 

 

 

 

 

 

 

 

 

 

 

 

 

.r. g. in
3&9».

. WW»?

9.. . (t.
3-}, I

. 4.5

. A .)__

s

HESlS

1063

This is to certify that the

dissertation entitled

The Relationship of Identity Status,
Perceived Parental Explanatory Style, and
Sex to Young Adult Attributional Style and

Attributions for Failure
presented by

Carla Marie Monestere

has been accepted towards fulﬁllment
of the requirements for

Ph . D . degree in Psychology

 

 

Major professor

Date 8/12/2002

 

MSU 1': an Afﬁrmative Action/Equal Opportunity Institution 0-12771

 

 

Ll

 

 

BRARY

Michigan State,
University

 

 

PMCE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

 

r91? E DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CiRC/DaieDue.p65-p.15

ﬂ

THE RELATIONSHIP OF IDENTITY STATUS, PERCEIVED PARENTAL
EXPLANATORY STYLE, AND SEX TO YOUNG ADULT ATTRIBUTIONAL
STYLE AND ATTRIBUTIONS FOR FAILURE '

By

Carla Marie Monestere

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PPHLOSOPHY
Department of Psychology
2002

ABSTRACT

THE RELATIONSHIP OF IDENTITY STATUS, PERCEIVED PARENTAL
EXPLANATORY STYLE, AND SEX To YOUNG ADULT ATTRIBUTIONAL
STYLE AND ATTRIBUTIONS FOR FAILURE
By

Carla Marie Monestere

This research placed attributional theory within a broader framework by
examining its relationships to developmental, familial, and societal factors. First, the
research examined sex differences and the constancy of attributional style (AS) across
content areas. It also reviewed the relationship of AS to identity status (ID) and sex
differences therein. Another focus was sex differences in the relationship of young adult
attributional style (YAS) to perceived maternal and paternal attributional styles. Finally,
it examined the relationship of these factors to attributions for and mood after failure.

There were 278 men, 494 women, and one unknown person in Phase I. These
participants completed questionnaires addressing demographics, AS, perceived parental
attributional style, depressed mood, and life stressors. Seventy-three volunteers from
Phase I then did Phase H which entailed completing one of three sets of anagrams (of
varying degrees of difficulty), making attributions for failure, and reporting on mood.

There was support for domain speciﬁc attributional diatheses and sex differences
therein. YAS scores for interpersonal events were signiﬁcantly (p < .05) lower than
scores for achievement events. Men’s intemality attributions for achievement events
were signiﬁcantly more external than the corresponding women’s attributions. YAS and
[D were unrelated. There were signiﬁcant positive relationships between YAS and
perceived paternal and maternal attributions for both achievement and interpersonal

events. When young adult attributions were regressed on the corresponding parental

attributions, perceptions of father generally accounted for more variance than perceived
maternal attributions. In a structural equation model of Phase I data, there were more
Signiﬁcant paths from perceived paternal attributions (as opposed to perceived maternal
ones) to young adult attributions. Compared to men’s, women’s overall attributions were
related to more diverse perceptions of parental attributions. There were fewer total paths
from perceived parental attributions to men’s overall attributions, but the number of paths
from perceptions of mothers and fathers was more equal. In Phase II, only perceived
maternal generality attributions for achievement events predicted attributions for failure.

The study revealed domain speciﬁc attributional differences and sex differences
that were consistent with socialization and indicate that societal factors may relate to
YAS. YAS and ID were not related. The data suggested that young adults utilize parents
as models and develop similar attributional styles. Findings also suggested that males and
females incorporate perceptions of parents differently. Contrary to hypotheses, only
perceptions of maternal attributional style predicted young adult attributions for failure.
This ﬁnding may reﬂect a developmentally appropriate connection to parents.

Findings suggest the importance of attending to sex differences, attiibutional
domains, and societal norms when studying AS. Results also point to the role of parents
in development of YAS and raise the possibility of developmental changes in whether an

individual’s AS (or perceptions of parents) will predict failure attributions.

COPyright by
CARLA MARIE MONESTERE
2002

ACKNOWLEDGEMENTS

An immense thank you to Elizabeth whose love, support, and unwavering
optimism made this work possible. You may not want me to gush, but you gave me the
time and space to do this work even when it was hard for you to do so. I appreciate it.

To Dozier W. Thornton, a phenomenal advisor, thank you for the gift of your
support, attention, and interest. Your constructive input and your ability to see the
positive helped me to reach my potential and become a psychologist. You are among the
most decent human beings I know and I am honored to have worked with you.

I am deeply grateful to my committee members: Norman Abeles, Ph.D., Bertram
Karon, Ph.D., and Lawrence van Egeren, Ph.D.. Each of you made a commitment to my
training, educated me, and gave insightful feedback that improved my work and skills.

Thank you, Aunt Ella. You have provided support on every level and I am
profoundly thankful. You enabled me to engage fully in this work and helped me to ﬁnd
out what I could do. Thank you for investing so much and helping me to grow.

I thank all my friends from CAPS at the U of M who helped me to rediscover the
joy in clinical work and expanded my view of psychology. I offer particular thanks to
Millie Tirado, Ph.D., Vicki Hays, Ph.D., Deb Kraus, Ph.D., Alice Brunner, Ph.D., Torn
Baez, Ph.D. and Lynne Rose, A.C.S.W.. Your thoughtful and caring guidance (especially
about perfectionism) as well as your prayers and encouragement sustained me.

I am thankful for all of the named and unnamed ﬁiends who have enriched my
life during the dissertation process. Thank you especially to Krista Suess (for humorand
walks), the Suess family, the Heimler-Rosen family, the Glew family, Karen Ryder,
Carol Hein-Creger, Grace Chow, Nancy Bischof, Adrianne Camero-Sulak, Maria
VandenBoom, Derek Zmich, David O’Dowd, Breeze, Dancer, Moe, Lucy, Jonah,

Aloysius, Asher, and Gus, Luti, and Ecclesia.

TABLE OF CONTENTS

LIST OF TABLES ..................................................... viii
LIST OF FIGURES ..................................................... ix
KEY TO ABBREVIATIONS ............................................. x
Chapter One ............................................................ 1
Chapter Two: Literature Review ............................................ 4
Theories of Causality ............................................... 4
Attributional Style for Achievement and Interpersonal Related Events:
Relationship to Sex ......................................... 10
Attributional Style and Perceptions of Parental Attributional Style .......... 14
Identity Status ................................................... 19
Attributional Style, Perceptions of Parental Attributional Style, Identity Status
and Response to Failure ...................................... 22
Chapter Three: Methods ................................................. 25
Method and Design ............................................... 25
Past Work ....................................................... 25
Participants ...................................................... 26
Procedures ...................................................... 27
Session I .................................................. 27
Session II ................................................. 28
Measures ....................................................... 29
Session I .................................................. 29
Session H ................................................. 33
Hypotheses ...................................................... 35
Section One Hypotheses ........................................... 35
Identity status ........................................ 35
Attributional style .................................... 35
Section One Analyses ............................................. 36
Section Two Hypotheses ........................................... 37
Section Two Analyses ............................................. 37
Section Three Hypotheses .......................................... 39
Section Three Analyses ............................................ 39
Section Four Hypotheses ........................................... 40
Section Four Analyses ............................................. 41
Chapter Four: Results ................................................... 43
Section 1 Results ................................................. 43
Section 2 Results ................................................. 47
Section 3 Results ................................................. 51
Section Four Results .............................................. 54
SEM Analysis for Phase 1 .......................................... 56
SEM Phase H Analysis ............................................. 59
Chapter Five: Discussion ................................................. 62
Attributional Style and Identity Status ................................. 62
Societal Roles and Norms: Relationship to Attributional Domain and Sex
Differences ................................................ 64

vi

Attributional Style and Family Factors ................................ 66
Relationship of Young Adult and Perceived Parental Attributions to Attributions

for and Mood after Failure .................................... 70
APPENDIX A: Dealing with missing identity status data ........................ 75
APPENDIX B: Figures .................................................. 76
REFERENCES ........................................................ 84

vii

LIST OF TABLES

Table 1: Regression of Women’s Attributions on BDI Scores by Attributional Domain
............................................................... 44

Table 2: Mean Attribution Scores by Sex and Attributional Domain ............... 44
Table 3: Regression of Men’s Attributions on BDI Scores by Attributional Domain . . 46

Table 4: Reggession of Young Adult Attributions on BDI Scores by Attributional

Domain ......................................................... 48
Table 5: Regression of Ideological and Interpersonal Identig Status on You_ng Adult

Attributions for Achievement ....................................... 50
Table 6: Reggession of Ideological and Interpersonal Identity Status on Young Adult

Attributions for Interpersonal Events .................................. 50

Table 7: Summm of Regressions of Young Adult Attributions on Perceived Paternal
and Maternal Attributions by Sex an_d by Attributional Dom_ain_g_ ............. 52

Table 8: Overview bLSex and Attributional Domain of Regrpssion Coefficients from
Regressing Young Adult Attributions onto Perceived Parental Attributions . . . 53

Table 9. Sex differences In the relationship of young adult attributions to perceived

 

parental attributions In the intemrsonal domain ......................... 54
Table 10: Sex Differences in the Relationship of Young Adult Attributions to Perceived

Parental Attributions in the Achievement Domain ....................... 55
Table 11: Unstandardized Parameter Estimates for the Men’s Model .............. 58
Table 12: Unstandardized Parameter Estimates for the Women’s Model ........... 59

Table 13: Unstandardized Parameter Estimates for Phase H F m Model ........... 61

viii

LIST OF FIGURES

Figure l: Latent variable multiple groups model showing relationships among
attributional styles and identity statuses by sex ......................... 76

Figure 2: Latent variable multiple groups model showing relationships among
perceived parental attributional styles, young adult attributional styles, and
identity statuses by sex ............................................ 77

Figure 3: Model testing failure attribution as predicted by young adult and perceived
parental attributions for achievement ................................. 78

Figure 4: Hypothesized model of the relationships of domain speciﬁc perceived
parental attributions to overall young adult attributions and identity statuses,
run separately for men and women ................................... 79

Figure 5: Final model for men’s overall attributional styles and domain-speciﬁc
perceived parental attributional styles ................................ 80

Figure 6: Final model for women’s overall attributional styles and domain—speciﬁc
perceived parental attributional styles ................................ 81

Figure 7: Model testing failure attributions as predicted by young adult and
perceived parental generality attributions for achievement events .......... 82

Figure 8: Final model testing failure attributions as predicted by young adult and
perceived parental generality attributions for achievement events .......... 83

ix

ANOVA
AS

ASQ

BDI

CFI

DACL
EASQ
EOMEIS
GFI

ID

ID-ID
IP-ID

LES
MANOVA
NFI

NNFI

PAQ

PEM
PMAS-ACH
PMAS-INP
PPAS-ACH
PPAS-INP
PPASQ
SEM

SES

VAS

KEY TO ABBREVIATIONS

Analysis of Variance

attributional style

Attributional Styles Questionnaire

Beck Depression Inventory

Comparative Fit Index

Depression Adjective Checklist

Expanded Attributional Styles Questionnaire

Extended Objective Measure of Ego Identity Status
Goodness of Fit Index

identity status

ideological identity status

interpersonal identity status

Life Experiences Survey

Multivariate Analyses of Variance

Normed Fit Index

Non-Normed Fit Index

Particular Attribution Questionnaire

Performance Evaluation Measure

perceived maternal attributional style for achievement events
perceived maternal attributional style for interpersonal events
perceived paternal attributional style for achievement events
perceived paternal attributional style for interpersonal events
Perceived Parental Attributional Style Questionnaire
Structural Equation Modeling

socioeconomic status

Visual Analog Scale

YA young adult

YAS young adult attributional style

YAS-ACH young adult attributional style for achievement
YAS-IN P young adult attributional style for interpersonal events

xi

Chapter One

Human beings have asked, “Why me?” for thousands of years. For example, one
hears it in the Book of Job as Job faces one form of adversity after another. Confronting
adversity and failure is part of the human experience, yet some individuals are crushed by
it while others overcome and ﬂourish. What permits a person to persevere and respond
constructively to failures and challenges?

Many theories have addressed the manner in which people respond to adversity
(Lazarus, 1991, 1993; Rotter, 1966; Abramson, Metalsky, & Alloy, 1989; Abramson,
Seligrnan, & Teasdale, 197 8). The reformulated theory of learned helplessness
(Abramson et al., 1978) has stimulated much research about people’s responses to
negative events. This model is based on the theory of learned helplessness (Seligman,
1975) which stems from animal research. This theory argued that organisms learn to give
up in situations where they cannot control the outcome. Abramson et al. (197 8) added an
attributional component to the model when the theory did not adequately account for
human behavior. They argued that when a person does not control an outcome, the
individual asks “Why?” and makes attributions about three causal dimensions
internality/extemality (Where is the cause?), stability/instability (How enduring is the
cause?), and globality/specificity (How widespread is the impact of the cause?)
Abramson et al. hypothesized that an internal, stable, and global attributional style
produces cognitive, emotional, and motivational deﬁcits that make a person vulnerable to
depression when stressors occur. Studies have shown relationships between attributional
style and depression (Peterson & Seligman, 1984), and health (Peterson, Seligman, &
Valliant, 1988), and a person’s response to success and failure (Metalsky, Halberstadt, &
Abramson, 1987). Studies of success and failure situations have found relationships
between attributions and the duration (Metalsky et al., 1987) and the severity of
depressed mood (Follette & Jacobson, 1987).

Although Abramson et al.’s (1978) model and much subsequent research testing it

focused on adults, it was a logical step to examine attributional style and its relationship
to mood and behavior in children (Kaslow, Rehm, & Siegel, 1984; Seligrnan, Peterson,
Kaslow, Tanenbaum, Alloy, & Abramson, 1984; Ward, F riedlander, & Silverman, 1987).
Given the limited scope of learned helplessness theory (Abramson et al., 1978; Seligman,
1975) and the relevance of attributional style to functioning in adults and children, it
would be good to articulate further the relationship of attributional style to development,
e.g. identity status (Marcia, 1966) and to familial inﬂuences, e.g. perceptions of parents.
Understanding the relationship of developmental and familial factors to attributional style
is important because attributional style can affect a young adult’s ability to respond
effectively to failure and thereby inﬂuence the trajectory of the individual’s life.
Additionally, given the relationship between attributional style and depression, and the
reality that depression is a signiﬁcant mental health concern in the US, understanding
more about what factors are related to attributional style could be immrtant for
improving prevention and treatment efforts.

This research was an effort to begin to put attributional theory within a broader
conceptual framework by exploring its relationship to developmental and familial factors.
More speciﬁcally, it examined the relationship of attributional style to identity status and
sex differences therein. Identity status (Marcia, 1966) is an indicator of young adult
development. It is a measure of a young adult’s exploration of and commitment to an
adult role and a set of beliefs in interpersonal and ideological realms. The study also 1
examined the relationship of attributional style to perceptions of parents and sex
differences in these relationships. Perceptions of parents may also be important given that
parents serve as models and teachers as children grow. Young adults rely on their parents
in many ways. It is possible that young adults’ perceptions of their parents’ explanatory
styles might relate to their own attributional styles. The research also studied the
constancy of attributional style by exploring whether individuals exhibit domain speciﬁc

attributional diatheses. For example, do individuals demonstrate different attributional

styles when responding to interpersonal events as compared to achievement events (both
of which are subject to strong sociocultural beliefs)? This study investigated sex
differences in attributional style for interpersonal events and for achievement/work
events. Additionally, the study examined the relationship of attributional style, perceived
parental attributional style, identity status, and sex to attributions for failure and
subsequent mood.

