesis. Instead, these predictions served as a way of assessing the internal consistency of the responses given by the subjects and, thereby, could lend more credence to the other results within the current study. The first hypothesis was an attempt to provide concurrent validity for the qualitative factors of the written descriptions of parents by comparing these factors to the corresponding ICL dimensions. In the second hypothesis, the subjects' ICL scores were placed in one of four quadrants of Leary's (1957) circumplex on the basis of the of their DOM and LOV coordinates. The






33

prediction was made that certain quadrants would be correlated more favorably with the outcome measures. Hypothesis 1. Interactional Effects

As the central hypothesis, the total network density

and the averaged conceptual level of object representations would be linked in an inverse relationship, whereby subjects with higher conceptual levels would have lower density networks and subjects with lower conceptual levels would have social networks with higher density. Thus, subjects who had greater deviations from the equilibrium of this inverse relationship would have higher levels of personal distress (i.e., poorer mood and more symptoms) and greater impairment in love and work.

So that this interaction could be observed further, the level of object representations was split into low, medium, and high levels. Thus, for the lower levels of object representations (scaled score=4.0 or lower), higher density would be significantly correlated with better scores on the measures of mental health. In the medium range (4.5-5.5) an average level of density would be related to better adjustment. And, for the higher levels of object representations (6.0 or above), a lower density network would be most adaptive.

Hypothesis 2. Univariate Comparisons

Subjects with higher levels of life change, smaller

total network size, higher total network density, and lower






34


averaged conceptual levels would have significantly higher levels of personal distress and significantly greater impairment in their ability to love and work.

Ideally, the variables of averaged conceptual level, total network size, total network density, and life change would be independent of all demographic variables, such as age, sex, socio-economic status (SES), and so on. Likewise, it was hoped that verbal fluency would be unrelated to the conceptual level of object representations, and that the variables on the ICL which are sensitive to response sets (i.e number of items checked, or NIC, and average intensity, or AIN) would not be correlated with the measures of mental health.

Hypothesis 3. Outcome Measures

As with most correlations to measures of mental health, the present study would demonstrate the total variance (R-squared) explained by the major independent variables (i.e averaged conceptual level and total network density), in association with the variable of life change, to be greater for the multiple linear regression models which use the measures of subjective distress (a mood scale and a symptom checklist) as their dependent variable than for the model which uses a role performance variable (grade point average). The total variance (R-squared) of the model using the relationship measure as the dependent variable would fall within the range of the total variance (R-squared) for the other models. All models would be significant.







35

Hypothesis 4. Life Change

The correlation between total network density and the measures of mental health and the correlation between the averaged conceptual level of object representations and mental health would still be significant after the effect of life change had been partialed out. Hypothesis 5. Qualitative Factors and ICL Dimensions

As an expression of the concurrent validity between the qualitative dimensions of the object representations and the interpersonal factors of the ICL, the striving factor would be similar to the DOM dimension of the ICL for the same sex parent. Likewise, the nurturant factor would correspond to that of LOV on the ICL.

Hypothesis 6. ICL Quadrants

After the ICL dimensions of LOV and DOM were reduced to their component parts, Love (L)-Hate (H) and Dominance

(D)-Submission (S), and four quadrants (DL, DH, SL, and SN) were formed, the quadrants from the ICL for father would be ranked so that, in ascending order, significantly better scores on the mental health measures would be found with SH (worst), SL, DH, and DL (best). In a similar fashion, the quadrants of the ICL for mother would be ranked to that, in ascending order, significantly better scores on the mental health measures would be found within SH (worst), DH, SL, and DL (best).















CHAPTER TWO
METHOD

Subjects

Of the 117 undergraduates at the University of Florida who participated in the present study, 65 were female and 52 were male. They were all single and 18 to 23 years old (average age=19.7 years). They stated that they had attended this university during the Spring or Summer semester of 1983 and that they were not in psychotherapy while participating in the study.

Measures

Category A. Measures of Life Change and Demographics

The variables assessed in Category A (life change and

demographics) could obfuscate the hypothesized relationships between the variables in Category B (level and quality of object representations, density and size of social networks) and those in Category C (degree of subjective distress and success in love and work); hence the need to control for the effects of these "given", life event, variables.

Life Change-College Schedule of Recent Events (CSRE)

Andrews and Tennant (1978), in reviewing the literature, concluded that increases in life change, or situation stress, seem to be related to a higher incidence of


36 -






37


psychological disorders and physical illness. The effects, however, are usually modest, which is consistent with Andrews and Tennant's (1978) contention that life change cannot be said to be the cause for nor does it appear to be a major determinant of physical or psychological problems. Life change is best viewed in terms of its contribution to the psychological and physical states that foster the development of ill-being. Life change probably demoralizes a person and increases a person's need for the benefits of social interaction (be it support or professional help). Of particular relevance to the present study, Smith (1977) and Molen (1978) have found that increased levels of life change, situational stress, in college freshmen, junior college students, and graduate nursing students have been associated with decreases in functioning on a role performance variable, namely grade point average (GPA). This positive correlation between life change and GPA is weak but significant in all the studies.

Hirsch (1979a) reported that a person's network density was independent of his or her experience of life change. The relationship between the conceptual level of object representations and life change is unknown, though it was hypothesized that they are independent. So, while it was thought that the major determinants of maladaptiveness were probably the intrinsic personality and sociodemographic attributes of the person, the level of life change needed to be controlled in the present study.






38

All of the above studies used the Holmes and Rahe (1967) Schedule of Recent Events (SRE) or a reasonable facimile thereof. Each event on this schedule was reduced to a score (expressed in "life change units"--LCJ), which was thought to represent that amount of disruption in a person's life that resulted from the particular event. Some of the 43 events on the SRE, however, were not appropriate for most college students (e.g., "son or daughter leaving home"). Anderson (1972) has developed the College Schedule of Recent Events (CSRE) in response to this deficiency of the SRE. The selection of these events and the LCU assigned to each event was obtained by consensus from the opinions of 284 students, using a procedure which was identical to that employed by Holmes and Rahe (1967). The results show high agreement among all groups within the sample, with the highest agreement being between college juniors and seniors (.97) and the lowest between freshmen and graduate students (.87). For males and females the agreement was .94. Two studies (Marx, Garrity, and Bowers, 1975; Liao, 1977) have used the CSRE to find the expected small but significant relationship between higher levels of LCU and increased incidence of health problems for college freshmen and second year pharmacy students.

For purposes of this study the CSRE was modified

slightly; all items (a total of seven) which pertained to being married were deleted because no subject in the present






39


study was married. Also, the subjects were asked to check the number of times during the last nine, months that any of the events occurred to them. This time interval was comparable with the interval Hirsch (1979a) had instructed his subjects to use on the SRE.

Demographic Variables

Age. Only subjects between the ages of 18 and 23,

inclusive, were used in the study because students within this range should have been concerned with the same psychosocial tasks, as opposed to older or younger students.

Class. Subjects were asked if they were classified as freshmen, sophomores, juniors, or seniors.

Sex. Note was made of the subject's sex. Edwards and Thacker (1979) reported that the females in their study obtained a higher GPA than the males.

Socioeconomic Status (SES). Hollingshead's Two Factor Index of Social Position--The ubiquitous effects of social class differences are well known in the social sciences. SES was measured in the traditional method, using Hollingshead's Two Factor Index of Social Position (Hollingshead, 1957). This instrument combines the educational and occupational status of the "head of the household" of the subject's family as a basis for placing the subject in one of five classes of SES. People in Class I were considered "upper class", those in Class II were "upper-middle class", Class III represented "middle class",






40


Class IV indicated "lower-middle class", and the people in Class V formed the "lower class".

Employed. Of interest here was whether or not the subject was, at the time of the study, employed to help finance his or her education. This financial necessity could have produced an additional source of stress above and beyond that tapped by the CSRE.

Credits. The amount of credits the subjects were

enrolled for during the semester of the study could have varied widely. Those subjects taking an unusually heavy load could have suffered as a result.

Domicile. Of concern here was whether or not the subject lived with his or her parents while attending college. It was thought that living at home could have restricted the subjects' identity formation by inhibiting their separation from their parents, which may have resulted in a general decrease in psychological well-being. Category B. Primary Measures

Object Representations

Procedure. The subjects were given a blank piece of

paper and asked to "describe your mother", with five minutes allotted for the description. They were given another five minutes to "describe your father". These descriptions were then scored for the conceptual level and the qualitative characteristics of the object representations. Verbal fluency was also scored.






41


Conceptual Level. Each description was given a scale score of 1-9 to denote the five conceptual levels. These levels are sensorimotor-preoperational (scaled score=l), concrete-perceptual (3), external iconic (5), internal iconic (7), and "conceptual" (9), and a description of these levels is provided on pages 10 through 12. The interrater reliability given by Blatt et al. (1981) for scoring the level of object representations ranged from .88 (between expert rater and trained rater) to .70 (between expert rater and untrained rater) to .85 (for all three raters).
Qualitative Characteristics. A written description of parent was rated on each of 13 characteristics according to the degree to which the subject was seen as attributing that characteristic to his or her parent. Each characteristic was rated on a scale from 1-7, with a score of 4 given when the characteristic was not relevant or was too vaguely described to be scored accurately. The 13 characteristics were as follows.
(1.) Affectionate--little affection=l, much affection=7

(2.) Ambitious-Driving--relatively non-ambitious and

driving=l, strongly ambitious and driving of self and

others=7
(3.) Malevolent-Benevolent--malevolent=l, benevolent=7

(4.) Cold-Warm--cold=l, warm=7
(5.) Degree of Constructive Involvement-- distinctive or

destructive, intrusive involvement=l, positive and






42


constructive involvement with encouragement of

autonomy and individuality=7

(6.) Intellectual--not at all=l, highly intellectual=7

(7.) Judgemental--non-judgemental=1, highly judgemental=7

(8.) Negative-Positive Ideal--negative ideal=1, positive

ideal=7
(9.) Nurturant--low nurturance=1, high nurturance=7 (10.) Punitive--non-punitive=1, highly punitive=7 (11.) Successful--failure=1, success=7 (12.) Weak-Strong--extremely weak=1, extremely strong=7 (13.) Degree of Ambivalence--little ambivalence=1, marked

ambivalence=3 (note this characteristic is only scored

1-3)
A more thorough description of each characteristic is found in Blatt et al. (1981). The interrater reliability (alpha coefficient) for all three raters ranged from .68 for "affectionate" to .93 for "successful".
Blatt et al. (1981) found that these 13 characteristics loaded on two factors. The first factor, parent as nurturant, was composed of nurturance (factor loading=.90), positive ideal (.90), benevolence (.88), warmth (.87), constructive involvement (.84), affectionate (.80), strength (.67), and successful (.65). The second factor, parent as striving, was made up of judgemental (.90), ambitious (.89), punitive (.88), intellectual (.82), ambivalence (.60), successful (.48), and strength (.48). The scale scores for






43


parent as nurturant and parent as striving were formed by summating the scale scores of the characteristics which loaded on that factor. Overall, the nurturant factor and striving factor had a reliability of .95 and .93, respectively, for all three raters.

Verbal Fluency. This variable was arrived at by coding the length of the typed description (e.g., 1-4 lines=l, 5-7 lines=2, 8-10 lines=3, . more than 19 lines=7). When factor analyzed along with the 13 other characteristics, verbal fluency formed a third, single-variable, factor. It accounted for 8 percent of the total variance (Blatt et al., 1981).

Interpersonal Checklist (ICL)

Description. The ICL has been found to be an efficient means for measuring salient interpersonal dimensions, and it has significantly outperformed several S-R scales in predicting to criterion variables which measure interpersonal behavior (Knudson and Golding, 1974). On the ICL, a 134-item adjective checklist, the subject was asked to circle those items which described the stimulus person (in this case, mother or father). One-hundred and twenty-eight of the items can either produce 16 interpersonal scores or be loaded on the 2 dimensions of Love-Hate (LOV) and Dominance-Submission (DOM). The present study used Form 4 of the ICL. The data were reduced to the LOV and DOM scores so each subject could be sorted into one of four quadrants.






44


Average Intensity (AIN) and Number of Items Checked

(NIC). These variables were used to check for response sets. Each adjective was ranked (1-4) according to its "intensity" of affective charge. For instance, "can be obedient" had an intensity of 1, while "spineless" had an intensity of 4 within the same category. The AIN consisted of the weighed number of adjectives over the total number checked. According to LaForge (1976), an AIN well above 2.00 represented a willingness to criticize the stimulus person; whereas, an intensity well below 2.00 suggested a tendency to idealize the person. AIN correlated -.74 with social desirability (LaForge, 1976). NIC was simply a total of the number of items endorsed by the subject. A very low NIC usually indicated that the subject had only a superficial engagement in the task or was describing someone who was not well known to the subject.

Social Networks

Social Network List. This instrument was used to

obtain the size of the social network. The subjects were asked to list up to 20 significant others in their life with whom they had had contact with (in person, by phone or letter) at least once in the prior 4-6 week period. These significant others were to be listed in the three categories of family members or relatives, friends, and professionals. The instructions requested that the subjects list only those people who were important sources of support, encouragement,






45

or guidance. In Hirsch (1979a) the average size of the network was 14 people (family and relatives=6, friends=7, and professional=l).

Support System Map. This instrument was used to

compute the density scores. The subjects were presented with a blank piece of paper, except for the word "map" at the top. They were instructed to put their name in the middle. They were then to put those individuals they mentioned on the Social Network List onto the map, putting nearest to their own name those individuals whom they felt "closest" to. For convenience sake, they were to put friends on one side of the page, family members and relatives on the other side, and professionals on the bottom. The subjects were then asked to draw lines between themselves and each person on the map. Next, they were asked to draw a line between those individuals whom they considered to have relationships with each other. The formula for computing the density of the family and relatives network, the friendship network, and the total network was as follows.

density=X/(N(N-l )/2)
where X=the number of actual relationships between members of the
subject's network
N=number of people in the subject's network.






46


The formula for nuclear family-friendship (NF-F) boundary density was as follows.

NF-F boundary density=X/(NF)(F)

where NF=number of nuclear family members F=number of friends

X=number of relationships existing between the nuclear family
members and the friends.

An example of how these two formulas can be used for computing the density scores of a hypothetical subject is presented in Appendix A.

Category C. Measures of Mental Health

Subjective Distress
Profile of Mood States (POMS). The construct validity of this 65-item instrument has been demonstrated in numerous studies with both clinical and normal populations and across levels of SES (McNair, Lorr, and Droppleman, 1971). Six factors have been derived from the POMS: tension-anxiety, depression-dejection, anger-hostility, vigor-activity, fatigue-inertia, and confusion-bewilderment. A total mood score was computed by subtracting scores on the vigor-activity scale from the combined score on the other five scales (equal weighting is given to all scales). Subjects were asked to fill out the POMS according to "how you usually feel".
Hopkins Symptom Checklist (HSCL). This instrument was a 58-item scale which had been shown in many studies to have






47

construct validity for both clinical and normal populations and across levels of SES (Derogatis, Lipman, Rickels, Uhlenhuth, and Covi, 1974). Five factors have been extracted from the HSCL: somatization, obsessive-compulsive rumination, interpersonal sensitivity, depression, and anxiety. A total symptom score was computed by combining the scores on all five scales (each receiving equal weighting).

Love (interpersonal exchange of positive affect) Scale of Feelings and the Behavior of Love (Love

Scale). This instrument was a 120-item scale devised by Swensen (1978) which can be used to rate the quality of any relationship (e.g., with spouse, mother, father, or closest friend of same or opposite sex). Six factors were produced by the Love Scale: verbal expression of affection, self-disclosure, toleration of the loved one, non-material evidence, unexpressed feelings, and material evidence. A total Love Scale score was computed by adding up the scores on all six factors (equal weightings). Subjects in the current experiment were asked to fill out the Love Scale on the person whom they felt closest to.

The Love Scale was thought to have been especially well suited for the current study. Grumper (1976) has found that juvenile offenders at the post-conformist stages of ego development have a higher score on the Love Scale for their best friend than do juveniles at the pre-conformist levels






48


of ego development. Kohlhepp (1976) also found that whereas the quality of the relationship as measured by the Love Scale tends to decrease over the course of marriage for spouses at the pre-conformist levels, the amount of "love" expressed by spouses at the post-conformist levels increases over the course of their marriage. Because ego development was thought to be independent of psychopathology (Waugh and McCaulley, 1981) and level of object representations was related to psychopathology, the level of object representations was proposed to have shown a similar, if not stronger, relationship to the Love Scale than that reported for ego development. Nickerson (1977) found that increased expression of love was a function of Bales personality factors of assertiveness (DOM) and positiveness (LOV), with more assertive and positive spouses tending to express more love to their spouses. Hence, subjects whose parents are in the high DOM-high LOV quadrant were hypothesized to have a tendency to score higher on the Love Scale. In addition, Swensen (1978) reported that the expression of love (as measured by the Love Scale) and marriage problems (as measured by the Marriage Problem Scale) are two separate orthogonal factors for normal couples. The POMS usually correlated highly with lists of (marriage) problems. Hence, the dimensions tapped by the POMS (and perhaps the HSCL) and the Love Scale were thought to be different.