The present research used an existing data set and therefore the sample, methods,
and measures were pre-set. The study focused on constructs assessed in the original

project.

Chapter Two: Literature Review
Theories of Causalig

Kelley's (1967) covariation theory, Rotter's (1966) locus of control model,
Weiner, Frieze, Kukla, Reed, Rest, & Rosenbaum’s (1971) model of causal ascriptions,
and Abramson et al.’s (1978) reformulated theory of learned helplessness addressed how
people explain events. In Kelley's (1967) model, people confronted by ambiguity or a
challenge to their assumptions will logically evaluate covariation along three dimensions:
distinctiveness, consistency over time, and consensus, and then make causal attributions.
People may attribute causality to the self, to others, or to the environment. Attributions
provide guidelines for future actions. According to Rotter (1966), stable individual
differences determine the nature of a person's causal inferences. Rotter argued that people
differ in their expectations about their ability to control positive and negative
reinforcements. Individuals who believe that they can control the occurrence of
reinforcers are said to have an internal locus of control. In contrast, people with an
external locus of control perceive something outside of themselves as controlling the
reinforcements they receive. These models incorporate an internal/extemal dimension
and can be applied to a variety of settings, but they merely allude to how attributions are
connected to mood. The covariation theory also requires a very logical analysis that is in
reality not error free (Crocker, 1981). In addition, by postulating an internal trait, Rotter's
theory neglects environmental factors.

Weiner et a1. (1971) proposed a model of causal ascriptions for achievement
behavior. In this model, people ascribe the outcome of achievement events to ability or
effort (internal factors) or to task difﬁculty or luck (external factors). They argued that
ascriptions to internal causes generate more intense positive and negative affective
reactions than ascriptions to external sources (Weiner, Heckhausen, Meyer, & Cook,
1972). Internal attributions for failure and external attributions for success might generate

negative moods. Later versions of the theory added attributions about stability and

controllability. Attributions about stability are hypothesized to determine expectations for
the future whereas controllability and locus of control (intemality/extemality) are
construed as determining emotional response. Expectations and mood determine future
performance. Weiner, Russell, & Lerman (1979) further articulated the relationship of
outcome, attributions, and affect. They argued that the objective outcome (success or
failure) produces positive or negative affects independently of ascriptions, which also
engender affect. This model, like those mentioned previously has an internal/external
dimension, but differs in that it speciﬁes the relationship of causal ascriptions to mood
and performance. This theory, however, does not address attributions about the
speciﬁcity of causes. It is also speciﬁc to achievement behavior.

Seligrnan’s (1975) conceptualization of depression, the theory of learned
helplessness, included an attributional component (Abramson et al., 1978). In its original
formulation (Seligrnan, 1975), the theory asserted that when an organism cannot control
an event (outcomes are independent of actions) deﬁcits in motivation, cognition, and
emotion occur. Therefore, when a person cannot control a situation, theory predicts that
the person will give up and become depressed. This theory does not explain when
helplessness will or will not generalize across situations, why negative events do not
always cause a depression, or how the loss of self-esteem frequently associated with
depression occurs (Peterson & Seligrnan, 1984). The major change in the reformulated
theory of learned helplessness (Abramson et al., 1978) was the inclusion of an
attributional framework; attributions about the causes of events were viewed as altering
expectations and thereby affecting mood. If a person fails to control an outcome, s/he
asks why it has occurred. The answer to this question determines the generality and the
chronicity of the person's response as well as the event's impact on the individual’s self-
esteem. The reformulation proposed that people make causal attributions along three
orthogonal dimensions: intemality vs. extemality (e.g. IS the cause something that is

within me or in the Situation?), stability vs. transience (e.g. Is the cause something that is

constant or temporary?), and globality vs. speciﬁcity (e. g. How pervasive is the causal
factor? Does it apply to one situation or to a variety?) Individuals who attribute negative
events to internal reasons are expected to suffer a loss of self-esteem. Individuals who
attribute negative events to a persistent cause are expected to remain depressed longer. If
one believes that a pervasive factor has caused a negative event, one is apt to Show
helplessness deﬁcits in a variety of settings. Therefore, individuals who attribute
causality to internal, stable, and global factors are most susceptible to becoming
depressed when confronted with negative life events; in effect, a diathesis-stress model
was proposed. In a subsequent reformulation, Abramson, Alloy, & Metalsky (1986)
argued that depressive reactions are "...more likely to occur, to be more intense, and to
last longer when negative life events are attributed to stable and global causes..."
(Metalsky et al., 1987, p. 386) as opposed to unstable, speciﬁc causes. The
intemality/extemality dimension is treated as inﬂuencing self-esteem, but is not the
predisposing factor (diathesis) for depression. Later research is consistent with this
perspective. Joiner and Rudd (1996) showed that speciﬁcity and stability load on the
same factor (attributional generality) and that intemality loads on a separate factor that
pertains to the individual’s external focus.

Abramson et a1. (1989) developed the theory of hopelessness depression to
respond to criticisms (Coyne & Gotlib, 1983) of the reformulated theory of learned
helplessness (Abramson et al., 1978). In this model, hopelessness is construed as the
proximal and sufficient cause of what they call hopelessness depression. The path to
hopelessness depression begins either with the occurrence of a negative event or with the
nonoccurrence of a positive event. The model then posits that three diatheses can interact
with the event and contribute to the development of depression. The ﬁrst diathesis is
attributing the event to internal, stable, global causes and attaching great importance to
the event. The second diathesis is inferring negative consequences based on the event.

The third diathesis is inferring negative characteristics about the self alter a negative

event occurs.

In contrast to the reformulated theory of learned helplessness (Abramson et al.,
1978), this model (Abramson et al., 1989) treats attributional style as a distal cause that
can inﬂuence a person’s attributions for a particular event. Attributional style is a
diathesis that only functions in the presence of a negative life event. Abramson et al.
(1989) also argued that a diathesis stress model implies that a person will be vulnerable
to a particular stressor if his/her attributional style for that type of event is depressogenic
(internal, stable, and global); there is a match between the stressor and the attributional
vulnerability. Thus, one can speak of domain speciﬁc attributional diathesis; for example,
a college student explains achievement, but not interpersonal, events using internal,
stable, and global causes.

Hopelessness theory is signiﬁcant in several ways. First, it attempts to put
attributional style in a broader conceptual framework. Second, it changes the role of
attributional style in understanding depression and points to the possibility that other
factors may inﬂuence it and the way it relates to mood. Third, hopelessness theory
suggests that any negative event (controlled or not) can serve as a trigger. Fourth, it
highlights the possibility that individuals may have content speciﬁc attributional style
vulnerabilities. Even so, the model does not emphasize the role or relationship of
developmental factors to explanatory processes.

In contrast to the other models the reformulated theory of learned helplessness
posits a chronic attributional style reﬂecting past experience and reinforcements. Like
Weiner et al.'s model (1979), it links cognition to mood. However, only the reformulated
theory of learned helplessness addresses motivation and self-esteem. In addition, the
reformulated theory of learned helplessness does not require rational causal thinking at
all times and is not limited to explaining one realm of behavior.

Although constructed to explain depression, the reformulated theory of learned

helplessness has also been used to examine attributions about and emotional responses to

failure (F ollette & Jacobson, 1987; Metalsky et al., 1987). Metalsky et al.- (1987) tested
the diathesis-stress and causal mediational components of the reformulated theory using a
sample of college students who had taken a midterm exam. For students who failed the
exam, a regression revealed that the averages of their combined scores on the speciﬁcity
and stability dimensions of the Attributional Style Questionnaire (the generality score)
predicted the duration of their depressed mood in a manner consistent with the
reformulated theory. Mood immediately after the test, hoWever, was solely a function of
test outcome. Although this ﬁnding is consistent with Weiner et al.'s (1979) assertions, it
may mean that people require time to decide what caused the failure and readjust mood
accordingly. Metalsky et al. also found that particular attributions made to address a
speciﬁc failure mediate the effect of attributional style on the duration of depressed
mood. Follette & Jacobson (1987) conducted a similar study using college students.
Although they found no relationship between attributional style (operationalized as the
average of the intemality, speciﬁcity, and stability scores) and depression immediately
after receiving test grades, the interaction of speciﬁc attributions about failing the exam
with expectations about performance did predict immediate depression. Additionally, in a
study of undergraduates receiving inaccurate failure feedback about their performance on
a general ability test, Houston (1995) found that generality scores (the average of
stability and globality items) for achievement events were predictive of change in
measured anxiety, but not in depression. These results are consistent with the idea that
attributional style can act as a diathesis for negative mood. Houston (1995) did not ﬁnd
support for causal mediation.

Researchers have begun to examine attributional style within a broader context.
Studies have demonstrated that a variety of factors including affect (Baumgardner &
Arkin, 1988; Brown, 1984), reinforcement contingencies (Weiner et al., 1971), self.-
efficacy (Houston, 1995), self-esteem (Metalsky, Joiner, Hardin, & Abramson, 1993) and

expectations (House, 1976) can alter the nature of a person's causal attributions. Other

researchers have started examining children's and adolescents’ causal attributions
(Boggiano, 1998; Dixon & Ahrens, 1992; Dweck, 1975; Lewinsohn, Rohde, Seeley,
Klein, & Gotlib, 2000; Nelson & Cooper, 1997). Even so, the relationship of
developmental and familial factors to adult functioning has been insufﬁciently studied.
Many studies have equated adults of all ages; psychosocial or developmental factors are
assumed to have no bearing on causal attributions. In effect, young adults seeking a role
in the world and closely tied to their parents are considered the same as elderly persons
who may be reﬂecting on their lives in a quest for meaning. Although a few studies have
examined sex differences (Handal, Gist, Wiener, 1987; Johnson, 1992; Whitley, Michael,
& Tremont, 1991), many studies have not; in effect these studies assumed that the
differences in socialization of women and men do not alter attributional style.

In summary, Abramson et al.'s (1978) model posited that people explain events
by making attributions along three dimensions - intemality/extemality, speciﬁcity, and
globality. A later reformulation (Abramson et al., 1986) argued that speciﬁcity and
stability are critical for determining whether or not a person becomes depressed and that
intemality inﬂuences self-esteem. Another model (Abramson et al., 1989) intended to
address the shortcomings of Abramson et al.’s (1978) reformulated theory suggested that
other factors can interact with attributional style and that the lack of control is not
essential. This model also suggested that individuals may exhibit domain speciﬁc
attributional vulnerabilities. For example, an individual may make internal, stable and
global attributions for negative events in his/her personal life, but he/ she may make
external, unstable, and Speciﬁc attributions for negative events at work. The reformulated
theory of learned helplessness is particularly useful because it may be applied to many
situations, reﬂects past experience, addresses motivation and mood, does not require
constant logical thought, and considers the environment. Studies have found that mood,
expectations, and reinforcements can alter attributions (Baumgardner & Arkin, 1988;

House, 1976; Weiner et al., 1971). However, limited work has examined if or how

developmental or familial factors relate to causal attributions in adults.

Attributional Style for Achievement and Intemrsonal Related Events: RelationS_h_ip&
_S_ex

The reformulated theory of learned helplessness (Abramson et al., 1978) proposed
that individuals who regularly attribute negative events to internal, stable and global
causes are vulnerable to depression. Much research using this theory assumes that
attributional style is consistent across different types of situations (e.g. achievement vs.
interpersonal). However, the theory itself did not address this point.

The theory of hopelessness depression (Abramson et al., 1989), a revision of the
reformulated theory of learned helplessness, suggested that attributional style may differ
across domains and therefore a person may exhibit a Speciﬁc vulnerability. Houston
(1995) was consistent with the notion of domain speciﬁc attributional diatheses. Using
Metalsky et al.’s (1987) attributional style questionnaire, Houston (1995) found that
attributional style for achievement events is more strongly related to emotional response
after failing an intellectual task than attributional style for interpersonal events is. Using
the Attributional Styles Questionnaire (Seligrnan, Abramson, Semmel, & von Baeyer,
1979), Berndt, Bemdt, and Kaiser (1982) assessed differences in attributions based on
content (achievement and afﬁliation) and sex. They found little support for differences
based on either dimension. However, the results of this study must be considered
tentative as the sample was small and the Attributional Styles Questionnaire has modest
reliability. Some researchers evaluating the reformulated theory of learned helplessness
also appear to have considered the concept of a speciﬁc attributional vulnerability. For
example, Peterson, Colvin, and Lin (1992) reported on two studies. One study used a
version of the Attributional Styles Questionnaire focused solely on academic events
when examining the relationship of attributional style to helplessness in students

grappling with academic challenges and disappointments.

10

Commonly used measures of attributional style (Peterson & Villanova, 1988;
Seligrnan et al., 1979; Metalsky et al., 1987) focus primarily on two domains,
achievement events and interpersonal events. For example, Peterson and Villanova’s
(1988) Expanded Attributional Style Questionnaire (which is used in this data set) asks
individuals to make attributions for 24 negative life events. Of these items, 7 are focused
on achievement events and 10 are focused on interpersonal events. Thus far, a review of
the literature has revealed no studies that used the Expanded Attributional Styles
Questionnaire (Peterson & Villanova, 1988) to examine the possibility of a domain
speciﬁc attributional vulnerability/diathesis.

The existence of societal processes and structures that steer men and women
toward different roles suggests that there will be sex differences in domain speciﬁc
attributional style. Citing the work of Robert V. Wells, Alice Clark, Talcott Parsons, Ruth
Bloch and others, Chodorow (1978) argued that the shift ﬁom an agrarian society to an
industrial economy in the US changed the division of labor to one in which women
became primarily responsible for childcare, emotional labor, and the household, and men
became primarily responsible for public matters and achieving in the paid labor force.
Chodorow asserted that this division of labor is related to the structure of the family, the
behavior of parents, and the psychological development of children. More speciﬁcally,
she argued that these factors give girls an experience of closeness and connection and
. boys an experience of distance (in relationship to the mother) that predisposes them to
engage in interpersonal or affective roles and extrafamilial and more impersonal
activities respectively.