49


Work (role performance variable)

Grade Point Average (GPA). The most parsimonious means of measuring the subjects' role performance as college students was chosen to be their GPA for the semester.

Procedure

The subjects were solicited for the study from the Introductory Psychology courses offered during the Fall semester of 1982. They received research credit for their participation, which was applied toward their final grade. A copy of the consent forms are found in Appendix A. Because no interview data was required, the subjects were run in groups with up to five members in each. First, the demographic data were collected. A copy of the demographic questionnaire constructed for this study can be found in Appendix B. Next, the subjects filled out, in sequence, the College Schedule of Recent Events, description of mother, description of father, ICL for mother, ICL for father, Social Network List, Social System Map, POMS, HSCL, and Love Scale. The total time for administering these instruments, with breaks, was approximately 2 hours. The GPA was obtained from the Registrar at the end of the semester. Results were mailed to those who requested them. The data were collected by two research assistants, who were trained by the experimenter. The set of instructions used by the assistants while conducting the data collection protocol can be found in Appendix C. A record of the questions they were






50


asked by the subjects and the answers the assistants gave to them are also found in Appendix C.

The descriptions of mother and father were scored blindly by the experimenter. Fifty descriptions were randomly selected and scored by an independent rater to confirm the reliability of the scoring procedure. The 27 examples given by Blatt et al. (1981) were used as training exercises. Special effort was paid to scoring the conceptual level because of its importance in this study. After the 50 descriptions were scored in the reliability check, the experimenter and the independent rater reviewed those descriptions on which they disagreed. An attempt was made to achieve consensus on these differences. In addition, the experimenter and the independent rater also reviewed the conceptual level of the other dozen cases on which the experimenter had some question.

A FORTRAN program was provided by LaForge (1976) for

reducing the raw data collected by the ICL. The subsequent data analysis was handled by the the computerized statistical analysis system (SAS, 1982) on an IBM 3081 at the Center for Information Technologies of Stanford University.















CHAPTER THREE
RESULTS

In the first section, the reliability correlations for the written descriptions of parents are presented, followed by a comparison of the mean scores of the measures used in the current study with scores reported in previous studies. In the last six sections, the results pertaining to the six hypotheses are set forth.

Interrater Reliability And Normative Comparisons

The interrater reliability correlations for the 13

qualitative descriptors, the length of description, and the conceptual level of the object representations in the reliability sample (n=50) compared favorably with the reliability intercorrelations reported in Blatt et al. (1981). When the 13 qualitative descriptors were reduced to the 2 main qualitative factors of "parent as nurturant" and "parent as striving", the Pearson product-moment correlations between the ratings of the experimenter and those of the independent rater were .82 and .71, respectively, for these 2 factors. The intercorrelations for the length of descriptor was .98. On the crucial ratings of conceptual level, the scores of the experimenter and the independent rater correlated .88 with



51 -






52


each other, which was identical with the intercorrelation that Blatt et al. (1981) found between the scores of their "expert rater" and "trained rater". These results are summarized in Appendix D.

In the present study, the scores derived from the descriptions of parents, the social network scales, the CSRE, the POMS, the HSCL, the Love Scale, and the ICL were fairly similar to the scores expected from the normative data provided by other authors. Very few significant differences were found when these measures were compared with the same measures in the previous studies.

Despite the exclusion of the items pertaining to

marriage, the average CSRE score was still significantly higher in the current study than in Marx et al. (1975). The distribution of scores in the two studies, however, was almost identical, with 15 percent of the subjects being in higher change group (i.e., they had scores which were one standard deviation above the mean), as compared to 14 percent in Marx et al. (1975) and 10 percent in the lower change group, as compared to 11 percent in the earlier study.

Although the mean total symptom score on the HSCL was higher for the present group of subjects (mean=68) than for those in the normative sample (m=51) reported by Derogatis et al. (1974), the HSCL scores in the present study were closer to those found in Hirsch's (1979a) sample of recent






53

widows and women returning to college (m=72). Both samples were significantly less than the mean score for the groups of anxious neurotics (m=90) or depressed neurotics (m=103) reported by Derogatis et al. (1974).

On the surface, the Love Scale scores, especially for females, appeared to be quite different in the present sample than in the norms provided by Swensen (1978). The main difference seemed to have been that subjects in the present study had far fewer "unexpressed feelings" (which counted negatively toward the total score) than in the earlier study. Perhaps, this difference was a reflection of the relationships which were being rated by the subjects. In the norms found in Swensen (1978), the subjects were asked to rate their relationship with their closest opposite sex friend, whereas in the present study they were instructed to rate their relationship with the person to whom they felt the closest. That females had higher scores than males in the current study was not surprising considering that 43 percent of the females chose to rate their relationships with their fiance or boyfriend whereas only 23 percent of the males had made this choice of relationships.

Many of the scores derived from the ICL were

significantly different in the present study from the norms offered by LaForge (1976). In particular, the number of items checked (NIC) were 8-11 items fewer than expected.






54


The average intensity (AIN) of the adjectives checked was, however, almost identical with the normative data. Although the scores for the present study were sometimes different on the LOV and DOM for mother and for father, the general direction of the LOV and DOM scores and their relationships with one another in the normative sample were maintained in the current study.

The the scores derived from the descriptions of parents were not significantly different in the present study than from what had been given as norms in Blatt et al. (1981).

On the surface, the means for social networks appeared to be very similar in the present study to those in Hirsch (1979a). The size of the social network was 13.7 in the present study, whereas, it was 13.9 in Hirsch (1979a). The total network density was .26 in the current study, compared to .27 in the earlier study. But, the density of the nuclear family/relatives network was higher in the present study (.85) than in Hirsch's (1979a) sample (.59). And, in a similar fashion, the network of friends had slightly higher density (.29) in Hirsch (1979a) than in the present study (.23).

The normative comparisons mentioned above can be found in Appendix D.

Next, the results pertaining to each of the hypotheses are reported. Each of the following six sections corresponds to one of these hypotheses.






55

Interactional Effects

It was proposed, as the central hypothesis of the

current study, that greater deviations from the equilibrium of an inverse relationship between total network density and the averaged conceptual level would result in poorer scores on the outcome measures. Greater deviations from this equilibrium, however, were not significantly correlated with any of the four outcome measures. To obtain these results, the total network density and the averaged conceptual level scores were standardized in the form of z-scores. Then for each subject, the z-score for the conceptual level was subtracted from the z-score for network density. The absolute value of this difference created an inverse relationship between the scores, such that a final score of zero would indicate a perfect fit between network density and conceptual level (e.g., higher network density with lower conceptual level) while a larger final score would represent a more divergent interaction (e.g., lower network density with lower conceptual level). This final score correlated -.07 with the POMS, -.13 with the HSCL, .03 with the Love Scale, and -.07 with GPA. Only the correlation with GPA was in the expected direction. The final score was not significantly correlated with any of the subscales on the POMS, the HSCL, or the Love Scale.

To further elucidate this pattern of results, the

conceptual level scores were divided into three groups. The






56

subjects with an averaged conceptual level of 4.0 or less
were placed in the low group (n=27), while those subjects

with a conceptual level of 6.0 or more were placed in the

high group (n=27). The remaining subjects, whose averaged

scores ranged from 4.5 to 5.5, were placed in the medium


Table 1
Means (M) and Standard Deviations (SD) of Selected
Variables for Three Groups of Averaged Conceptual
Level (CL) of Object Representations


lower medium higher
CL CL CL
(n=27) (n=63) (n=27)
Variables M(SD) M(SD) M(SD)

Network Density 0.28(0.18) 0.27(0.13) 0.26(0.12)
Network Size 13.85(3.81) 13.62(4.24) 13.63(3.99)
Nurturant--Mother 0.65(5.70) -0.55(6.09) 0.64(5.49) Nurturant--Father -0.55(6.27) 0.27(5.75) -0.08(5.59) Striving--Mother -0.03(2.26) -0.13(3.35) 0.34(2.77) Striving--Father* -1.52(3.03) 0.12(2.73) 1.24(3.11) DOM--Mother* 6.95(5.26) 3.93(6.43) 1.57(7.14)
DOM--Father 8.97(5.43) 10.32(5.13) 9.81(5.29)
LOV--Mother 3.83(7.52) 3.96(7.78) 6.22(8.15)
LOV--Father 2.77(9.93) -0.28(7.50) -0.99(7.49)
GPA 2.77(0.82) 2.61(0.68) 2.89(0.84)
POMS 43(19) 42(25) 43(20)
HSCL 71(18) 66(15) 68(12)
Love Scale 191(25) 189(30) 193(26)
CSRE 1248(574) 1018(569) 1053(611)
*=difference in means between low CL and high CL
are significant at p<.05
Note. Network Size and Network Density are both
Total scores.


group (n=63). In Table 1, the means and standard deviations

within each of the three groups are summarized with respect






57


to certain selected variables, variables which were chosen for their overall importance in the study. As can be seen in Table 1, the most striking result is the "flatness" of the distribution of network density scores across the three groups; the mean for network density varies from .28 to .27 to .26 across the three groups. Only the means for the qualitative factor of striving factor from the description of father and the DOM dimension on mother are significantly different across the three groups. Subjects tended to represent their father as more striving and to describe their mothers as less dominant as the conceptual level of the subjects' group increased. Thus, Table I illustrates that the differences between the three groups of conceptual level scores existed only for two of the qualitative dimensions and not for network density or for any of the outcome measures.

The scores for total network density were also divided into three groups. Subjects with total network density scores of .16 or less were placed in the lower density group (n=20), while subjects with total network density scores of .33 or greater were assigned to the higher density group (n=30). The remainder (n=58) fell into the medium density group. The cutting scores for the higher and lower groups were chosen so that approximately 30 subjects would be placed in each of these two groups. Though no direct predictions had been made about these three groups, it was






58

assumed that subjects in the higher density group would have

lower averaged conceptual level scores while those in the

lower density group would have higher averaged conceptual


Table 2
Means (M) and Standard Deviations (SD) of Selected Variables for Three Groups of Total Network Density


lower medium higher
total total total
(n=29) (n=58) (n=30)
Variables M(SD) M(SD) M(SD)

Conceptual Level 4.97(1.14) 4.96(1.11) 4.90(1.17) Network Size* 16.83(3.84) 12.72(3.72) 12.47(3.35)
Nurturant--Mother 0.70(5.78) -0.45(6.17) 0.20(5.36) Nurturant--Father -1.18(6.90) 0.33(5.39) 0.49(5.44) Striving--Mother 0.01(2.86) -0.01(3.20) 0.00(2.75) Striving--Father 0.01(2.98) -0.03(3.25) 0.05(2.65) DOM--Mother 4.95(6.55) 2.97(6.65) 5.38(6.26)
DOM--Father* 11.62(4.07) 9.49(5.36) 9.00(5.70)
LOV--Mother 5.18(8.39) 3.14(7.21) 6.28(8.11)
LOV--Father 0.72(8.84) -0.86(7.49) 3.36(8.23)
GPA 2.65(0.77) 2.75(0.74) 2.68(0.79)
POMS 48(29) 42(19) 39(22)
HSCL 71(16) 68(16) 65(13)
Love Scale 193(24) 188(30) 193(27)
CSRE 1196(538) 1042(635) 1038(516)
*=difference in means between low CL and high CL
are significant at p<.05
Note. Network Size is a Total score and Conceptual
Level-isan Averaged score.


level scores. As can be seen in Table 2, the means for the

conceptual level scores are nearly flat across all three
groups, just as the means for the total network density

scores had been evenly distributed across the three groups

of conceptual level scores in Table 1. In Table 2, total






59


network size is significantly larger (p<.0005) for the lower density group than for the other two groups. In addition, the DOM of the subjects' father is significantly higher (p<.05) in the lower density group than in the higher density group. Thus, Table I and Table 2 illustrate that averaged conceptual level and total network density seem to be unrelated to each other. This lack of relationship between total network density and averaged conceptual level is further substantiated by the -.05 correlation between these two main variables (see Table 4).

Univariate Comparisons

In this section, the results pertaining to the second hypothesis are presented. First, the intercorrelations between the outcome measures and the major variables of social networks and object relations, as well as life change, are examined. Second, the influence of the control variables, which include both the demographic variables and the variables of length of the description (of the subjects parents), number of items checked (NIC) and average intensity (AIN) from the ICL, is analyzed. Additional intercorrelations and other analyses are presented so as to shed more light on the outcome of the central hypothesis.

It was predicted that the variables of life change, total network size, total network density, and averaged conceptual level would each be correlated with the four outcome measures. Pearson product moment correlations were






60


performed between each of these variables. As can be seen



Table 3
Summary of the Correlations Between the Outcome Measures
and the Variables of Life Change, Total Network Size,
Total Network Density, and Averaged Conceptual Level


Total Total Averaged
Outcome Life Network Network Conceptual
Measures Change Size Density Level

POMS .26** -.07 -.l9* .04
HSCL .32*** -.03 -.18 -.001
Love Scale .21* .06 -.01 .04
GPA -.10 -.08 -.04 .07

*=p<.05. **=p<.005. ***=p<.0005.



in Table 3, the variables of total network size and averaged conceptual level are not significantly correlated with any of the four outcome measures. Total network density is significantly correlated (p<.05) with the POMS (a mood scale), but in a direction which was opposite to what had been predicted. This finding was also visible in Table 2. Although the differences between the measures were not significant, the distribution of means of the POMS scores across the three groups of density became smaller as the density of the group became larger. Now, Table 3 shows that, overall, as density of the network increases, the scores on the POMS become significantly lower (i.e., better). There was also a tendency (p<.06) for subjects in higher density networks to also produce better scores on the






61


HSCL (a symptom checklist). Thus, the results found in Table 3 suggest that higher density networks are correlated with lower levels of subjective distress. The variable of life change is also significantly with the P0OMS and the HSCL. As seen in Table 3, subjects who reported higher levels of life change in the nine months prior to the study had significantly (p<.005) better mood scores and significantly (p<.005) fewer symptoms. Unexpectedly, higher life change scores are significantly correlated (p<.05) with better Love Scale scores. Table 3 also shows that the role performance variable, GPA, is not correlated with any of these key variables.

Tables 4 through 9 present a more detailed analysis of the above results. In Table 4, the intercorrelations



Table 4
Correlation Matrix for the Variables of Life Change,
Total Network Size, Total Network Density, and Averaged Conceptual Level



Total Total
Life Network Network
Variables Change Size Density

Total Network Size .11
Total Network Density -.08 ~ 3*
Averaged Conceptual Level -.07 -.01 -.05

*=p<.Q5. **=p<.005. ***=p<.0005.



between each of the independent variables are reported. As expected, total network size and total network density are






62


correlated -.31 (p<.005) with each other. As can be observed in Table 4, total network size, total network density, and averaged conceptual level appear to be independent of the life change variable. As mentioned in the previous section, total network density and averaged conceptual level are only correlated -.05 with each other.

Table 5 is a summary of the intercorrelations between



Table 5
Correlation Matrix for the Outcome Measures


Outcome Love
Measures POMS HSCL Scale

HSCL .66*** --Love Scale -.20* -.12 -GPA -.25* -.09 .05
*=p<.05. **=p<.005. ***=p<.0005.



each of the outcome measures. This Table shows that the POMS is significantly correlated with all the other outcome measures. The POMS is, not surprisingly, strongly correlated with the other measure of subjective distress, the HSCL. Table 5 shows, though, that the HSCL is not related to the other two outcome measures. On the other hand, lower (i.e better) POMS scores are related to higher Love Scale scores and higher GPA's. Table 5 also indicates that although all the other intercorrelations were in the expected direction none were significant. Thus, Table 5






63


seems to suggest that the POMS shares a significant amount

of variance with the other three outcome measures.