Despite laws ensuring equal opportunity for women in employment, data suggest
that the division Chodorow (1978) described still exists even though the gap may be
narrowing. Data suggest that the rate of women’s participation in the paid labor force has
been increasing since 1960 and almost equals the rate for men in some age groups, but in

1999 the participation rate for men was still higher in every age group and in each marital

11

status category (U .S. Bureau of Labor Statistics Bulletins 2217 & 2340 and unpublished
data reported in the US. Census Bureau, 2000, Table 651, p. 408). For the civilian,
noninstitutional labor force (over age 16 years), 74.7% of men werein the labor force as
compared to 60% of women (U .S. Bureau of Labor Statistics Bulletin 2307 and
Employment and Earnings (January issues) cited in US. Census Bureau, 2000, Table
645, p. 404). In 1990, which is closer to the time of data collection, 76.4 % of men and
57.5% of women were in the labor force (U .8. Bureau of Labor Statistics Bulletin 2307
and Employment and Earnings (January issues) cited in US. Census Bureau, 2000, Table
645, p. 404).) Additionally, some data (U .8. Bureau of Labor Statistics, Employment and
Earnings (January issues), and unpublished data cited in the US. Census Bureau, 2000,
Table 669, p. 416-418) suggest that women are encouraged to pursue ﬁelds that are
related to caring, nurturing, and supporting others. In 1999, for the US. as a whole,
60.4% of those in service occupations were women, 9.0% of those in precision
production, crafts, or repair positions were women, 24.1% of operators, fabricators, and
laborers were women, 63.8% of technical, sales, and administrative support personnel
were women, and 49.5% of those in managerial and professional specialties were
women. The difference is even more obvious when considering more detailed data. For
example, in 1999, 10.6% of engineers were women, but 85.7% of health assessing and
treating professionals (e. g. nurses, dietitians, and speech therapists) were women.
Traditional sex role stereotypes, “. . .consensual beliefs about the differing
characteristics of men and women in our society...” (Rosenkrantz, Vogel, Bee,
Broverman, and Broverman, 1968, p. 287) are consistent with the aforementioned
division of labor. For example, Rosenkrantz et al. reported that at least 75% of the men
and 75% of the women in their sample (r_r = 154) agreed that men are competitive,
logical, conﬁdent, ambitious, skilled at business, make decisions easily, act as leaders,
and know the ways of the world. In contrast, at least 75% of the men and 75% of the

women agreed that women are tactful, gentle, aware of others’ feelings, need security,

12

express tender feelings, and are talkative. Similarly, based on a combination of interview
data (p = 20) and extant literature and research, Komarovsky (1950) argued that parents
treat boys and girls differently to prepare them for roles inside and outside the home
respectively. She asserted that males are encouraged to be competitive, independent, and
dominant whereas girls are relatively sheltered and encouraged to attend to kinship
responsibilities. Several studies also suggest that men and women share these beliefs
(Broverman, Broverman, Clarkson, Rosenkrantz, & Vogel, 1970; Kitay, 1940;
Femberger, 1948; Rosenkrantz et al., 1968).

Given our societal structure and the messages associated with sex roles, it seems
likely that males and females might develop attributional styles that help them to meet
societal roles and obligations. Thus, males may be more likely to develop an attributional
style which allows them to persist after experiencing negative work or achievement
related events and females may be more likely to develop an attributional style which
allows them to persist after experiencing negative relational/affective events.

Research on sex differences in attributional style has shown conﬂicting and
complex results. Using a sample comprised of adolescents, young adults, and middle-
aged persons, Johnston and Page (1991) found that men’s responses to the Attributional
Style Questionnaire (Abramson et al., 1978) were signiﬁcantly more internal, stable and
global than were women’s responses. Other studies have found that scores on the causal
dimensions do not differ by sex (Bemdt et al., 1982; Handal et al., 1987; Johnson, 1992;
Whitley et al., 1991). However, studies have also suggested that the relationship of
attributional style to depression differs for men and women (Handal et al., 1987;
Johnson, 1992). These latter studies point to the importance of examining the relationship
of attributional style to mood, and to other indicators of functioning, separately for men
and women even when there are no sex differences on the causal dimensions.

In summary, although much work has assumed that attributional style will be

consistent across content domains, the reformulated theory of learned helplessness is

13

ambiguous about this point. Measures used to assess attributional style focus primarily on
two domains —— achievement and interpersonal matters. Some studies have begun to
examine whether there are content speciﬁc attributional diatheses. Other studies have
begun to examine sex differences in attributional style. However, these studies have not
addressed the possibility that these content speciﬁc attributional diatheses may differ

based on sex.

Attributional Sgle and Perceptions of Parental Attributicpral Style
The reformulated theory of learned helplessness (Abramson et al., 1978) did not

 

clearly articulate how attributional style develops or how attributional style relates to
developmental factors. Researchers, however, have applied the construct of attributional
style to children and adolescents. For example, Dixon and Ahrens (1992) found that an
attributional style-stress interaction helps to account for depression in children.
Lewinsohn et al. (2000) found, in a study of adolescents, that a “positive attributional
style” was associated with remaining nondepressed aﬁer recovering ﬁom major
depression. Further, some theoretical work and research has begun to address the
acquisition of attributional style (Haines, Metalsky, Cardamone, & Joiner, 1999; Kaslow,
Rehm, Pollock, Siegel, 1988; Seligrnan et al., 1984). Haines et a1. (1999) argued that the
experiences that an individual has in childhood (e.g. feedback from signiﬁcant adults)
can contribute to the development of a pessimistic world view and be reﬂected in
attributional style. The research on the role of parents in the development of attributional
style is conﬂicting. Kaslow et a1. (1988) found no relationship between children’s
attributional style and. parental attributional style/cognitive distortions. Seligrnan et a1.
(1984), however, found that children’s attributional styles were related to those of their
mothers, but not their fathers.

While the aforementioned literature suggests that a parent’s behavior could affect

a child’s attributional style and thus contribute to the development of an adult’s

l4

attributional style, psychoanalytic theory (Erikson, 1963) pointed to the importance of
parents and the young adult’s perceptions of them for young adult development. The
tasks of young adulthood include becoming more independent of parents and selecting a
. role in the world (Erikson, 1963). Young adults' perceptions of their parents are related to
the development of autonomy, their psychological health, and their involvement in
interpersonal relationships (Frank, Pirsch, & Wright, 1991; Frank, Pirsch, Wright,
Jacobson, & Pedigo, 1990). It seems possible that perceptions of parents may also affect
young adults' causal attributions. Monestere and Thornton (1993, 1994) reported that
young adult attributional style is positively correlated with the combined perceptions of
maternal and paternal attributional style and with the perceptions of each parent’s
attributional style individually. The magnitude of the relationship of young adult
attributional style to perceived parental attributional style differs based on parental
marital status (Monestere & Thornton, 1994) and on identity status (Monestere &
Thornton, 1993).

A young adult's perceptions of parental attributional styles may serve as a model
of how parents explain events as well as a model of how to explain events in general.
People might refer to these percepts or schemas when they are unsure of how to explain
an event. And, when faced with a particular type of event (e.g. achievement,
interpersonal), young adults may consider their parents’ styles of explaining events
within the same domain/content area. Use of these percepts would be especially
important in young adulthood because young adults confront many new experiences,
beliefs, and choices. Given the frequency of these novel experiences, perceptions of
parental attributional style might greatly inﬂuence a young adult's attributional style,
coping mechanisms, and choice of roles and beliefs, with Signiﬁcant implications for
later life.

It is hypothesized that there will be sex differences in the relationship of young

adult attributional styles to perceived parental attributional style. It is argued that three

15

factors might affect the magnitude of the aforementioned relationship. First, the
availability of the model (mother/father) when the individual was growing up could
affect the relationship of young adult attributional style to perceived parental attributional
style. Bandura (1977) argued that the number and type of opportunities a person has to
observe a model will affect observational learning. Thus, the more opportunities the
individual had to observe the parent explain or address speciﬁc kinds of situations, the
more likely it is that the individual will learn about a model’s beliefs and ways of
handling various situations. Data from the sample used in this study suggest that the
young adults had more contact with and opportunities to learn from their mothers. More
speciﬁcally, 73% of the participants had married parents and given cultural norms it
seems likely that they spent more time with their mothers. Data from the participant
background questionnaire and other data from the US government are consistent with this
assertion. More speciﬁcally, for the sample as a whole, mothers did not start working
until the child was an average of 7.29 years old (S1; = 4.44). The data also indicate that
the most common time for mothers from every marital status to resume full-time
employment was when the child started elementary school. Additibnally, for the largest
group of participants (those with married parents), nearly 45% of the mothers were
employed only part-time or not at all (29.8% part-time, 14.1% never). Other than
occupation, no data were gathered about paternal employment, an omission suggesting
the strength of the assumption that fathers work full-time. Data ﬁom the US government
also suggest taht children are likely to spend more time with mothers than with fathers.
Government data suggest that nearly all married men are employed. In 1990, more than
96% of married men aged 25-44 years old were employed in contrast to 69-74% of
married women in the same age category (U .S. Bureau of Labor Statistics Bulletins 2217
& 2340 and unpublished data reported in the US. Census Bureau, 2000, Table 651, p.
408)

Second, what a parent symbolizes or represents (roles / beliefs / expectations) to

16

' the child might also affect magnitude of the relationship of young adult attributional style
to perceived parental attributional style. Parsons and Bales (1955) asserted that families
play a key role in shaping children into members of society. This book argued that
women and men ﬁll different roles in society and different functions within the family.
More speciﬁcally it argued that men perform instrumental functions in the world and at
home and that women function primarily at home and perform expressive functions
within the family. Thus, men assume agentic roles and focus on accomplishing goals.
Women, in contrast, primarily address the emotional and relational needs of the family.
Similarly, Chodorow (1978) argued that as the primary caretaker of children, mothers
give children experiences of closeness and nuturance whereas fathers demonstrate
competence in the world outside the family and give children an experience of
independence. Although societal changes have narrowed the differences between men’s
and women’s societal roles, labor statistics and the sex role stereotypes noted earlier
suggest that these traditional roles could affect the relationship of young adult
attributional style to perceived parental attributional style. Thus it seems likely that
young adults may perceive their mothers as representing nurturance and the social realm
and be more likely to consider how their mothers would explain interpersonal events.
Repeated consideration could lead young adults to explain interpersonal events in a
manner similar to their mothers’ styles. It also seems likely that fathers may represent
agency and achievement and that young adults may be more apt to consider how a father
might explain achievement events and thereby develop a similar attributional style for
achievement events.

The third factor posited to affect the magnitude of the relationship of young adult
attributional style to perceived parental attributional style is sameness. Kagan (1958,
1971) argued that children want to be competent and powerful like their parents and
come to identify with them, perceiving themselves as especially similar to the same sex

parent. He said that children identify with a model when they believe that they share

17

physical or psychological qualities with the model and when they vicariously experience
feelings that are appropriate to the model’s experiences. Kagan asserted that children
desire the positive features of the model as well as the beneﬁts resulting from them. As a
result, he thought that children take on the behaviors and the attitudes of the model.
Kagan asserted that while children can perceive similarities on their own, others also
highlight similarities for them. He said that children are reinforced by others for imitating
the parent and to some extent are also reinforced by their own behavior.

Mischel ( 1970) also extensively reviewed the research pertaining to sex typing
and socialization, concluding that social learning processes and reinforcement play an
important role in the development of sex role identity. He noted that parents are
important models of sex role behavior and of behavior in general. His review also
suggested that individuals pay more attention to and recall more about models of the
same sex.

In summary, the reformulated theory of learned helplessness (Abramson et al.,
1978) does not specify how attributional style is developed. Researchers have begun to
examine attributional style in children and work is being done to ascertain what factors,
e.g. parents, might play a role in the development of an individual’s attributional style. It
is possible that young adult attributional style is related to young adults’ perceptions of
their parents’ explanatory styles. More speciﬁcally, it is argued that young adults may
consider how parents explain events in general (as well as particular types of events)
incorporate this information into their own attributional style for similar events. Young
adults’ perceptions of parental attributional style may also inﬂuence their causal
attributions for particular failure events. It is proposed that three factors (availability,
symbolism, sameness) will affect the magnitude of the relationship of young adult

attributional style to perceived parental attributional style.

18

Identig Status
Erikson (1963) asserted that adolescents and young adults must deﬁne the role

they want to assume in the adult world they are entering; he described this as achieving
an identity. In selecting a role, young adults must decide what they believe and like in
ideological (political, religious, work) and interpersonal (friendship, love) realms. As part
of this process, ErikSOn thought that young adults integrate perceptions of parents and
parental beliefs and perceptions of themselves and their own beliefs, with societal
mesSages and roles.

Marcia (1966) proposed that young adults' resolutions of this process vary along
two orthogonal dimensions: exploration/crisis and commitment. The exploration/crisis
dimension assesses how actively young adults explore or select roles and beliefs. Young
adults who actively engage in this process are described as being in crisis. The
commitment dimension refers to how ﬁrmly or permanently young adults hold their
beliefs and their adult roles. Young adults who hold their beliefs strongly and have
deﬁned a role are said to have made a commitment. Using these dimensions, he
described four resolutions to the identity versus identity confusion stage. These four
identity statuses are identity achieved, moratorium, foreclosed, and diffuse. Individuals
who have actively explored roles and beliefs and made a commitment to them are termed
identity achieved. Individuals in moratorium actively explore beliefs and roles but have
not yet made a commitment to them. Adults who have not explored their beliefs and role
options, but who have made a commitment are in the foreclosed group. (These adults
typically adopt their parents' beliefs and enter roles prescribed by their families.) Adults
who have not explored beliefs and role options and who have not made a comrrritrnent to
a set of beliefs or a role are termed diffuse.

There has been debate in the literature about how well Marcia’s (1966) identity
status construct captures Erikson’s thoughts about identity development (Coté & Levine,

1988a, 1988b; Waterman, 1988). Research has also examined whether and how people

19

move from one identity status to another and the stability of the identity statuses (Adams
& Fitch, 1982; Waterman, 1982; Waterman, Geary, & Waterman, 1974). Several studies
indicate that in general, the percentage of individuals in the identity achievement
category increases and the percentage of individuals in the diffusion status decreases with
age, as one might expect if there is a progression from diffusion to achievement (Adams
& Fitch, 1982; Waterman, 1982; Waterman et al., 1974). However, when movement is
studied at the level of the individual, there is more variation. This work has prompted
researchers and theorists to consider whether the identity statuses might best be construed
as a typology rather than stages showing movement toward identity achievement. Despite
questions about how to regard identity status, it seems to be a reasonable mechanism for
determining whether individuals have explored beliefs and riles and made reﬂective
identity commitments (Waterman, 1988).

Although the early research addressing identity status emphasizes development in
the ideological (politics, religion) realm based on all male samples, later research
assesses identity development in both the ideological and the interpersonal (sex,
ﬁiendship) realms and includes both men and women in the samples. The research using
identity status to compare identity development in men and women has yielded
conﬂicting results. Assessing identity using a measure focused on ideology, Adams and
Fitch (1982) found that men were more likely to be in the identity achieved and
moratorium categories than women. In contrast, using a measure of identity status that
encompasses both the ideological and interpersonal domains, Grotevant and Adams
(1984) found evidence that women were more likely to be identity achieved than men in
the ideological and interpersonal domains and on the total identity scale. Although the
differences are statistically signiﬁcant, men’s and women’s scores differed by only a
small amount. In a third study, Bilsker, Schiedel, and Marcia (1988) found no sex
differences in overall identity status or within the domains assessed. They reported

comparable numbers of men and women in each identity status category. Reviewing

20

multiple studies, Waterman (1999) found evidence to suggest that women were more
likely than men to be in the identity achieved and moratorium categories for the
interpersonal domains, but not for work related domains.