In Table 6, the variable of conceptual level is



Table 6
Correlations Between the Subscales of the Outcome
Measures and the Conceptual Level Scores



Outcome Conceptual Level Scores
Measures Mother Father Averaged

POMS--Total -.06 .14 .04
Tension-Anxiety .02 .11 .07
Depression-Dejection -.08 .04 -.03
Anger-Hostility .03 .09 .06
Vigor -.08 -.24* -.17
Fatigue -.15 .07 -.04
Confusion-Bewilderment -.17 .02 -.08

HSCL--Total -.06 .06 -.001
Somatization -.07 -.01 -.05
Obsession -.08 .11 .02
Interpersonal -.07 .06 -.004
Depression -.10 .03 -.04
Anxiety .04 .08 .07

Love Scale Index -.02 .08 .04
Verbal Expression -.05 .05 .00001
Self-Disclosure .01 .12 .07
Toleration .07 .08 .08
Non-Material Evidence -.13 -.09 -.12
Unexpressed Feelings -.08 -.15 -.13
Material Evidence -.09 -.06 -.08

GPA .06 .07 .07
*=p<.05. **=p<.005. ***=p<.0005.



examined in greater detail with respect to the subscales of

the outcome measures. The conceptual level scores are

reported in three ways: the conceptual level of the written






64


description of the mother, the conceptual level of the written description of father, and the averaged conceptual level of both descriptions. The conceptual level of the description of mother correlates .70 (p<.0005) with the conceptual level of father. Averaged conceptual level is correlated .93 (p<.0005) with the conceptual level of mother and .91 (p<.0005) with the conceptual level of father. The POMS has six subscales (tension-anxiety, which correlates .74 with the total POMS score; depression-dejection, .84; anger-hostility, .62; vigor, -.64; fatigue, .75; confusion-bewilderment, .83), the HSCL has five (somatization, which correlates .81 with the total H-SCL score; obsession, .84; interpersonal, .75; depression, .79; anxiety, .89), and the Love Scale has six (verbal expression, which correlates .86 with the Love Scale Index, self-disclosure, .72; toleration, .70; non-material evidence, .54; unexpressed feelings, -.69; material evidence, .45). As can be seen in Table 6, only one of the 51 possible correlations between the three types of conceptual level scores and the 17 subscales is significant. Evidently, subjects with higher conceptual levels on their descriptions of father had significantly (p<.05) less vigorous moods (note that the vigor subscale of the POMS counts negatively toward the total POMS score). The otherwise general lack of significant correlations in Table

6 suggests that the results presented in Table 3 and the






65


means shown in Table 1 are correct; conceptual level appears

to be generally unrelated to the outcome measures.

In Table 7, total network size, total network density,



Table 7
Correlations Between the Subscales of the Outcome Measures and the Variables of Life Change, Total
Network Size, and Total Network Density



Total Total
Outcome Life Network Network
Measures Change Size Density

POMS--Total .26** -.07 -.19*
Tension-Anxiety .36*** -.11 -.13
Depression-Dejection .22* -.03 -.18*
Anger-Hostility .26* -.01 -.22*
Vigor .03 .08 .01
Fatigue .15 -.03 -.11
Confusion-Bewilderment .31** .05 -.22*

HSCL--Total .32*** -.03 -.18
Somatization .37*** -.003 -.12
Obsession .25* -.06 -.15
Interpersonal .22* .04 -.14
Depression .25* -.07 -.16
Anxiety .17 -.03 -.12

Love Scale Index .21* .06 -.01
Verbal Expression .15 .05 -.0004
Self-Disclosure .33*** .06 -.15
Toleration .31** .01 .04
Non-Material Evidence -.12 .12 .08
Unexpressed Feelings .05 .05 .10
Material Evidence .08 .08 .20*

GPA -.10 -.08 -.04

*=p<.05. **=p<.005. ***=p<.0005.



and the variable of life change are correlated with each of

the 17 subscales, with strikingly different results. As






66


indicated by Table 7, life change is significantly correlated with four of the six POMS subscales, with five of six of the HSCL subscales, and with two of the six Love Scale subscales. Subjects who reported greater degrees of life change in the nine months before the study appeared to have greater levels of tension-anxiety, depression-dejection, anger-hostility, and confusion-bewilderment. The mood subscale which are sensitive to activity level (i.e., vigor and fatigue) are not significantly correlated to life change. Subjects with higher life change also had symptoms of somatization, obsession, interpersonal difficulties, and depression. The positive correlation between higher levels of life change and the love scale appears to have been largely due to these subjects feeling as if they had more self-disclosing and tolerant relationships with their significant other.

Table 7 also supports some of the other results that were summarized in Table 3. As can be seen in Table 7, total network size is not only unrelated to any of the total socres of the outcome measures it is also unrelated to any of the subscales of the outcome measures. On the other hand, Table 7 permits a more careful analysis of the significant negative correlation between total network density and the POMS. A constellation of three subscales seems to have produced this result. As seen in Table 7, subjects with lower density networks were more likely to






67


have moods characterized by feelings of depression-dejection, anger-hostility, and confusion-bewilderment. Although none of the correlations between total network density and the five HSCL subscales approach significance, a remarkable consistency exists among these correlations. This consistency is reflected by the correlation of total network density with the total HSCL score being larger than any of the correlations between this density score and the HSCL subscales. Table 7 also shows a significant (p<.05) correlation between total network density and the material evidence of love as expressed on the Love Scale. The lack of even remotely significant correlations between total network density and the Love Scale subscales makes the one significant correlation appear to be an isolated phenomenon. Overall, the results presented in Table 7 appear to offer further substantiation of the findings seen in Table 3.

In Table 8, the subscales of the outcome measures are correlated with three other types of network density scores in an attempt to shed some further light on the unexpected correlations between total network density and subjective distress. Total network density is correlated .25 (p<.05) with family/relative network density, .67 (p<.0005) with friendship network density, and .55 (p<.0005) with nuclear family-friendship (NF-F) boundary density. Family/relative network density was not significantly correlated with either






68




Table 8
Correlations Between the Subscales of the Outcome
Measures and the Variables of Family/Relative Network
Density, Friendship Network Density, and Nuclear
Family-Friendship (NF-F) Boundary Density


Density Scores

Family/ FriendOutcome Relative ship NF-F
Measures Network Network Boundary

POMS--Total -.20* -.09 -.05
Tension-Anxiety -.13 -.05 -.01
Depression-Dejection -.23* -.12 -.03
Anger-Hostility -.17 -.08 -.03
Vigor .08 -.01 .02
Fatigue -.11 -.07 .04
Confusion-Bewilderment -.18 -.08 -.18

HSCL--Total -.13 -.12 -.09
Somatization -.10 -.16 .0001
Obsession -.12 -.08 -.10
Interpersonal -.08 -.02 -.07
Depression -.13 -.14 -.05
Anxiety -.07 -.09 -.10

Love Scale Index .13 .02 -.04
Verbal Expression .12 -.01 -.01
Self-Disclosure .06 -.04 -.13
Toleration .01 .04 .02
Non-Material Evidence .11 .08 -.01
Unexpressed Feelings -.07 .06 .07
Material Evidence .14 .16 .06

GPA .03 -.10 .11
*=p<.05. **=p<.005. ***=p<.0005.



friendship network density (.14) or with NF-F boundary

density (-.03). Friendship network density was correlated

.24 (p<.05) with NF-F boundary density. As can be seen in

Table 7 and Table 8, the family/relative network density






69

scores appeared to have behaved, with respect to the POMS subscales, in much the same manner as the total network density scores. Subjects in higher density family/relative networks had significantly (p<.05) lower (i.e., better) total POMS scores. As was the case with total network density, a three subscale constellation (depression-dejection, anger-hostility, and confusion-bewilderment) tended to be negatively associated with family/relative network density, although only the depression-dejection subscale was significant (p
As reported in Table 8, nuclear family-friendship

(NF-F) boundary density was not significantly correlated with any of the outcome measures or their subscales. Hirsch (1979a) had found NF-F boundary density to be the most influential of the various density scores. One factor that may have lessened the influence of NF-F boundary density in the current study is that only 49 percent of the subjects in the current study had a NF-F boundary density score higher than .00. In other words, only about one half of the subjects had any relationships between their family/relative







70


network and their friendship network. In Hirsch (1979a), 85 percent of the subjects reported at least one relationship between a friend and a family member. A possible mediator of this lack of boundary density in the current study could have been that, because many of the subjects were college students who lived away from home, they had little opportunity for their family and friends to meet.