Some work has also explored the relative contributions of different domains to
identity development in women and men. According to Bilsker et al. (1988), identity
status in the ideological realm was the best predictor of men’s overall identity
development. In contrast, identity status for the interpersonal realm was the best predictor
of women’s overall identity development. This study reported no sex differences with
respect to occupation. Women participants rated interpersonal identity development as
more important to their identity than men did.

If an individual explores, he or she must actively analyze beliefs and choices. The
person must compare parental, personal, and social beliefs and guidelines. Engaging in
this process may have several consequences - increased analytical skills, a more well
deﬁned sense of one's own beliefs as well as parental and societal beliefs, and a greater
awareness of one's role in the world and what one wants. In addition, the skills associated
with active exploration may allow individuals to make more accurate attributions and to
avoid inappropriately assigning blame to themselves. Thus, as the level of failure
increases, identity achieved and moratorium individuals will be better able to recognize
the underlying causes and make accurate attributions than foreclosed and diffusion status
individuals, who may be more apt to make attributional errors with increasing frequency.

Exploration also suggests that people will develop beliefs or skills that will allow
them to continue exploring even when confronting difficult or challenging situations. An
explanatory style that decreases the likelihood of an individual experiencing cognitive,
motivational, or affective deﬁcits is an asset. Thus, it seems likely that many individuals
who explore have external, unstable, and speciﬁc attributional styles for uncontrollable,
negative events. A number of studies are consistent with the idea that individuals who

explore develop attributes, beliefs, and skills that facilitate exploration and resilience.

21

According to Grotevant and Adams (1984), individuals in the achievement category were
inclined to confront and to engage problems while individuals in the foreclosed and
diffuse groups use less effective coping strategies. Marcia (1966) reported that
individuals in the identity achieved group perform better on a stressful concept
attainment task. Additionally, Marcia (1967) found that persons in the identity achieved
and moratorium groups are less vulnerable to self-esteem manipulation when receiving
false feedback (positive or negative) on a concept attainment task that they were told was
indicative of their intelligence and academic success. Although Marcia’s (1966, 1967)
studies had relatively small sample sizes (26 and 72 respectively) they suggest that
individuals in the achievement and moratorium statuses frmction better in stressful
problem solving Situations and may be less vulnerable to negative emotional
consequences in response to positive or negative feedback about their performance.

In summary, Erikson (1963) argued that a major task of adolescence and young
adulthood is the formation of a cohesive sense of identity. Marcia (1966) developed the
construct of identity status to assess young adults’ resolutions to Erikson’s identity versus
role confusion stage. Researchers have used this construct to assess development in men
and women and the results have been conﬂicting. It is also proposed that the analytical
skills associated with high identity status enable individuals to make more accurate
causal attributions about failure. F urtherrnore, it is argued that exploration is associated

with an optimistic attributional style.

Attributional Style, Perceptions of Parental Attributional Sgle, Identiﬂ Status and
Resppnse to Failure

Although it is often treated as being consistent across content areas, it is proposed
that a person’s attributional style will differ for interpersonal and achievement events. It
is hypothesized that the differences will be consistent with the norms of mainstream US

culture with respect to the gender appropriateness and importance of these activities.

22

More speciﬁcally, it is proposed that young adults will develOp attributional styles that
will allow them to persist in adhering to the normative teachings of the dominant culture.
For example, mainstream US culture prioritizes success in the work world (over success
in interpersonal relationships) for males (Parsons & Bales, 1955). Thus, males would
beneﬁt from an attributional style (an external, unstable, and speciﬁc style) that allows
them to persist after achievement or work failures and it is hypothesized that men’s
attributional styles will be more external, unstable, and speciﬁc for negative achievement
or work events than for interpersonal ones. Conversely, given the importance that society
places on women’s interpersonal relationships (Parsons & Bales, 1955), it is
hypothesized that women will exhibit a more external, unstable, and speciﬁc attributional
style for uncontrollable negative interpersonal events than for uncontrollable
achievement or work events.

It is also proposed that young adults’ perceptions of their parents’ attributional
styles for achievement and interpersonal events will be positively related to their own
attributional styles for the same domains. It is argued that when young adults explain an
event they will consider the way their parents might explain the event and, through
repetition, develop attributional styles similar to their parents’. It is also argued that the
magnitude of these relationships will differ based on availability of the model (Bandura,
1977), what the parent represents (Chodorow, 1978; Parsons & Bales, 1955), and the
sameness of the young adult and the model (Kagan, 1958, 1971; Michel, 1970).

It is hypothesized that attributional style (Abramson et al., 1978) is related to
identity status (Marcia, 1966). More speciﬁcally, it is hypothesized that attributional style
begins to develop during childhood and can therefore affect a person’s approach to the
identity versus identity confusion stage (Erikson, 1963). It is proposed that children who
routinely attribute negative events to external, unstable, and speciﬁc causes will
experience fewer cognitive, emotional, and motivational deﬁcits (Abramson et al. 1978)

and, as young adults, will be better able to explore roles and beliefs as they are less likely

23

to become depressed after encountering setbacks. Thus, it is hypothesized that an
external, unstable, and speciﬁc attributional style for negative events will be associated
with the identity statuses that require exploration (achieved and moratorium).
Conversely, it is proposed that an internal, stable, and global explanatory style for
negative events will be associated with identity statuses not marked by exploration.

It is hypothesized that the magnitude of the relationship between attributional
style and identity status will differ based on domain. Given similarity of content, when
young adults face choices about beliefs and roles in the ideological domain, it is argued
that achievement attributional style will play a greater role than interpersonal
attributional style in determining whether and how much they explore. Thus it is
hypothesized that achievement attributional style will be more closely related to
ideological identity status than to interpersonal identity status. Similarly, given
overlapping content, when young adults explore beliefs and roles in the interpersonal
domain, it is hypothesized that interpersonal attributional style will play a greater role
than achievement attributional style in determining if and how much they explore.
Therefore, it is hypothesized that interpersonal attributional style will be more closely
related to interpersonal identity status than to ideological identity status.

It is asserted that attributional style, perceptions of parental attributional style, and
identity status will be related to young adults’ responses to failure and that there will be
sex differences. It is hypothesized that each of the aforementioned variables will be
directly related to attributions for failure. It is also hypothesized that identity status will
moderate the relationship of attributional style and perceived parental attributional style
to young adults’ attributions for and mood after failure. Compared to attributional style
and identity status scores for the interpersonal domain, it is also proposed that
attributional style for achievement events and identity status scores for the ideological
domain will be more strongly associated with attributions for failure and subsequent

mood.

24

Chapter Three: Methods
Method and Desigp

The ﬁrst phase of this research was a correlational study. Questionnaires were
used to examine the hypothesized relationships among perceptions of parental
attributional styles and a young adult's attributional style and identity status. This
research set the stage to evaluate the impact of perceived parental attributional style and
young adult attributional style on young adults' attributions after a failure experience.
The second phase of this research was an experiment examining the relationship of
identity status to attributions about varying degrees of failure. Participants were assigned
to a high, intermediate, or low failure condition; the level of failure was manipulated by
varying task difficulty. An attempt was made to have equal numbers of men and women
in each failure condition and to have equal numbers of individuals from the identity
statuses that are high and low in exploration in each failure condition. (Given the limited
sample size, however, the failure levels were collapsed; in other words, the high,

intermediate, and low failure groups were combined in analyses.)

Past Work

Two sets of analyses have been done using this data set. The ﬁrst examined a)
whether identity status moderates the relationship of young adult attributional style
(YAS) to perceived parental attributional style for mothers and fathers combined and b)
the relationship of identity status and level of failure to attributional style. The second
study examined a) whether parental marital status moderated the relationship of YAS to
perceived parental attributional style for fathers and mothers separately and b) the
relationship of YAS and perceived parental attributional Style to attributions for failure.
These studies differ from the present one in the following ways. First, the current study is
more speciﬁc in its examination of the constructs. Previous work addressed overall

attributional style and overall identity status as opposed to content-speciﬁc domains.

25

Second, the present study emphasizes the role of sex differences in understanding the
relationship of attributional style to perceived parental attributional styles and to identity
status. Third, identity status takes a different place in the current model consistent with
the idea that attributional style accounts for some of the variation in identity status.
Fourth, previous work has not addressed the emotional response to failure and, ﬁnally,

the analytic approach is different from that used in the past.

Participants

Seven hundred seventy-three individuals (494 females, 278 males, 1 undeclared)
participated in Phase I of the study and serve as the sample for the present research. On
average, participants were 19.45 years old (S_D = 1.22 year) and had attended Michigan
State University for 2.01 years (p = 771, S_D = 1.03 year). The socioeconomic status
scores (SES) for fathers (M = 6.5, S_D = 2.14) and mothers (M = 4.89, SD = 3.02)
indicate that most participants were from middle class backgrounds. (On Hollingshead's
(1957) scale, SES is rated from 1 to 9 where 9 is the highest SES.) Most fathers were
technicians, semi or minor professionals, managers, teachers, or owners of small to
medium sized businesses and farms. Mothers typically were in clerical or sales positions
or were skilled laborers, craftspersons, or owners of small businesses and farms. The
mean level of education was 3.62 (S_D = 1.29) for fathers and 3.29 (SD = 1.15) for
mothers indicating that fathers and mothers typically had some college education.
Women and men differed on the following variables: age (1 (768) = 5.90, p < .001; mean
for men = 19.79 year, mean for women = 19.26 year), father's SES (1 (701) = 2.27, p <
.024; mean for men = 6.75, mean for women = 6.37), father's education (t (767) = 2.41, p
< .016; mean for men = 3.77, mean for women = 3.54 for women). Women and men did
not differ with respect to mother's SES or mother's education.

Individuals who completed Phase I and then participated in Phase H (27 males, 46
females) did not differ signiﬁcantly from individuals who completed only Phase I with

26

respect to age, class, mother's education, mother's SES, or father's education. The
composition of the Phase I and the Phase H samples did not differ with respect to gender
(x2 =.03020, _d_f = 1, p < .86204). The mean SES score for fathers of Phase H participants
(6.96) was signiﬁcantly greater than that of fathers of Phase I participants (6.45)

(t (91.82) = -2.24, p<.03).

The high exploration identity status groups (achievement, moratorium) and low
exploration identity status groups (diffusion, foreclosure) participating in Phase H did not
differ in terms of age (L (24.19) = -.71, p < .486), class (3 (52) = .26, p < .792), father's
SES (t (48) = -.27, p < .79), father's education (1 (52) = .22, p < .828), mother's SES (t
(49) = .20, p < .842), or mother's education (1 (52) = 1.79, p < .079).

Procedures

Session I

Participants completed a packet of questionnaires during a 2 1/2 hour group testing
session. At the beginning of this session, subjects were given an informed consent sheet
for session I, an informed consent sheet describing session H, and a recontact form. They
were instructed to complete the recontact form only if they wished to participate in
session H. They were also told that submitting the session H informed consent sheet and
recontact form did not commit them to participating or guarantee that they would be
asked to participate. The questionnaire packet consisted of the Personal and Family
Background Questionnaire, the Expanded Attributional Styles Questionnaire (Peterson &
Villanova, 1988), the Beck Depression Inventory (Beck, Ward, Mendelson, Mock, &
Erbaugh, 1961), the Perceived Parental Attributional Style Questionnaire - father form
(Monestere, 1993), the Extended Objective Measure of Ego Identity Status (EOMEIS)
(Grotevant & Adams, 1984), the Perceived Parental Attributional Style Questionnaire -
mother form (Monestere, 1993), and the Life Experiences Survey (Sarason, Johnson, &

Siegel, 1978). Half of the packets had the questionnaires in this sequence and half had

27

them in the reverse order.

After the initial session, a researcher scored the BDI, the LES, and the EOMEIS
for all persons volunteering for session H. To protect the well being of participants,
individuals with scores of 10 or above on the BDI or scores above 13 on the negative
change scale of the LES were eliminated. Identity status was also coded to ensure that
equal numbers of low and high exploration individuals were assigned to each of the three
failure conditions. An effort was made to assign equal numbers of women and men to

each failure condition.

Session 11

Research assistants contacted eligible volunteers by phone to schedule session H
testing appointments. Scheduling was to continue until 75 men and 75 women completed
session H. This goal was not reached because of time constraints. The intermediate
failure condition was also eliminated to reduce the time needed to conduct the study and
to ensure adequate numbers of participants in the other failure conditions. The research
assistants conducting session H did not know participants’ names Or the contents of their
session I questionnaires. During session H, subjects performed an anagram solving task
(form a, b, or c) and immediately responded to a visual analog scale measuring satisﬁed
mood. They then received accurate failure feedback about their performance, waited 5
minutes, completed three questionnaires (the ED], the PAQ, and the Performance
Evaluation Measure (PEM)). The PEM was done ﬁrst and the BDI and PAQ were
counterbalanced. After completing the questionnaires, participants were debriefed. The
Depression Adjective Check List (Lubin, 1965) was administered after the debrieﬁng.
Subjects were recontacted by phone approximately 8 hours after the session to address
any remaining questions or concerns. Two participants were scheduled at the same time
whenever possible, but tested separately. This arrangement allowed the examiner to

justify the brief 5-minute waiting period after the experimental task without using

28

deception.
Measures
Session I
The Personal and Family Backggound Questionnaire requests demographic
information including the participant's age, race, occupation, education, religion, family
structure, and current living situation as well as parents' occupations and marital status.
The Expanded Attributional Sgyles Questionnaire (EASQ) (Peterson & Villanova,

1988) assesses attributional style and consists of 24 hypothetical negative events.
Respondents are asked to imagine that the event is happening to them, write down one
major cause, make causal attributions, and rate how much control they would have in that
situation. Respondents make the causal attributions using a seven-point scale for each
dimension: intemality (7) vs. extemality (1), stability (7) vs. instability (1), and globality
(7) vs. speciﬁcity (1). To score each attributional dimension, the item ratings for the
dimension are summed and divided by the number of items (Peterson & Villanova,
1988). A generality score may also be computed by summing the item ratings for the
globality and stability dimensions and dividing by the number of items in both scales
(Metalsky et al., 1987). Attributional style may also be operationalized as the average of
the scores for the three attributional dimensions (F ollette & Jacobson, 1987). The three
attributional dimensions have adequate internal consistency; coefﬁcient alphas were .66
for intemality, .85 for stability, and .88 for globality (Peterson and Villanova, 1988). The
EASQ also demonstrated good predictive validity. In a study of 140 college students,
Peterson and Villanova (1988) found that scores on the explanatory style dimensions
were signiﬁcantly and positively correlated with attributions for actual bad events. The
intemality and globality dimensions were the best predictors of intemality and globality
ratings for actual events (r = .32, p < .001 and g = .36, p < .001 respectively). However,
scores on the stability and globality explanatory dimensions predict stability ratings for
actual events equally well (r = .18 and _r_ = .18, p <.05 for both).

29

The Perceived Parental Attributional Sgle Questionnaire (PPASQ) (Monestere,
1993) is a modiﬁcation of the EASQ (Peterson & Villanova, 1988) and assesses

respondents’ perceptions of their parents' explanatory styles. The PPASQ consists of 24
hypothetical negative events. Young adults are asked to imagine that their parents are
experiencing each event and write down one major cause. Then they are asked what they
think their parents’ attributions would be and how much control their parents would
experience in each situation. (Participants complete one from the mother’s perspective
and one from the father's perspective.) The measure is scored in the same manner as the
EASQ.

The Beck Depression Inventogy (Beck et al., 1961) (BDI) is a 21 item self-report
measure that evaluates the presence and severity of depression by assessing depressive
symptomatology, such as loss of appetite, sleep disturbances, feelings of sadness, guilt,
pessimism, failure, dissatisfaction, or self-hatred. Each symptom is rated on a four point
scale from 0 (not at all) to 3 (severe). The sum of the item scores is the total score. Total
scores above 10 indicate depression. The BDI has demonstrated solid internal
consistency in psychiatric and nonpsychiatric populations. In a meta-analysis, Beck,
Steer, & Garbin (1987) reports a mean coefficient alpha for 9 psychiatric samples of .86
and a mean coefficient alpha for 15 nonpsychiatric samples of .81 . Test-retest reliability
(over different intervals) varies between .48 and .86 for psychiatric samples and between
.60 and .90 for nonpsychiatric samples (Beck et al., 1987). Comparisons with DSM-HI
criteria indicate sufﬁcient content validity (Beck & Steer, 1987). Although it is not
intended as a diagnostic tool, the BDI does differentiate certain depressive disorders
(Steer, Beck, Brown, Berchick, 1987). Beck et al. (1987) also ﬁnd evidence of construct
validity; BDI scores correlate with hopelessness. In their meta-analysis, Beck et a1.
(1987) ﬁnd support for concurrent validity; on average BDI scores correlate .72 with
clinical ratings of depression for psychiatric patients and .60 with ratings of

nonpsychiatric samples.

30

The Life Exmriences survey (Sarason et al., 1978) (LES) is a 60—item self-report
measure examining life changes that have occurred over a one year period. In addition to
reporting which events they have experienced and when (0 to 6 months ago or 7 to 12
months ago), respondents indicate if they viewed events positively or negatively and the
impact of the events. Respondents rate the impact on a scale of -3 (extremely negative) to
+3 (extremely positive). Three scores can be derived. The negative change score is the
sum of the impact ratings for all negative events. The positive change score is the sum of
the impact ratings for all positive events. The total change score is the sum of the positive
and negative change scores. The LES has moderate test-retest reliability. In two samples
(I; = 34, p = 58), Sarason et al. (1978) reports the following signiﬁcant (p < .001 for all
cases) test-retest correlations over 5- and 6-week intervals: for positive change scores .19
and .53, for negative change scores .56 and .88, and for total change scores .63 and .64.
Studies have also found evidence of construct validity. For example, the negative and
total change scores are signiﬁcantly correlated with anxiety, personal maladjustment, and
depression (Sarason et al., 1978).

The Expepded Obiective Measure of Ego Identig Status (EOMEIS) (Grotevant &
Adams, 1984) is a standardized measure of adolescent identity development. The
measure's 64-items assess whether or not a young adult has clearly deﬁned beliefs in four
interpersonal domains (friendship, dating, sex roles, recreation) and four ideological
domains (politics, religion, occupation, and philosophical life style) and whether or not
these beliefs were achieved through searching and questioning (as opposed to being
adopted unquestioningly from one's parents). Each of the eight domains contains eight
items. Of the eight items in each domain, there are two items that characterize each of the
four identity statuses. Responses to each item are made on a 6-point Likert scale where
A = strongly agree and F = strongly disagree. There are three conceptually based sets of
scales in the EOMEIS - the ideological, the interpersonal, and the total identity scales.

For each identity status, ideological scale scores are obtained by summing the scores of

31

the items that characterize a given identity status across the four ideological domains
(e.g. sum the scores of the two political, two occupational, two religious, and the two life
style items pertaining to moratorium). Interpersonal scale scores are derived in the same
manner. The total identity scale scores are obtained by summing the ideological and
interpersonal scale scores for each identity status. Using two samples (p = 317, p = 274),
Grotevant and Adams (1984) found that the EOMEIS had adequate internal consistency
(coefficient alpha ranged between .37 and .77 for the ideological and interpersonal
subscales and .42 to .84 for the total identity scale) and test-retest reliability over a four
week interval (.59 to .82 for ideological and interpersonal scales and .63 to .83 for the
total identity scale). To establish content validity, 10 graduate student raters familiar with
Marcia's identity statuses assigned the items to a status. Mean interrater agreement across
all raters on all items was 96.5%. With respect to factorial validity, three separate factor
analyses indicated that the EOMEIS makes consistent distinctions among the identity
status groups and has two distinct domains (interpersonal & ideological). The EOMEIS
also appears to have discriminant validity; correlations with constructs that might account
for variability in identity status (e. g. academic achievement, vocabulary, social
desirability) were small and accounted for only 6.25% of the variance in identity status.

The classiﬁcation rules outlined in Adams, Shea, & Fitch (1979) were used to
determine participants' identity statuses at the end of the data collection. To determine
overall identity status classiﬁcation, ﬁrst the sample mean for each of the four identity
statuses is computed (e.g. sum of the ideological and interpersonal scale scores for
moratorium divided by 16 (the number of items in the scales)). The cutoff for each
identity status scale equals one standard deviation above the mean. After establishing the
cutoffs for the sample, Adams et al. (1979) recommended the following strategy for
classifying individuals. 1) Classify persons to a particular identity status if a) their score
on that identity status scale is above the cutoff and b) their scores on the three other

identity status scales are below their respective cutoffs. 2) Classify a person as

32

moratorium if all of their identity status scale scores are below the cutoffs. 3) Classify an
individual as being in transition if more than one scale score was above the cutoff. Thus
an individual who scores above the cutoffs for diffusion and foreclosure would be called
diffusion-foreclosure. Individuals with three or more scores above the respective cutoffs
were considered unclassiﬁable. Grotevant and Adams (1984) used these rules to
determine an individual’s ideological and interpersonal identity status classiﬁcations as
well. During data collection for the current study, temporary identity status assignments
were made using scale means and cutoffs from another sample of college students
(Grotevant & Adams, 1984). This was necessary because the sample means (needed to
determine identity status) were not available when individuals were being assigned to the
experimental conditions. When data collection was complete, means and cutoffs were
determined for the current sample.

This classiﬁcation system was used in analyses that required ordinal variables.
The algorithm outlined on page 43 was developed in order to treat identity status as a
continuous variable in the proposed models. This approach of working with identity
status in both ways reﬂected the discussion about the nature of the identity status

classiﬁcations.

Session II

For the Anams Task, three sets (labeled Forms A, B, and C) of 30 anagrams
were used. They differed in the numbers of difficult and easy anagrams. The most
difficult set included 6 easy and 24 difﬁcult anagrams. The intermediate set included 11
easy anagrams and 19 difﬁcult anagrams. The easiest set included 15 easy and 15
difﬁcult anagrams. All of the easy anagrams were used in previous research (Feather &
Simon, 1971; Zuckerman and Allison, 1976). The difficult anagrams were based on
uncommon words that have been used in puzzles (Hunter, 1986). Every anagram could

be organized to spell an English word. The anagrams were sequenced so that a greater

33

proportion of difﬁcult ones were located toward the end of the task. Each anagram
consisted of 6 letters and was presented on a 3" x 5" ﬁling card. Participants recorded
their responses on an answer sheet. Participants had 30 seconds to work on each anagram
initially, but had only 20 seconds per anagram by the last item. The participant was
seated at a desk with a cassette player. When played, the audiocassette beeped at
speciﬁed intervals to signal that the participant should begin the next anagram whether or
not she or he had solved the current one. Participants received accurate feedback about
their performance after completing this task.

The Visual Analog Scale is a one item, 7-point Likert scale that assesses how
satisﬁed or down the subject feels after completing the anagrams task. It consists of
seven equally spaced plus signs running in a horizontal line across an otherwise blank 8”
x 11” sheet of paper. The endpoints were labeled ‘very satisﬁed mood’ and ‘very down
mood’.

The Particular Attribution Questionnaire (PAQ) is a 12 item measure exploring
the causal attributions that young adults have made about their performance on the
anagrams task. It examines the same attributional dimensions explored by the ASQ as
well as beliefs about effort, luck, exam difficulty, and ability. (It is a modiﬁed version of
a measure used in Metalsky et al., 1987. This was entitled the Test Performance
Questionnaire.)

The Performance Evaluation Measure consists of two items and evaluates the
effect of the experimental intervention. The ﬁrst item asks the subject to rate how well
they performed on a scale of 1 (very badly) to 10 (very well). The second item asks the
subject to rate on a scale of 1 (very happy) to 10 (very unhappy) how satisﬁed they were
with their performance.

The Depression Adjective Checklist (DACL) (Lubin 1965, 1967) assesses the

 

individual's current level of depressed affect. Form A consists of 32 nonoverlapping

adjectives. Scores above 15 indicate a signiﬁcant level of depression (Levitt & Lubin,

34

1975). The DACL (form A) has good internal consistency; the Split-half reliability
coefficient is .82. The measure also has strong discriminative validity; F -tests revealed
signiﬁcant differences (p < .0005) among normal controls, nondepressed patients, and
depressed patients. Patient group means were higher than the control mean and depressed

patients had the highest group mean (Lubin, 1965).

Hymtheses

Based on the reviewed theories, this project tested hypotheses in four sections.
Section One tested hypotheses that speciﬁed differences between men and women with
respect to identity status and attributional style. Section Two proposed hypotheses related
to the correlational structure of variables collected speciﬁcally between young adult
attributional style and ideological and interpersonal identity statuses. Section Three
hypothesized relationships between perceived parental attributional style for both
interpersonal and achievement events, young adult attributional style for achievement
and interpersonal events, and identity status for both the ideological and interpersonal
domains. Section Four speciﬁed hypotheses regarding the relationships between
attributional style, failure attribution, mood, and depression following an anagrams task
that was experimentally manipulated to control the level of failure. Each set of
hypotheses is followed by a description of the statistical methods that were used to test

each hypothesis.

Section One Hymtheses
Identig status.

1.1 In the interpersonal domain, there will be signiﬁcantly larger numbers of women (as
opposed to men) in the identity achieved and the moratorium statuses.

1.2 In the ideological domain, there will be signiﬁcantly larger numbers of men (as
opposed to women) in the identity achieved and moratorium groups.

Attributional sgle.

1.3 Women’s attributional style scores for interpersonal events will be more signiﬁcantly
external, unstable, and speciﬁc (lower) than their attributional style scores for negative

35

achievement events.
1.4 Men’s attributional style scores for achievement events will be signiﬁcantly more
external, unstable, and speciﬁc (lower) than their attributional style scores for negative
interpersonal events.

1.5 Women will have a signiﬁcantly more external, unstable, and speciﬁc (lower)
attributional style than men for negative interpersonal events.

1.6 Men will have a signiﬁcantly more external, unstable, and speciﬁc (lower)
attributional style than females for negative achievement events.

Section One Analyses
The planned analyses for sedtion one are described below. For hypotheses 1.1 and

1.2, identity statuses will be determined for each individual based on a set of
predetermined cutoff scores. Procedures are outlined in Adams et al. (1979). These
scores will be converted to a categorical scale: 1 = identity diffusion; 2 = identity
foreclosure; 3 = identity moratorium; and 4 = identity achievement. Chi-square analysis
will test categories by sex to determine if the number of women and men vary in each
category.

Multivariate Analyses of Variance (Stevens, 1992) will be used to test hypotheses
1.3 through 1.6. MANOVA runs univariate analyses of variance with a type I error rate
equal to one statistical test. It also allows the pooled variance of all dependent variables
to be evaluated as one construct. Two-tailed tests will be used. Both attributional styles
(interpersonal and achievement) contain three scales each (internal-external, global-
speciﬁc, stable-unstable). These are expected to vary by sex. Thus, two MANOVAS will
be run for each attributional style. The three attributional style dimensions will be the
dependent variables and sex will be the independent variable, thus comprising a one-way
MANOVA with three dependent variables. Differences on the scales between women
and men in the hypothesized direction will provide evidence for the hypotheses.
Depression, as measured by the Beck Depression Inventory (Beck et al., 1961), will be
used as a covariate for the analyses explicated above. Per Cook and Campbell (1979),

MAN OVAs will also be conducted to test the interaction of the independent variable and

36

the covariate. This step will ensure that the assumptions underlying the MANOVA
analysis are not being violeted and will facilitate interpretation. Analyses will be carried
out using SAS (SAS, Inc., 2000)

Section Two Hypptheses
2.1 The mean attributional style scores for individuals in the achievement and
moratorium statuses will be signiﬁcantly more external, unstable, and speciﬁc (lower)

than the mean scores for persons in the diffusion or foreclosure statuses for both the
ideological and interpersonal realms.

2.2 Mean attributional style scores for negative achievement events will be signiﬁcantly
lower external, unstable, and speciﬁc) for the achieved and moratorium identity statuses
in the ideological domain than for the diffuse and foreclosed groups in the ideological
domain.

2.3 Mean attributional style scores for negative interpersonal events will be signiﬁcantly
lower external, unstable, and speciﬁc) for the achieved and moratorium identity statuses
in the interpersonal domain than for the diffuse and foreclosed groups in the interpersonal
domain.

2.4 The correlation between achievement attributional style scores and identity status for
the ideological domain will be signiﬁcantly greater than the correlation between
achievement attributional style scores and identity status for the interpersonal domain.
2.5 The correlation between interpersonal attributional style and identity status for the

interpersonal domain will be signiﬁcantly greater than the correlation between
interpersonal attributional style and identity status for the achievement domain.

Section Two Analyses
The following describes the analytic plan for the section two hypotheses.

Multivariate Analyses of Variance will be used to test hypotheses 2.1, 2.2, and 2.3.
Identity will be converted to a categorical scale: 1=identity diffusion; 2 = identity
foreclosure; 3 = identity moratorium; and 4 = identity achievement. This is an ordinal
scale. The three attributional style subscales (internal-external, global-speciﬁc, stable-
unstable) will be modeled as dependent variables. Contrasts will test the hypotheses that
the identity scores 3 and 4 will have lower means in the three attributional subscales than
identity levels 1 and 2. Depression, as measured using the BDI (Beck et al., 1961), will

serve as a covariate for the analyses explicated above.