The possibility that proximity influenced the NF-F

boundary density scores is examined in Table 9. One of the



Table 9
Summary of the Differences on Nuclear Family-Friendship
(NF-F) Boundary Density for the Subjects who Lived at
Home versus the Subjects who Lived Away from Home



Live At Live Away
Home From Home
Conditions (n=21) .(n=96)

Percent with NF-F 48 49
Boundary Density
Density > .00

Mean NF-F Boundary .07 .07
Density--All Subjects

Mean NF-F Boundary .15 .14
Density--Subjects
with Density > .00



demographic variables, domicile, was defined by whether or not the subject lived with his or her parents while attending the university. In Table 9, comparisons are made, for the variable of NF-F boundary density, between those






71


subjects who lived at home (n=21) versus those who lived away from home (n=96). As revealed in Table 9, the numbers between the two groups are remarkably similar, despite the difference in sample sizes. The percentage of subjects who had a NF-F boundary density greater than .00 (i.e., who had a relationship between their family and friends) was 48 for the subjects who lived at home and 49 for those who lived away from home. Of those subjects who had a NF-F boundary density score above .00, the mean density score was .15 (n=10) for those who lived at home and .14 (n=47) for those who lived away from home. The mean NF-F boundary density for all subjects was .07 for each of the two domicile groups. Thus, Table 9 appears to suggest that whether the subjects lived at home or not had little influence on their NF-F boundary density scores.

In Tables 10 through 14 the qualitative variables of object relations are compared first to each other, then to the outcome measures (and their subscales), and finally to the variables of life change, total network size, total network density, and averaged conceptual level.

Each of the qualitative factors derived from the

subjects' written description of their parents cannot be combined and averaged in the same fashion as the conceptual level scores. The striving factor derived from the description of mother is only correlated .10 with the striving factor derived from the description of father. The







-72


correlation on the nurturant factor between the description of father and the description of mother is .31 (p<.0005). This correlation, although significant, is still well under the .70 correlation between conceptual level for mother and the conceptual level for father. Of the four intercorrelations between striving (on mother and father) and nurturant (on mother and father), only the correlation between striving on mother and nurturant on father is significant (r=.29, p<.005). Thus, it seems that the qualitative factors of parent as striving on (mother and father) and parent as nurturant (on mother and father) should not be reduced to summary scores. Similarly, DOM and LOV from the subjects' completion of the ICL on their mother cannot be combined with the DOM and LOV from the ICL on their father because too much information would be lost. The subjects used the ICL to describe their mother and their father. The adjectives they used were then reduced to scores on two dimensions, the dominance-love dimension (DOM) and the love-hate dimension (LOy). Neither the correlation between DOM for mother and DOM for father (r=.02) or LOV for mother and LOV for father (r=.09) were significant. Only LOV for father and DOM for father were significantly correlated (r=.35, p<.0005). The other three intercorrelations, LOV for father and DOM for mother (r=-.03), LOV for mother and DOM for mother (r=-.07), and DOM for father and LOV for mother (r=.03), were not






73

significant. Thus, with only one exception, these dimensions appear to be independent of each other.

In Table 10, the qualitative factors of "parent as


Table 10
Correlations Between the Qualitative Factors of
"Parent as Striving" and "Parent as Nurturant"
and the ICL Dimensions (DOM and LOV)


DOM LOV
Variables Mother Father Mother Father

Striving--Mother .28** .04 -.04 .12
Striving--Father .04 .35*** .08 -.29**
Nurturant--Mother .25** -.01 .46*** .11
Nurturant--Father .09 -.02 .08 .52***

*=p<.05. **=p<.005. ***=p<.0005.


striving" and "parent as nurturant" are compared to the ICL dimensions, LOV and DOM. As expected, the striving factor and DOM and, likewise, the nurturant factor and LOV are significantly correlated. These correlations, as represented by the diagonal axis (top left to bottom right) of Table 10, are significant when the mother-mother and father-father comparisons are made. This set of intercorrelations is examined further in Table 30. None of the mother-father correlations were significant. Two other significant correlations are seen in Table 10. Striving on father is correlated -.29 (p<.005) with LOV for father, which might possibly be associated with DOM for father. The .25 (p<.05) correlation between nurturant on mother and DOM for mother is readily explanable.







74


Tables 11 and 12 report on the intercorrelations

between the outcome measures (and their subscales) and the

qualitative variables of object relations. In Table 11, the



Table 11
Correlations Between the Subscales of the Outcome Measures and the Variables of Parent As Striving and Parent As Nurturant



Parent As Parent As
Striving Nurturant
Outcome
Measures Mother Father Mother Father

POMS--Total -.15 -.003 -.24* -.21*
Tension-Anxiety -.09 -.03 -.28** -.15
Depression-Dejection -.17 -.09 -.21* -.14
Anger-Hostility -.05 .12 -.22* -.10
Vigor .10 -.11 .06 .14
Fatigue -.05 -.02 -.09 -.13
Confusion-Bewilderment -.20* -.08 -.28** -.29**

HSCL--Total -.05 -.11 -.14 -.23*
Somatization -.01 -.22 -.11 -.15
Obsession -.05 -.04 -.15 -.26**
Interpersonal .03 -.04 -.05 -.11
Depression -.12 -.13 -.16 -.21*
Anxiety -.03 -.04 -.08 -.16

Love Scale Index -.04 .01 -.04 .01
Verbal Expression .03 -.002 .02 .05
Self-Disclosure -.05 .08 -.16 -.16
Toleration -.02 -.04 -.17 .05
Non-Material Evidence -.07 -.16 .14 .17
Unexpressed Feelings .05 -.12 .08 .11
Material Evidence .01 -.08 .10 .17

GPA -.17 .002 .02 -.001

*=p<.05. **=p<.005. ***=p<.0005.



qualitative factors derived from the descriptions of parents

(i.e., parents as striving and parents as nurturant) are






75

correlated with the subscales of the outcome measures. As can be observed in Table 11, only one correlation is significant (p<.05) for "striving". Evidently, subjects whose description of their mother was high on the striving factor are less likely to report a confused and bewildered mood. The "nurturant" factor is more often significantly correlated with the outcome measures.

Table 11 illustrates that subjects whose description of their mother is high on the nurturant factor reported feeling less tension-anxiety, depression-dejection, anger-hostility, and confusion-bewilderment. As can be seen in Table 11, the nurturant factor derived from the description of the subjects' father is also significantly (p<.05) correlated with the measures of subjective distress (i.e., the POMS and the HSCL). Although all the subscales on the POMS are in the proper direction (i.e., suggesting that higher nurturant scores are related to better mood scores) only the confusion-bewilderment subscale is significantly (p<.005) correlated with the nurturant factor on father. The HSCL, and in particular the subscales that represent interpersonal difficulties and depression, is also significantly correlated with the nurturant factor on father. Thus, Table 11 seems to suggest that higher scores on the qualitative factor of "parent as nurturant" tend to correlate with reports of better mood and, in some cases, fewer symptoms. The scores for~ the qualitative factor of "parent as striving" seem less influential.






76


In Table 12, the correlations between the two



Table 12
Correlations Between the Subscales of the Outcome
Measures and the ICL Dimensions (DOM and LOV)


Outcome DOM LOV
Measures Mother Father Mother Father

POMS--Total -.34*** .06 -.09 -.14
Tension-Anxiety -.20* .04 -.02 -.09
Depression-Dejection -.26** .02 -.03 -.03
Anger-Hostility -.19* .15 -.09 -.27**
Vigor .36*** .07 .07 .07
Fatigue -.18 .10 -.12 -.13
Confusion-Bewilderment -.31** .02 -.10 -.08

HSCL--Total -.26** .02 .08 -.01
Somatization -.21* .04 .06 .01
Obsession -.14 -.04 .05 -.03
Interpersonal -.16 .01 .05 .07
Depression -.31** .01 .02 -.05
Anxiety -.25* .002 .17 -.05

Love Scale Index .06 -.04 .09 .18*
Verbal Expression .09 -.04 .04 .12
Self-Disclosure -.10 -.02 .04 .04
Toleration -.02 -.06 -.04 .18*
Non-Material Evidence .14 -.03 .20* .31**
Unexpressed Feelings -.01 -.10 .01 .06
Material Evidence .14 -.20* .21* .33***

GPA -.03 -.13 .03 -.02

*=p<.05. **=p<.005. ***=p<.0005.



dimensions of the ICL and the subscales of the outcome

measures are presented. As seen in Table 12, DOM for mother

and LOV for father seemed to each have a stronger

relationship with the outcome measures than did DOM for

father or LOV for mother. All but one of the subscales of






77


the POMS, with the exception of fatigue (p<.06), and three of the five subscales on the HSCL are significantly correlated with DOM for mother. With the exception of the anger-hostility subscale, DOM for father, LOV for mother, and LOV for father are not correlated with the measures of subjective distress. LOV for father is significantly correlated (P<.05) with the Love Scale index. Subjects who described their fathers as high on the Love dimension tended to report relationships which were more tolerant and allowed for more material and nonmaterial expression of love. LOV for mother is also positively correlated (p<.05) with material and nonmaterial expression on the Love Scale, but not with the total Love Scale score. The significant negative correlation between the material expression subscale of the Love Scale and DOM for father and between the anger-hostility subscale of the POMS and LOV for father seem to be isolated phenomenon. Thus, Table 11 appears to suggest that DOM for mother and LOV for father are the most sensitive dimensions of the ICL when related to the outcome measures.

In Table 13 the qualitative factors from the written

descriptions of parents is compared to the variables of life change, total network size, total network density, and averaged conceptual level. Only two correlations were significant. Table 13 confirms that, as also reported in Table 1, striving for father is significantly correlated







78



Table 13
Correlations Between the Variables of "Parent as
Striving" and "Parent as Nurturant" and the Variables
of Life Change, Total Network Size, Total Network
Density and Averaged Conceptual Level



Parent As Parent As
Striving Nurturant
Variables Mother Father Mother Father

Life Change .03 -.07 -.31** -.16
Total Network Size .08 -.06 .14 .10
Total Network Density .05 .03 -.01 .12
Conceptual Level .06 .34*** -.04 .08
*=p<.05. **=p<.005. ***=p<.0005.

Note. Conceptual Level is an Averaged score.


(p<.0005) with conceptual level. The nurturant factor on mother is correlated -.31 (p<.005) with the variable of life change. Subjects who described their mothers as high on nurturance tended to report less life change. This finding forces one to reconsider the results presented in Table 11. A comparison of Table 7 and Table 11 reveals that the variable of life change was correlated with the same subscales of the POMS as the nurturant factor on mother. At the same time, however, life change was significantly (p<.0005) correlated with the HSCL while the nurturant factor was not. Thus, while life change may have had some bearing on the interaction between the nurturant factor and the POMS, it is difficult to delineate a clear causative relationship.







79

In Table 14 the ICL dimensions are compared to the same



Table 14
Correlations Between the ICL Dimensions (DOM and LOV) and the Variables of Life Change, Total Network Size,
Total Network Density, and Averaged Conceptual Level



DOM LOV
Variables Mother Father Mother 'Father

Life Change .002 .19* -.10 -.02
Total Network Size .18 .15 .10 .07
Total Network Density .12 -.11 .05 .22*
Conceptual Level -.27** .03 .14 -.12

*=p<.05. **=p<.005. ***=p<.0005.

Note. Conceptual Level is an Averaged score.



four variables as above. Again, the relationship seen in Table 1 between DOM for mother and conceptual level is confirmed in Table 14. The relationship between the ICL dimensions and total network density that was seen in Table

2 is not, however, supported by Table 14. In Table 2 the means of DOM for father had been significantly different in the lower density group as compared to the higher density group. As shown in Table 14, DOM for father and total network density are correlated only -.11 with each other. On the other hand, LOV for father is correlated .22 (p<.05) with total network density. Table 14 also indicates that DOM for father is significantly correlated (p<.05) with life change. While the nurturant factor and life change were significantly correlated and shared a similar pattern on the







80

POMS, DOM for father appeared to be quite dissimilar from life change on the POMS. Thus, the impact of life change on the qualitative variables of object relations appears to be difficult to interpret.

Tables 15 through 28 are concerned with the control variables, which consist of three separate groups of variables. First, in Tables 15, 16, and 17 the variables in each of these three groups are compared within each group and then across the groups. Next, the variables in the first of these groups, the demographic variables, are compared in Table 18 to the variables of life change, total network size, total network density, and averaged conceptual level. Next, in Tables 19 and 20 the demographic variables are compared to the subscales of the outcome measures. In Tables 21 and 22 the demographic variables are correlated with the qualitative variables of object relations. In Tables 23, 24, and 25 a similar set of comparisons is made for the second group of control variables, the description length scores. In the final three tables of this section, Tables 26, 27, and 28, another three part comparison is made for the ICL control variables.

The first set of control variables are the demographic variables, which include the subjects' age, academic classification (or "class"), sex, SES, employment status (or "employed"), total number of semester credits (or "credits"). and domicile (i.e., whether the subjects lived







81

at home or not). These variables are described in the second chapter (pages 39 to 40). The significant intercorrelations among the variables that are presented in



Table 15
Correlation Matrix for the Demographic Variables


Variables Age Class Sex SES Employed Credits

C l a s s 7 1 * . .. .. .. .. .
Sex .18* .11 ...........
SES .02 -.13 .03 .........
Employed .31** .30** -.03 .08 ......
Credits .25* -.05 .01 .02 -.21" --Domicile -.03 .04 .14 -.01 .10 -.09
*=p<.05. **=p<.005. ***=p<.0005.

Note. For Class, l=sophomore and 3=senior; for
Sex, 0=males and 1= females; for SES, 1=highest and
5=lowest; for employed, 0=no and 1=yes; for domicile,
0=live away from home and l=live at home.


Table 15 are not unexpected. Subjects who were older tended to be in the upper academic classes (r=.71, p<.0005), to be employed (r=.31, p<.005), and to be registered for more credits (r=.25, p<.05). Those in the upper academic classes were more likely to be employed (r=.30, p<.005). And, the subjects who were employed had registered for fewer semester credits (r=.21, p<.05).
The description length scores represented the second

group of control variables. These scores were suggested by Blatt et al. (1981) to be a rough estimate of verbal fluency. The length of the subjects written description of






82

their mother was correlated .75 (p<.0005) with the length of their description of their father. Thus, these variables were combined and averaged. Both the length of the description of mother and the length of the description of father were correlated .93 (p<.0005) with this averaged score.
In Table 16 the demographic variables are correlated



Table 16
Correlations Between the Demographic Variables
and the Description Length Scores



Description Length Scores
Variables Mother Father Averaged

Age .001 .10 .05
Class .002 .05 .03
Sex .41*** .29** .38***
SES -.18 -.13 -.17
Employed .06 .05 .06
Credits .13 .05 .09
Domicile .11 .04 .08
*=p<.05. **=p<.005. ***=p<.0005.

Note. For Class, l=sophomore and 3=senior; for
Sex, 0=males and 1= females; for SES, highest and
5=lowest; for employed, 0=no and l=yes; for domicile,
0=live away from home and l=live at home.


with the description length scores. As seen in Table 16 only the variable of sex was significant. Women tended to write lengthier descriptions of their parents than did men. This finding was consistent with the norms presented by Blatt et al. (1981). A slight but nonsignificant trend was






83

seen in Table 16 between SES and length. This trend suggests that subjects in the higher social classes tended to produce longer descriptions.

The third group of control variables were derived from the ICL. These variables are the number of adjectives the subjects checked to describe their parents ("number of items checked", or NIC) and the average intensity (AIN) of the adjectives the subjects used in the description. while the subjects seemed to have a highly similar NIC score for both mother and father (r=.75, p<.0005), the AIN scores were not significant between the ICL on mother and the ICL on father. NIC and AIN were significantly correlated with each other both within (NIC for mother-AIN for mother, r=.29, p<.005; NIC for father-AIN for father, r=.25, p<.05) and across (NIC for father-AIN for father, r=.24, p<.05; NIC for father-AIN for mother, r=.21, p<.05) the sex of the parents.

In Table 17 the ICL control variables are correlated

with the other two groups of control variables. As seen in Table 17 only two correlations were significant. Subjects in lower academic classes and in lower socio-economic classes used significantly more intense adjectives to describe their mother. None of the correlations between the ICL control variables and the description length scores were significant. Thus, Tables 16 and 17 seem to suggest that each group of control variables accounts for a separate portion of the overall variance.






84



Table 17
Correlations Between the ICL Control Variables
(Number of Items Checked, NIC, and Average
Intensity, AIN) and the Demographic and
Description Length Variables



NIC AIN
Variables Mother Father Mother Father

Demographic
Age .11 .05 -.12 -.08
Class .03 .01 -.18* -.07
Sex .04 -.05 .07 -.01
SES .12 .07 .23* -.08
Employed .05 -.05 -.08 -.01
Credits -.03 -.02 -.01 -.05
Domicile -.14 -.16 -.15 -.06

Description Length
Mother .05 -.11 -.01 -.05
Father .14 .06 .04 .02
Averaged .10 -.02 .02 -.02
*=p<.05. **=p<.005. ***=p<.0005.