37

A latentvariable model (Loehlin, 1992; Marcoulides & Schumacker, 2001) will
be estimated to evaluate hypotheses 2.4 and 2.5. (See Figure 1.) Each subscale of the
EASQ Young Adult Attributional Style for Achievement (internal-external, global-
speciﬁc, stable-unstable) will serve as an indicator for a latent variable called
Achievement Attributional Style. The same structure will be applied to create a latent
variable called Interpersonal Attributional Style. The scales of the EOMEIS will be
reformulated to a continuous scale with low values indicating low levels of identity
exploration (i.e. diffusion) and high scores indicating high levels of identity exploration
(i.e. achievement). Eight items on each scale of the EOMEIS (achievement and
ideological) indicate one of four dimensions of identity status (diffusion, foreclosure,
moratorium, and achievement). Scored on a 0 to 5 scale, a weighted sum of the items will
be used to order the categories by degree of identity exploration. The algorithm is: X =
1(Sum (diffusion items)/40) + 2(sum (foreclosure items)/40) + 3 (sum (moratorium
items)/40) + 4 (sum (achievement items)/40). This will create a scale which has possible
values from 0 to 10. Extremely low scores will indicate diffusion and high scores will
indicate achievement. A manifest variable called Interpersonal Identity Status will use
items from the four interpersonal domains of the EOMEIS. A manifest variable called
Ideological Identity Status will use items from the four ideological domains. A maximum
likelihood solution will be used to solve the model displayed in Figure l. The
standardized regression coefﬁcients will be analyzed to provide support for the three
hypotheses explicated above. Signiﬁcance for paths will be held at a p < .05 two-tailed
test. The residual matrix will be examined for asymptotically standardized values greater
than 2.00. This indicates places of poor model ﬁt and modiﬁcations can be made to the
model to improve ﬁt. Paths between errors in the measurement models will be added to
improve ﬁt. Structural model modiﬁcations can also be made between the latent
variables. Model ﬁt will be determined by observing that the C2 /df is less than or equal to
2. The Comparative Fit Index (CF I) (Bentler, 1988), Goodness of Fit Index (GFI), Non-

38

Normed and Normed ﬁt indices (NNFI and NFI, respectively) (Bentler & Bonnet, 1980,
1987) will also be evaluated. A multiple groups model (Bentler, 1995) will be used to
evaluate gender between each estimated regression weight. The strength of the paths will
be evaluated for each gender group and if the difference between the two weights is
signiﬁcant, two paths will be produced, one for each gender. If they are not statistically
Signiﬁcant, they will be modeled as one path. Final models will be interpreted in the
discussion. All SEM models will be run using the Mplus Software (Muthen & Muthen, .
1998)

Section Three Hypptheses

3.1 The correlation between men’s achievement attributional style and perceived patemal
achievement attributional style will be signiﬁcantly greater than the correlation between
men’s achievement attributional style and perceived maternal achievement attributional
style. The reverse is hypothesized for women.

3.2 For both men and women, the correlation between young adult interpersonal
attributional style and perceived maternal interpersonal attributional style will be
signiﬁcantly greater than the correlation between young adult interpersonal attributional
style and perceived paternal interpersonal attributional style.

3.3 The relationship of perceived parental interpersonal attributional style to young adult
interpersonal attributional style, when both are of the same sex, will be signiﬁcantly
greater than the same relationship when the young adult is of the opposite sex.

3.4 The relationship of perceived parental achievement attributional style to young adult

achievement attributional style, when both are of the same sex, will be signiﬁcantly
greater than the same relationship when the young adult is of the opposite sex.

Section Three Analyses

The planned analyses for section three are outlined below. Four latent variables
will be added to the model presented in Figure 1. The Perceived Parental Attributional
Style Questionnaire (PPASQ) will assess each of these variables for each parent. Two
latent variables for each parent will assess Perceived Parental Attributional Style for
Achievement and Interpersonal Events. Three indicators will measure each latent
variable. This extended model is presented in Figure 2. A maximum likelihood

estimation technique will be used to evaluate each model, and sex differences will be

39

evaluated by using a multiple-groups modeling approach. This technique will allow paths
for each group to be derived if it is found that they are statistically different or of
different valences. For paths in which no sex difference is observed, one path will be

used to explain the variance between the variables.

Section Four Hypptheses

For these analyses, only 73 subjects of the original 773 were used. These subjects
were administered an anagrams task in which the subject failed at least 50 percent of the
time. The hypotheses and analyses examined the extent to which attributions about their
failure on the anagrams task are predicted by young adult attributional style and
perceived parental attributional style. For these analyses and hypotheses, it was assumed
that perceived paternal and maternal achievement attributional style will be more
strongly related to attributions for failure than perceived maternal and paternal
interpersonal attributional styles. Thus in the following models, only achievement
attributional styles are used. Given the conﬂicting ﬁndings about the intemality
dimension and the research indicating the importance of the speciﬁcity and stability
dimensions, in the next hypotheses, only these two constructs (speciﬁc-global and stable-
unstable) for the perceived parental attributions and the young adult attributions are used
to describe achievement attributions.
4.1 Men’s scores for intemality, globality, and stability for failure on the anagrarns task
(Particular Attribution Questionnaire) will be signiﬁcantly lower (more external, speciﬁc,

and unstable) than women’s scores.

4.2 Men’s scores on the BDI will be signiﬁcantly lower after anagrams failure than
females’ scores will be.

4.3 Achieved and moratorium status individuals will report a signiﬁcantly more satisﬁed
mood after failing (on the visual analog scale) than individuals in the foreclosed and
diffuse groups.

4. 4 Identity achieved and moratorium status individuals will exhibit signiﬁcantly more
external, unstable, and speciﬁc attributions on the failure task than the diffuse and
foreclosed groups will.

4.5 Manifest Variable Multiple Groups Model:

40

The following hypotheses are depicted in Figure 3. This ﬁgure illustrates a set of
relationships between variables collected in Sessions I and H, and indicates the
relationships between Perceived Parental Attributions, Young Adult Achievement Status,
Attributional Response to Failure, Depression and Satisfaction. Ideological Identity
Status will be seen as a dichotomous variable grouping diffuse and foreclosure as one
group and moratorium and achievement as the other. Relationships between the variables
proposed in the model will be evaluated based on Identity Status classiﬁcation. It is
hypothesized that identity achieved and moratorium status individuals will exhibit a more
external, unstable, and speciﬁc attributional style and diffuse and foreclosed groups to
exhibit a more internal, stable, global attributional style.

Path 1: Perceived Paternal Attributional Style for Achievement (PPAS-ACH)
will predict young adult Attributional Style for Achievement (YAS-ACH).

Path 2: Perceived Maternal Attributional Style for Achievement (PMAS-ACH)
will predict Young Adult Attributional Style for Achievement (YAS-ACH).

Paths 3 & 4: A direct relationship will be observed between both paternal and
maternal perceived parental attributional style and Attributional Failure outside of Young
Adult Attributional Style for Achievement.

Path 5: Young Adult Style for Achievement will predict Attributional Failure.

Path 6: Increased use of global and stable attributions for failure will lead to an
increase in depression scores.

Path 7: Increased use of global and stable attributions for failure will lead to
decrease in mood.

Identity interaction with paths 6 and 7 will be shown by little difference observed
in the emotional responses of achieved and moratorium persons (ideological domain) to
low and high levels of failure. It is hypothesized that the magnitude of the diffuse and
foreclosed individuals’ responses will increase greatly with increased failure.

Section Four Analyses
The planned analyses for section four are described in the following paragraphs.

Hypothesis 4.1 will be tested with a Multivariate Analysis of Variance procedure. Sex
will serve as the independent variable and the attribution subscales (global-speciﬁc and
stable-unstable scales) as the dependent variables. Gender differences will be observed
between mean levels of the dependent variable. Hypothesis 4.2 will be tested using an
Analysis of Variance to test the gender difference between Beck Depression Inventory
Scores measured after the anagrams task. Using a dichotomized version of the Identity
Scale (0 = Diffuse or Foreclosure; 1 = Moratorium or Achieved), the moratorium and
achieved group will be expected to have a signiﬁcantly lower mean on the visual analog
scale in an Analysis of Variance testing Hypotheses 4.3. Hypothesis 4.4 will use the same

dichotomized version of the Identity Scale to test the means of the global-speciﬁc and

41

stable-unstable scores of Young Adult Scales of Attributional Style.

For Hypothesis 4.5, a multiple-groups structural equation model will be run.
Manifest variables of Perceived Parental Attributional Style were will be created using
weighted sums of the measurement models derived in the ﬁrst two models. Standardized
regression coefficients will be multiplied by the items of the scales and summed. The
factor scores will then be used in the structural equation models. Latent variables will not
be used in these models because this sample has a much lower 11. Reducing the number of
variables used in the variance/covariance matrices will allow for these models to be
estimated. Making sure that each variable in the model is supported by ten subjects (i.e.
n/k ratios 10) will insure that the model derived will be stable and generalizable to the
larger population.

A maximum likelihood estimation method will be used to estimate all parameters.
A multiple group variable will be used to evaluate the need to include a path for each of
two levels of identity, (i.e. high and low, 1 and 0, diffusion/foreclosure and
moratorium/achieved) between each estimated regression weight. The strength of the
paths will be evaluated for each group and, if the difference between the two weights is
signiﬁcant, two paths will be produced, one for each level of identity. If they are not
statistically signiﬁcant, they will be held equal (modeled as one path). Signiﬁcance for
paths will be held at a p < .05 two-tailed test. The residual matrix will be examined for
asymptotically standardized values greater than 2.00. This indicates places of poor model
ﬁt and modiﬁcations can be made to the model to improve ﬁt. Fit functions will be
determined by observing that the chi-square degrees of freedom do not exceed two times
the value of the model chi-square. CFI, GFI, NNFI and NF I will be evaluated. Final

models are interpreted in the discussion.

42

Chapter Four: Results
Section 1 Results

1.1 In the interpersonal domain, there will be signiﬁcantly larger numbers of women (as
opposed to men) in the identity achieved and the moratorium statuses.

The chi-square analysis showed that sex and interpersonal identity status were not
independent (x2(4, p = 759) = 12.59, p < .01). There were signiﬁcant main effects for
both sex and for interpersonal identity status (Goodman & Kruskal Tau’s .003 and .017
respectively and p < .04 and p < .013 respectively). Contrary to predictions, there was a
larger percentage of men than women in the identity achieved category.

1.2 In the ideological domain, there will be signiﬁcantly larger numbers of men (as
opposed to women) in the identity achieved and moratorium groups.

The numbers of men and women in the achieved and moratorium statuses in the
ideological domain did not differ signiﬁcantly. Thus, hypothesis 1.2 is not supported.

1.3 Women’s attributional style scores for interpersonal events will be signiﬁcantly more
extemal/unstable/speciﬁc (lower) than their attributional style scores for negative
achievement events. ,

A MANOVA with the BDI score as a covariate was used to assess whether mean
scores on the three attributional dimensions (intemality, stability, globality) differed by
attributional domain (achievement, interpersonal). To run this analysis, it was necessary
to stack the cases in the data set (thereby doubling the sample size) to create a
blocking/grouping variable. The MAN OVA revealed a signiﬁcant main effect for
attributional domain (£(3,837) = 32.71 , p = .001, A = .89) and the BDI covariate was
signiﬁcant (£(3, 837) = 13.63, p =.000, A =.95). Univariate tests revealed that mean
scores for intemality and globality for interpersonal events (4.05 and 3.67 respectively)
were signiﬁcantly lower than those for achievement events (4.61 and 3.94 respectively)
as hypothesized (£(l,839) = 92.67, p = .000; £(1,839) = 13.88, p = .000 respectively).
Another MANOVA including the same variables demonstrated that the BDI x
attributional domain interaction was not signiﬁcant. Regressions showed that BDI scores

accounted for a signiﬁcant amount (p < .05) of the variance in each attributional style

43

dimension in both attributional domains with the exception of intemality for achievement

events as shown in Table 1.

 

 

 

 

Table 1
Regr_ession of Women’s Attributions on BDI Scores by Attributiopal Dom_ai_n
Atribution _B_ S_E_B If t p
Interpersonal Domain
Intemalityl .01 .01 .10 2.08 .03 74
Stability2 .02 .01 .20 4.14 .0000
GlobalityJ .04 .01 .25 5.35 .0000
Achievement Domain
Intemali " .01 .01 .05 .97 .3340
Stabili .02 .01 .15 3.17 .0016
Globality" .02 .01 .15 3.17 .0016

1. 50,423) = 4.36, p = .0374, 32 = .10 4. £0,428) = 0.94, p = .3340, 32 = .05
2. 50,420) = 17.16, p = .0000, 32 = .04 5. £0,428) = 10.07, p = .0016, 112 = .15
3.1:(1,420)= 28.61, 2 = .0000, R’ = .25 6. £0,428) = 10.05, p = .0016, E = .15

 

Note: With the exceptTon ofE values, ﬁgures are rounded to two decimarplaces.

Dependent t-tests were also run to examine whether women’s attributions for
intemality, stability, and globality differed across attributional domain (interpersonal,
achievement). Means for all three attributional dimensions in the interpersonal domain
were signiﬁcantly (p < .05) lower than the corresponding value in the achievement

domain (See Table 2). Thus, both sets of analyses provide support for hypothesis 1.3.

 

 

Table 2

Mean Attribution Scores by Sex and Attributional Domain

Sex Attributional Domain

Women Achievement (SQ) Interpersonal (SD)
Intemality 4.63, (0.88) 4.05, (0.78)
Stability 4.23, (0.87) 4.13,, (0.79)
Globality 3.96, (1.11) 3.68b (1.04)

Men
Intemality 4.43, (0.97) 4.06b (0.81)
Stability 4.17, (0.90) 4.16, (0.78)
Globality 4.05,l (1.02) 3.77,, (0.88)

 

Note. Means in the same row having different subscripts differ at p < .05 using a t-test.

44

1.4 Men’s attributional style scores for achievement events will be signiﬁcantly more
external, unstable, speciﬁc (lower) than their attributional style scores for negative
interpersonal events.

A MANOVA with BDI score as a covariate was used to assess whether mean
scores on the three attributional dimensions (intemality, stability, globality) differed by
attributional domain (achievement, interpersonal). To nm this analysis, it was necessary
to stack the cases in the data set (thereby doubling the sample size) to create a
blocking/ grouping variable. There was a signiﬁcant main effect for attributional domain
(£(3,451) = 9.56, p < .001, A = .94) and the BDI covariate was signiﬁcant (£(3,451) =
3.20, p = .023, A = .98). Univariate tests revealed that mean scores for intemality and
globality for interpersonal events (4.07 and 3.75 respectively) were signiﬁcantly lower
than those for achievement events (4.43 and 4.03 respectively) (£(1,453) = 18.46, p = .
.000; £( 1,453) = 9.93, p = .002 respectively). A separate MANOVA showed that the BDI
x attributional domain was not signiﬁcant. Regressions revealed no signiﬁcant
relationship between BDI score and any attributional dimension in the achievement
domain. Regressions Showed a signiﬁcant relationship of BDI to men’s attributions for
intemality and globality, but not stability, in the interpersonal domain (See Table 3).

Dependent T—tests were also run to examine whether men’s attributions for
intemality, stability, and globality differed across attributional domain (interpersonal,
achievement). Men’s intemality and globality scores for interpersonal events were
signiﬁcantly (p < .05) lower than the corresponding score for achievement events.
Stability scores did not differ across domains (See Table 2). Thus, there is not support for
hypothesis 1.4.

1.5 Women will have a signiﬁcantly more extemal/unstable/speciﬁc attributional style
than men for negative interpersonal events.

A MANOVA with BDI scores as a covariate was used to assess whether mean
scores on the three attributional dimensions (intemality, stability, globality) differed by
sex. The main effect was not signiﬁcant although the BDI covariate was signiﬁcant (£(3,
628) = 12.24, p = .000, A = .94). A separate MANOVA using the same variables showed

45

that the BDI x sex was not signiﬁcant. Regressions conducted separately by sex showed
that BDI scores accounted for a signiﬁcant (p < .05) amount of variation in each of the
three attributional dimensions with one exception, men’s attributions for stability (See
Tables 1 and 3). Thus, hypothesis 1.5 is not supported.

1.6 Men will have a signiﬁcantly more external/unstable/ speciﬁc attributional style than
females for negative achievement events.

A MANOVA with BDI scores as a covariate was used to assess whether mean
attributional style scores for achievement events (intemality, stability, globality) differed
by sex. There was a signiﬁcant main effect for sex (£(3, 660) = 3:26, p < .02, A = .98)
and the BDI covariate was signiﬁcant (£(3, 660) = 4.39, p = .005, A = .98). Univariate
test results showed that men’s mean score (4.43) for intemality was signiﬁcantly lower
than women’s (4.61) (£(1,662) = 4.98, p < .03). A separate MANOVA using the same
variables found that the BDI x sex was not signiﬁcant. BDI scores did not account for
variation in men’s attributions for achievement events; all regressions were not
signiﬁcant. BDI did account for signiﬁcant variation in women’s attributions for stability
and globality in the achievement domain (see Tables 1 and 3). Therefore there is some

support for hypothesis 1.6.

 

 

 

Table 3 ,
Reggession of Men’s Attributions on BDI Scores by Attributional Domain
Attribution B - __S_E B 15 t J
Interpersonal Domain ‘
IntemalityI .02 .01 .16 2.39 .0177
Stability2 .01 .01 .1 l 1.64 .1032
Globality3 .02 .01 .14 2.16 .0321
Achievement Domain
Intemality4 .01 .01 .08 1 .22 .2242
Stability’ .00 .01 .03 0.43 .6644
Globality" .01 .01 .08 1.21 .2271

1. £(1,219) = 5.71, p = .0177, _R’ = .16 4. £(1,233) = 1.49, p = .2242, Bf = .08
2. £(1,219)=2.68,p=.1032, RZ=.11 5. £(1,233)=0.l9,p=.6644, 32:03
3. £(1,219) =4.65, p= .0321, R} = .14 6. £(1,233) = 1.47, p= .2271, 32 = .08

 

Note: With the exception of p values, ﬁgures are rounded to two dECImal places.

46

Section 2 Results
2.1 The mean attributional style scores for individuals in the achievement and
moratorium statuses will be signiﬁcantly more external, unstable, and speciﬁc (lower)
than the mean scores for persons in the diffusion or foreclosure statuses for both the
ideological and interpersonal realms.

For hypothesis 2.1, two MANOVAS with BDI score serving as a covariate in each
were used to examine whether attributions (intemality, stability, and globality based on
the full EASQ) differed by identity status (treated as a 4 level ordinal variable). The ﬁrst
MANOVA did not ﬁnd a signiﬁcant main effect for ideological identity status, but the
BDI covariate was signiﬁcant (£(3, 648) = 8.45, p = .000, A = .96). A MANOVA
conducted to examine the BDI score x ideological identity status interaction found that it
was not signiﬁcant. Regressions indicated that BDI scores accounted for a signiﬁcant
amount of the variance in each attributional dimension using items from the entire EASQ
(£(3,682) = 9.45, p < .0001, A = .96).

The second MANOVA which used interpersonal identity status as an independent
variable found no signiﬁcant main effect. The covariate (BDI score) was signiﬁcant (£(3,
645) = 8.39, p = .000, A = .96). Another MANOVA showed that the BDI x interpersonal
identity status was not signiﬁcant. Regressions showed that the BDI score accounted for
a signiﬁcant amount of variance in the three overall attributional style dimensions (£ (3,
682) = 9.45, p < .0001 , A = .96). Therefore, hypothesis 2.1 is not supported.

2.2 Mean attributional style scores for negative achievement events will be signiﬁcantly
lower (external, unstable, speciﬁc) for the achieved and moratorium identity statuses in

the ideological domain than for the diffuse and foreclosed groups in the ideologi
domain.

To test hypothesis 2.2, a MANOVA was done in which attributions for
intemality, stability, and globality in the achievement domain were dependent variables,
ideological identity status was the independent variable, and BDI score was the covariate.
There was not a signiﬁcant main effect for ideological identity status, but the BDI
covariate was signiﬁcant (£(3,629) = 3.51, p = .01, A = .98). A MANOVA examining the

BDI x ideological identity status interaction showed that it was not signiﬁcant.

47

Regressions demonstrated that the BDI accounted for signiﬁcant variation in some

attributions for achievement events (see Table 4). Hypothesis 2.2 is not supported.

2.3 Mean attributional style scores for negative interpersonal events will be signiﬁcantly
lower (external, unstable, speciﬁc) for the achieved and moratorium identity statuses in
the interpersonal domain than for the diffuse and foreclosed groups in the interpersonal
domain.

 

 

 

 

Table 4

Re ssion of Youn Adult Attributions on BDI Scores b Attributional Domain

Attribution ; S_E _B t p

Interpersonal Domain _
Intemalityl .01 .00 3.12 .0019
Stability2 .02 .00 4.25 <.0001
GlobalityJ .03 .00 5.36 <.0001

Achievement Domain
Intemali " .01 .00 1.77 .0764
Stabili .01 .01 2.83 .0049
Globality" .02 .01 3.16 .0017

1. £(l, 631) =.9.75, p = .0019, 32 = .01 4. £0,663) = 3.15 , p = .0764, 32 = .01
2. £(1, 631) = 18.06, p = <.0001, 32 = .03 5. EU, 663) = 7.99, p = .0049, 32 = .01
3. F(1, 631) = 28.75, p = <.0001, 1:2 = .04 6. £(l, 663) = 9.96, p = .0017, 32 = .02

 

Note: With the exception of p values, ﬁgures are rounded to two decimal places.

To test hypothesis 2.3, a MAN OVA was conducted using intemality, stability, and
globality scores for the interpersonal domain as dependent variables, interpersonal
identity status as the independent variable, and BDI score as the covariate. The main
effect was not signiﬁcant, but the covariate was (£ (3,595) = 10.34, p = .000, A = .95). A
MAN OVA done to investigate the BDI score x interpersonal identity status term revealed
that it was not signiﬁcant. Regressions showed that BDI score had a direct effect on all
three attributional style dimensions (£ (3, 629) = 11.72, p < .0001, A = .95) (see Table 4).
Hypothesis 2.3 is not supported.

2.4 The correlation between achievement attributional style scores and identity status for
the ideological domain will be signiﬁcantly greater than the correlation between
achievement attributional style scores and identity status for the interpersonal domain.
2.5 The correlation between interpersonal attributional style and identity status for the

interpersonal domain will be signiﬁcantly greater than the correlation between
interpersonal attributional style and identity status for the ideological domain.

48

Although a latent variable model was planned to evaluate hypotheses 2.4 and 2.5
(Figure 1), it was not possible to analyze the model for two reasons. First, the algorithm
originally proposed to create a continuous identity status variable with the items of the
EOMEIS by weighting the responses with an ordinal scale was ineffective in producing a
scale where high values represented greater exploration during identity development.
Instead, the participants’ responses dictated the overall level of the score. For example, if
someone responded with high values on all foreclosure (status 2) items, the overall score
could be higher than someone who scored achievement (status 4) items highly (relative to
their own responses for other statuses) but in absolute values lower than the ﬁrst person.
It was decided that it was best to treat identity status as an ordinal scale moving from
diffusion to achievement and to use the Adams et al. (1979) scoring guidelines. Second,
it was not possible to create the hypothesized achievement and interpersonal attributional
style factors. More speciﬁcally, the intercorrelations between attributional style
dimensions across the proposed attributional domains (e. g. the correlation of intemality
for achievement and for intemality for interpersonal events) were much greater than any ,
common variance among the two proposed attributional domains. The placement of
residual covariances between the individual style pairs was not sufﬁcient to produce an
adequate solution. Therefore, regression equations were used to examine the
relationships outlined in hypotheses 2.4 and 2.5. Independent variables entered
simultaneously into all equations. The conﬁdence intervals (95%) were examined to see
if the regressions differed signiﬁcantly from each other. Ideological identity status and
interpersonal identity status were regressed separately onto intemality, stability, and
globality attribution scores for the achievement domain (see Table 5).

Neither regression was signiﬁcant. Ideological identity status and interpersonal
identity status were then regressed separately on intemality, stability, and globality scores
for the interpersonal domain of attributional style (see Table 6). Only the regression of

interpersonal identity status on attributions for the interpersonal domain was signiﬁcant

49

(32: .013, 13(3, 657) = 2.93, p < .03).

Table 5
Regression of Ideological and Interpprsonal Identity Status on Young Adult Attributions
for Achievement

 

 

 

Attribution 1}: _E B B _t_ p
Ideological Identity Status _
Intemality -.Ol .03 -0.01 -0.22 .82
Stability .01 .04 0.01 0.25 .80
Globality -.01 .03 -0.01 -027 .78
Interpersonal Identity Status
Intemality .00 .03 .00 .1 l .91
Stability .05 .04 .05 1.14 .25
Globality -.02 .03 -.03 -0.74 .46

 

Notezfigures are rounded to two decimal places.

Table 6

Regression of Ideological and Intemersonal Identig Status on Young Adult Attributions
for lntermrsonal Events

 

 

 

Attribution B S_E B [5 I p
Ideological Identity Status
Intemality -.00 .05 -0.00 -0.05 .96
Stability .09 .05 0.09 1.95 .05
Globality -.07 .04 -0.09 -1 .99 .05
Interpersonal Identity Status
Intemality -.10 .04 -.09 -2.30 .02
Stability .05 .04 .05 1.1 1 .27
. Globality -.05 .04 -.06 -1.37 .17

 

Note: Figures are rounded to two decimal places.

There was no support for hypothesis 2.4; since the regressions were not
signiﬁcant, the regression coefficients are essentially zero and do not differ from each
other. There was support for hypothesis 2.5. Given that the regression of interpersonal
identity status on the attributions for interpersonal events was signiﬁcant and the
regression of ideological identity status on the same variables was not, one can infer that

the regression coefficients from the ﬁrst equation are signiﬁcantly greater than those of

50

the second (which are effectively zero).

Section 3 Results
Since achievement and interpersonal attributional style factors were not created,
the planned structural equation modeling (Figure 2) was not done and regressions were
performed to examine hypotheses 3.1 — 3.4. Independent variables entered the regression
equations simultaneously.
3.1 The correlation between men’s achievement attributional style and perceived paternal
achievement attributional style will be signiﬁcantly greater than the correlation between

men’s achievement attributional style and perceived maternal achievement attributional
style. The reverse is hypothesized for women.

3.2 For both men and women, the correlation between young adult interpersonal
attributional style and perceived maternal interpersonal attributional style will be
signiﬁcantly greater than the correlation between young adult interpersonal attributional
style and perceived paternal interpersonal attributional style.

To test hypothesis 3.1, young adult scores on each attributional dimension
(intemality, stability, and globality) for the achievement domain were regressed
separately on perceived maternal and perceived paternal scores for the same dimension
and domain. Separate analyses were done for males and females. For example, men’s
scores for globality in the achievement domain were regressed onto perceived paternal
and perceived maternal globality scores for the achievement domain. To test hypothesis
3.2, the aforementioned steps were repeated for the interpersonal domain. Each
regression conducted to test 3.1 and 3.2 was signiﬁcant at the .0001 level (Table 7).

Examination of the regression coefﬁcients and their conﬁdence intervals (95%)
revealed little support for hypothesis 3.1 (Table 8). For men, in the achievement
attributional domain, only the regression coefﬁcient for perceived paternal attributions
for globality was signiﬁcantly (p < .05) greater than the corresponding coefficient for

perceived maternal attributions for globality. For women in the achievement domain, the

regression coefﬁcients for perceived paternal attributional style were signiﬁcantly (p <

51

.05) greater than those for perceived maternal attributional style for both stability and
globality. There were no other signiﬁcant differences.

There was no support for hypothesis 3.2. The regression coefﬁcients for

Table 7
‘ Summm of Regressions of Young Adult Attributions on Perceived Paterﬂl and

Maternal Attributions by Sex and by Attributional Domain

 

 

Attribution Rf £
Achievement Domain
Men
intemality .09 13.34
. stability .34 62.97
globality .37 73.19
Women
intemality .10 22.90
stability .28 83.29
globality .40 141.89
Interpersonal Domain
Men
intemality .09 1 0.82
stability .34 59.71
globality .35 62.68
Women
intemality .1 1 24.29
stability .3 l 90.51
globality .32 89.10

 

Note: All ﬁgures rounded to two decimal places. All models signiﬁcant at p < .000H

perceptions of fathers’ attributions were consistently larger than those for perceptions of
mothers’ attributions. For women, the differences between the coefﬁcients were
signiﬁcant (p < .05) on each attributional dimension. For men, only the regression
coefficient for perceptions of paternal attributions for globality was signiﬁcantly (p <
.05) greater than the corresponding maternal regression coefﬁcient (see Table 8).

3.3 The relationship of perceived parental interpersonal attributional to young adult
interpersonal attributional style when both are of the same sex will be signiﬁcantly
greater than the same relationship when the young adult is of the opposite sex.

3.4 The relationship of perceived parental achievement attributional style to young adult
achievement attributional style when both are of the same sex will be signiﬁcantly

greater than the same relationship when the young adult is of the opposite sex.

Regressions were used to evaluate hypotheses 3.3. and 3.4. Young adult

52

attributions (intemality, stability, and globality) for each attributional domain
(achievement and interpersonal) were regressed individually on perceived maternal
attribution scores for the same attributional dimension and the same attributional domain.
Table 8

Overview by Sex an_d Attributional Domain of Reggssion Coefﬁcients from Regressing
Young Adult Attributions onto Perceived Parental Attributions

 

 

 

 

Attributional ,
Dimension Predictor _B_ SE1; [3
Achievement Domain
Men
intemality paternal .16 .06 .18
maternal . 1 9 .06 .19
stability paternal .28 .06 .30
maternal .33 .06 .35
globality paternal“ .55 .06 .56
maternal .07 .06 .08
Women
intemality paternal .17 .04 .22
maternal . 16 .04 .1 8
stability paternal“ .36 .04 .40
maternal . 1 8 .04 .19
globality paternal“ .43 .04 .45
maternal .26 .04 .26
Interpersonal Domain
Men
intemality paternal .16 .06 .18
maternal .1 5 .06 .1 7
stability paternal .30 .06 .35
maternal .28 .06 .3 1
globality paternal“ .45 .06 .51
maternal .1 1 .06 . 1 2
Women
intemality paternal“ .22 .04 .26
maternal .12 .04 .14
stability paternal“ .37 .05 .40
maternal .20 .05 .22
globality paternal“ .41 .05 .39
maternal .26 .05 .25
f p< .05

Note: All numbers rounded to 2 decimal places

53

This was done separately for men and for women. The procedure was then repeated using
perceived paternal attribution scores as the predictor.