Note. For Class, l=sophomore and 3=senior; for
Sex, 0=males and 1= females; for SES, l=highest and
5=lowest; for employed, 0=no and l=yes; for domicile,
0=live away from home and l=live at home.


Table 18 reports on the correlations between the

demographic variables and the variables of life change,

total network size, total network density, and averaged

conceptual level. Only one correlation was significant in

Table 18. It appears that subjects in the lower academic

classes report more life change. Otherwise, these four

variables appear to be independent of the demographic

variables.






85




Table 18
Summary of the Correlations Between the Demographic Variables and the Variables of Life Change, Total Network Size, Total Network Density, and Averaged Conceptual Level


Total Total Averaged
Outcome Life Network Network Conceptual
Measures Change Size Density Level

Age -.12 -.03 .04 -.04
Class -.23* .06 .11 -.07
Sex -.05 -.08 -.005 .07
SES .12 -.12 .10 .03
Employed -.01 -.13 .09 -.01
Credits -.02 -.01 .11 .08
Domicile -.15 .09 .08 .05
*=p<.05. **=p<.005. ***=p<.0005.

Note. For Class, l=sophomore and 3=senior; for
Sex, 0=males and 1= females; for SES, l=highest and
5=lowest; for employed, 0=no and 1=yes; for domicile,
0=live away from home and l=live at home.


In Tables 19 and 20 the demographic variables are

compared with the subscales of the outcome measures. Three of the demographic variables, class, sex, and credits, appear in Tables 19 and 20 to have produced some interesting results. As seen in Table 19 subjects in the upper academic classes appear to report less dysphoric moods. More specifically, the upper academic class subjects, as with the older students, report significantly (p<.05) less fatigue. The subjects in the upper academic classes also tend, though not significantly, to attain better grades. Table 20 shows that subjects who had registered for more credits also have






86




Table 19
Correlations Between the Subscales of the Outcome
Measures and the Demographic Variables-Part A (Age, Class, Sex, and SES)



Outcome
Measures Age Class Sex SES

POMS--Total -.08 -.20* -.17 -.08
Tension-Anxiety .004 -.16 -.18 .12
Depression-Dejection .03 -.09 -.22* -.06
Anger-Hostility .01 -.07 -.25* .06
Vigor .14 .16 -.01 .24*
Fatigue -.25* -.25* -.02 -.12
Confusion-Bewilderment .06 -.07 -.20* .02

HSCL--Total .06 -.09 -.06 .01
Somatization .04 -.05 -.03 .01
Obsession .18* -.01 -.15 .11
Interpersonal .02 -.10 .05 -.04
Depression -.01 -.14 -.07 -.04
Anxiety .06 -.02 -.02 .02

Love Scale Index -.16 -.15 .36*** .07
Verbal Expression -.18 -.15 .27*** .12
Self-Disclosure -.10 -.14 .31** -.06
Toleration -.03 -.05 .22* .07
Non-Material Evidence -.10 -.07 .29** .02
Unexpressed Feelings .17 .12 -.19* -.02
Material Evidence .02 -.02 .19* .14

GPA -.06 .17 .08 -.09

*=p<.05. **=p<.005. ***=p<.0005.

Note. For Class, 1=sophomore and 3=senior; for
Sex, 0=males and 1= females; for SES, l=highest and
5=lowest.


significantly (p<.0005) higher GPA's. Table 19 reports

that, as previously mentioned in the section on normative

comparisons, female students had significantly (p<.0005)

higher Love Scale scores, a finding which was true for every






87




Table 20
Correlations Between the Subscales of the Outcome
Measures and the Demographic Variables-Part B (Employed, Credits, and Domicile)



Outcome
Measures Employed Credits Domicile

POMS--Total .01 .06 -.14
Tension-Anxiety .02 .15 -.12
Depression-Dejection -.02 -.02 -.20*
Anger-Hostility .14 .02 -.06
Vigor -.02 .01 -.01
Fatigue -.06 .1 -.06
Confusion-Bewilderment -.03 .03 -.17

HSCL--Total .06 -.04 -.16
Somatization .05 -.03 -.15
Obsession .03 -.07 -.11
Interpersonal .08 -.03 -.04
Depression .05 -.05 -.16
Anxiety -.003 .05 -.13

Love Scale Index -.02 .03 .17
Verbal Expression -.01 -.02 .24*
Self-Disclosure .06 -.06 .12
Toleration .07 -.06 .08
Non-Material Evidence -.11 .11 .10
Unexpressed Feelings .08 -.15 -.03
Material Evidence -.001 -.04 .10

GPA -.02 .34*** .07
*=p<.05. **=p<.005. ***=p<.0005.

Note. For employed, 0=no and l=yes; for domicile,
0=live away from home and l=live at home.


subscale (note that "unexpressed feelings" counts negatively

toward the total score). As seen in Table 19, women also

appeared to report on the POMS that they felt significantly

(p<.05) less depression-dejection, anger-hostility, and






88

confusion-bewilderment. Tables 19 and 20 also report five other significant correlations; however, these correlations do not appear to represent any coherent pattern of relationships among the variables. Thus, Tables 18, 19, and 20 appear to suggest that several of the demographic variables, while being of little influence on the main variables of the current study, do have a significant influence on some of the outcome measures.

Tables 21 and 22 show that some of the demographic


Table 21
Correlations Between the Qualitative Factors of
Object Relations and the Demographic Variables-Part A (Age, Class, Sex, and SES)


Variables Age Class Sex SES

Striving--Mother -.04 -.09 .002 -.02
Striving--Father -.05 -.004 .05 -.16
Nurturant--Mother .15 .23* .11 -.20*
Nurturant--Father .11 .09 .07 -.12
DOM--Mother -.03 .07 -.01 -.04
DOM--Father -.21* -.08 .01 -.05
LOV--Mother .19* .23* .16 -.08
LOV--Father .14 .04 .27** .03
*=p<.05. **=p<.005. ***=p<.0005.

Note. For Class, l=sophomore and 3=senior; for
Sex, 0=males and 1= females; for SES, 1=highest and
5=lowest.


variables can influence the qualitative variables of object relations. Table 21 suggests that the most consistent interaction (r=.27, p<.005) is found between women and LOV






89


from the ICL for father. Tables 12 and 19 have indicated that each of these variables is related to the Love Scale. When these results are combined, it appears that women tend to describe their fathers higher on LOV and to report on the Love Scale a better rapport and liason with a significant other. Though six other significant correlations are


Table 22
Correlations Between the Qualitative Factors of
Object Relations and the Demographic Variables-Part B (Employed, Credits, and Domicile)



Variables Employed Credits Domicile

Striving--Mother -.05 -.06 .02
Striving--Father .004 -.01 -.09
Nurturant--Mother .03 -.16 .09
Nurturant--Father -.19* -.06 .09
DOM--Mother .03 -.10 .09
DOM--Father -.13 .04 -.18
LOV--Mother .15 -.12 .10
LOV--Father -.1 -.07 .05
*=p<.05. **=p<.005. ***=p<.0005.

Note. For employed, 0=no and 1=yes; for domicile,
0=live away from home and 1=live at home.


reported in Tables 21 and 22, the impact of the other demographic variables on the qualitative variables does not appear to be related to a larger schema.

The description length scores, as control variables, appear to be less influential than the demographic variables. Table 23 indicates that the variables of life change, total network size, total network density, and






90



Table 23
Correlations Between the Description Length Scores
and the Variables of Life Change, Total Network Size,
Total Network Density, and Averaged Conceptual Level


Description Length Scores
Variables Mother Father Averaged

Life Change -.01 -.002 -.01
Total Network Size .15 .09 .13
Total Network Density -.05 -.03 -.04
Averaged Conceptual Level -.05 .01 -.05


averaged conceptual level are not significantly correlated with the description length scores. Thus, as predicted, the conceptual level of the written description appears to be independent of the length of that description.

In Table 24 only two of the the subscales of the

outcome measures are related to the length of the written descriptions of parents. As seen in Table 24, subjects who wrote lengthier descriptions appeared to report significantly (p<.05) less confused and bewildered moods. And, subjects who had longer descriptions after 5 minutes of describing their mother and father also tended to report fewer symptoms of obsession. (Of course one wonders how this last correlation may have been different if the obsessive subjects were given unlimited amounts of time to describe their parents.) Thus, Table 24 suggests that, although the description length scores were related to the outcome measures in ways that are easily understood, the