Each regression testing hypotheses 3.3 and 3.4 was Signiﬁcant at least at p <
.0003 indicating that perceived maternal and perceived paternal attributional styles for
achievement and perceived maternal and perceived paternal attributional styles for
interpersonal events accounted for signiﬁcant amounts of variation in young adult
attributional style for achievement events and for interpersonal events respectively.
Examination of the conﬁdence intervals (95%) for the regression coefficients reveals
little support for 3.3 and 3.4. There was one signiﬁcant regression supporting 3.4. When
women’s globality scores for achievement events were regressed on perceived maternal
attributions for globality in the achievement domain, the regression coefﬁcient was
signiﬁcantly greater than the corresponding regression coefficient for men (Tables 9 and
10).

Table 9

Sex differences in the relationship of young adult attributions to mrceived parental
attributions in the intemrsonal domain

 

 

 

 

Attributional Males Females
Dimension B SE B B SE B
Perceived Paternal
Intemality .20 .05 .24 .04 n.s.
Stability .46 .05 .48 .04 n.s.
Globality .52 .05 .55 .04 n.s.
Perceived Maternal
Intemality .23 .06 .1 8 .04 n.s.
Stability .45 .05 .38 .04 n.s.
Globality .41 .05 .48 .04 n.s.

 

Note: Numbers rounded to two decimal places

Section Four Results
4.1 Men’s scores for intemality, globality, and stability for failure on the anagrams task
(particular attribution questionnaire) will be signiﬁcantly lower (more external, speciﬁc,
unstable) than women’s scores.

An AN OVA as opposed to the planned MANOVA was used to test this

54

hypothesis. Given the limited number of items comprising the intemality (1 item),
stability (2 items), and globality (3 items) scales on the PAQ, the scales were combined
to enhance the reliability of this outcome measure. This created a composite measure of
attributional style for failure on the anagrams task. The overall attributional style score
from the PAQ served as the dependent variable and sex was the independent variable.
The AN OVA revealed no sex differences and revealed no support for hypothesis 4.1.
4.2 Men’s scores on the BDI will be signiﬁcantly lower after anagrams failure than
female’s scores will be.

The AN OVA was performed with post-failure BDI score as the dependent
variable and sex as the independent variable. The main effect was not signiﬁcant.
Therefore, hypothesis 4.2 is not supported.

4.3 Achieved and moratorium status individuals for the ideological domain will report a

signiﬁcantly more satisﬁed mood after failing (on the visual analog scale) than
individuals in the foreclosed and diffuse groups for the ideological domain.

Table 10

Sex differences in the relationship of young adult attributions to mrceived parental
attributions in the achievement domain

 

 

 

 

Attributional Males Females
Dimension B SL1; B BLT;
Paternal
Intemality .24 .05 .20 .04 n.s.
Stability .49 .05 .43 .04 n.s.
Globality .60 .05 .56 .04 n.s.
Maternal
Intemality .24 .06 .21 .04 n.s.
Stability .46 .05 .38 .04 n.s.
Globality .38 .05 .52 .04 p< .05

 

Note: NumRrs rounded to two decrfnal places

An AN OVA was used to test hypothesis 4.3. Satisﬁed mood (V AS) was the
dependent variable. A dichotomized version of the ideological identity status scale was
the independent variable. Results were not Signiﬁcant and thus hypothesis 4.3 is not

supported.

55

4.4 Identity achieved and moratorium status (ideological domain) individuals will exhibit
signiﬁcantly more external, unstable, and speciﬁc attributions on the failure task than the
diffuse and foreclosed groups (ideological domain) will.

An AN OVA using PAQ score as the dependent variable and ideological identity
status as the independent variable. Identity status was treated as a dichotomous variable.

The main effect was not signiﬁcant. Therefore, there is no support for hypothesis 4.4.

SEM Analysis for Phase 1
When the ﬁrst of the models proposed in Chapter 3 were run, it was apparent that

several issues prevented the models from converging. One was that the interpersonal and
achievement distinctions among the attributional styles were not distinct sources of
variance in the models. Instead, the similarities between globality, stability and
intemality overwhelmed the common variance across the achievement and interpersonal
domains. The second reason these models did not run was that the relationship between
identity status and attributional styles (regardless of ideological/interpersonal or
intemal/stable/ global dimensions respectively) was essentially random, making any of
these models impossible to derive. Thus, the solution was to ﬁnd models in which
relationships could be modeled reliably and where SEM would provide a different and
unique way of displaying the results that the other analysis presented so far would have
missed. Therefore, young adult attributional styles were regressed on perceived parental
attributional styles within the three dimensions of intemality, stability, and globality. The
model was run saturated within each attributional style dimension and the non-signiﬁcant
paths were deleted and the model was reﬁned.

The ﬁrst group of subjects, 27 8 men and 494 women, were selected for inclusion
in the models. Missing data were not allowed in the model estimation procedures, thus
232 men were included in the ﬁrst model and 372 women were included in the second
model. A maximum likelihood solution in SAS (SAS Institute Inc. 1999) was used to

ﬁnalize the models and all paths not achieving signiﬁcance were eliminated. Figure 4

56

shows the general hypothesized model run ﬁrst for each sex and Figures 5 and 6 were the
ﬁnal models based on signiﬁcance. In this analysis, the overall young adult attributional
styles (based on all 24 EASQ items thus collapsing achievement and interpersonal
domains with other types of events) were regressed on the attributional styles of each
perceived parent while keeping the achievement and interpersonal domains separate. In
this way, the makeup of overall young adult attributional style would be predicted by the
perceptions of the domain-speciﬁc styles of each parent. Interpersonal and Ideological
Identity status were regressed on the young adult attributional styles.

An initial run of the model presented in Figure 4 revealed that the Identity Status
scores were not signiﬁcantly related to any of the variables they were being regressed
upon. Thus, these variables were eliminated. The models were pared down by
eliminating paths that were not signiﬁcant on the basis of high residual pairs in the
residual matrix. Paths, which were initially eliminated, were replaced when other paths
were removed when warranted by the residual matrix. The chi-square test, the NF], NNFI
and the CPI were examined to Show model ﬁt and thus, the process was complete when
adding or removing paths resulted in no signiﬁcant change to these ﬁt functions.

The model for the men (Figure 5) included only one path that was not signiﬁcant,
mother’s achievement global attributional style to young adults global attributional style.
This was needed in order to improve model ﬁt. Even though the path itself is not
signiﬁcant, it still models enough variances as to reduce high values in the residual
matrix. Covariances were established between the error terms of young adult (YA)
intemality and YA stability, and YA stability and YA globality. The model ﬁt indices
revealed a good to excellent ﬁt (x2 = 53.56, d_f = 27, p = .0017; CFI = .9828, GFI = .9689,
NNFI = .9421 and NFI = .9673). The chi-square was signiﬁcant even though the chi-
square value was no greater than two times its degrees of freedom. This is usually the
case with model with larger sample sizes. The other ﬁt indices should be near 1.00 and

most of these were very close.

57

Standardized parameter estimates are included on each path. The unstandardized
parameter coefﬁcients, standard errors and t values are included in Table 11. The ﬁnal
model for men indicated that YA intemality was predicted by perceptions of mother’s
interpersonal intemality and perceptions of father’s achievement intemality. YA stability
was predicted by perceptions of father’s interpersonal stability and perceptions of
mother’s achievement stability and YA Globality was predicted by perceptions of

father’s achievement globality and perceptions of father’s interpersonal globality.

 

 

 

 

 

 

 

 

 

Table 11

Unstandardized Parameter Estinrates for the Men’s Model
Path Parameter Standard r

Estimate Error value

Perceived Paternal Achievement Intemality -) YA Internal AS .0982 040] 2,45
Perceived Maternal Interpersonal Intemality -D YA Internal AS , 1477 .0515 2.87
Perceived Paternal Interpersonal Stability -D YA Stability AS .3107 .0421 7.38
Perceived Maternal Achievement Stability 4 YA Stability AS .2512 .03 95 6.35
Perceived Paternal Achievement Globality -9 YA Global AS .3107 1 .05 52 5.63
Perceived Paternal Interpersonal Globality 4 YA Global AS .2607 .0540 4.83
Perceived Maternal Achievement Globality 4 YA Global AS .0528 .1476 1.1 l

 

 

 

 

 

The model for the women (Figure 6) did not include any paths that were not

Signiﬁcant. An additional path was added from the disturbance terms of YA intemality

and YA globality in addition to the same two covariances included in the men’s model.

The same process was used to derive this model from the larger one and the ﬁt functions
indicated a good to excellent ﬁt (x2 = 56.56, df = 27, p = .0007; CFI = .9898, GFI =
.9807, NNFI = .9605 and NFI = .9812). The unstandardized parameter coefficients,

standard errors and p values are included in Table 12.

The ﬁnal model indicated that women’s intemality was predicted by perceptions

of father’s interpersonal internal attributional style (AS) and perceptions of father’s

58

 

achievement internal AS, women’s YA stability was determined by perceptions of
father’s interpersonal stability, perceptions of father’s achievement stability and
perceptions of mother’s interpersonal stability. Women’s globality was determined by
perceptions of both parents’ attributions for both attributional domains (achievement,

interpersonal) for globality.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Dirglfaritiardized Parameter Estimates for Lh_e Women’s Model
Path Parameter Standard r
Estimate Error value

Perceived Paternal Achievement Intemality 4 YA Internal AS .0927 .0339 2.73
Perceived Paternal Interpersonal Intemality 4 YA Internal AS .0836 .0397 2.12
Perceived Paternal Achievement Stability 4 YA Stability AS .2289 .0405 5.65
Perceived Paternal Interpersonal Stability 4 YA Stability AS .1639 .0487 3.37
Perceived Maternal Interpersonal Stability 4 YA Stability AS .1268 .0374 3.39
Perceived Paternal Achievement Globality 4 YA Global AS .2382 .0406 5.87
Perceived Paternal Interpersonal Globality 4 YA Global AS .1486 .0464 3.20
Perceived Maternal Achievement Globality 4 YA Global AS .1846 .0421 4.38
Perceived Maternal Interpersonal Globality 4 YA Global AS .0981 .0467 2.10

 

 

 

SEM Phase H Analysis

A structural equation model was used to examine the relationship of perceived
parental and young adult generality scores for’achievement events to attributions for and
emotions after failure. Scores on the DACL were substituted for the BDI scores. There
were two reasons for this. First, some work (Metalsky et al., 1987; Weiner et al., 1979)
suggests that the impact of an attribution can be time delayed and that initial reaction is
based solely on the positive or negative nature of the event. The DACL was administered
later in the protocol. Second, there was greater variation in the DACL scores than in the

59

BDI scores. The BDI asks for mood over a two week period so it was probably not
sensitive enough for the intended purpose and the DACL, which asks about immediate
mood, would be better able to capture the respondent’s emotional state. One further
modiﬁcation was made to the planned analyses. Identity status was omitted, as previous
analyses showed it was not related to young adult or perceived parental attributional
styles.

The initial model run was performed on 27 men and 47 women. Since missing
data were not allowed in this analysis, the analysis was only performed on 67 subjects, 24
men and 43 women. The initial model run is proposed in Figure 7. The model was run
with all variables included even if no paths were signiﬁcant to them. This was done
because both a theoretical and an experimental design were involved here and the order
of placement of the variables was time-dependent. The ﬁnal model is shown in Figure 8.
This model was run with SAS and modiﬁed with high residual values and chi-square
tests. The model converged with a maximum likelihood solution and its ﬁt indices
appeared to suggest a good ﬁt (x2 = 13.96, pf = 7, p = .0519; CFI = .9042, GFI = .9427,
NNFI=.7946 and NFI = .8407). Only four paths were signiﬁcant: a) Perceived Paternal
Generality (stability + globality) and Young Adult Generality, a) Perceived Maternal
Generality (stability + globality) and Young Adult Generality, c) Perceived Maternal’s
Generality to Young Adult failure attributions and (1) Failure attribution to scores of
satisfaction versus down mood as measured by the visual analog scale. Standardized
coefficients are Shown in Figure 8, the unstandardized parameter estimates standard

errors and _t_ values are shown in Table 13.

60

Table 13

Unstandardized parameter estimates for phase II ﬁnal model

 

 

 

 

 

 

 

 

 

 

 

 

Path Parameter Standard 1
Estimate Error value
Perceived Paternal General ity 4 YA Generality .6118 .1249 4.90
Perceived Maternal Generality 4 YA Generality .1782 .1202 1.48
Failure Attribution 4 Visual Analog Scale .0671 .0289 2.32
Failure Attribution 4 Depression Score .0619 .0861 0.72
YA Generality 4 Failure Attribution .7338 1.19 0.62
Perceived Paternal Generality 4 Failure Attribution -1 .75 56 1.4104 -1.24
Perceived Maternal Generality 4 Failure Attribution 2.4635 1.1817 2.084

 

61

 

Chapter Five: Discussion

The reformulated theory of learned helplessness (Abramson et al., 1978) does not
specify whether or how causal attributions relate to developmental, familial, or
sociocultural factors. The present study begins to put the reformulated theory of learned
helplessness in a broader conceptual framework. In this research we examined the
relationship of attributional style to identity status (Marcia, 1966), the constancy of
attributional style across content areas, and the possible effects of societally assigned
roles and cultural norms on attributional style. In addition, the relationships of young
adult attributional style to perceptions of maternal and paternal attributional style were
reviewed to clarify further the role of family factors. Finally, the relationship of
perceived parental and young adult attributional styles to young adult attributions for, and
mood after, failing an anagram task were studied. The results suggest that attributional
style is not related to the formation of identity status and raise the question of how
attributional style might relate to other aspects of development. The ﬁndings suggest the
importance of understanding young adults’ perceptions of parents and the relationship of
these perceptions to young adult response to failure. Additionally, the results suggest the
value of attending to differences in explanatory style based on content area and of
articulating the relationship of societal roles and norms to an individual’s causal

reasoning.

Attributional Style and Identig Status
There was no support for the hypothesis that young adult explanatory style might
hinder or facilitate identity exploration and development. Young adults’ attributions fOr
negative events did not differ on any attributional dimension based on identity status.
This was true even when considering attributions and identity statuses with similar
content, e.g. attributions for interpersonal events and interpersonal identity status that

were hypothesized to be more closely related. Thus, it appears that a depressogenic

62

attributional style (internal, stable, global attributions for negative events) does not limit
identity exploration and that a more optimistic style (Seligrnan, 1990) (external, unstable,
speciﬁc attributions for negative events) does not directly facilitate exploration.
Therefore, it seems that attributional style does not account for the tendency of
individuals in the achievement and moratorium identity statuses to conﬁ'ont problems
actively and effectively (Grotevant & Adams, 1984), to perform better on stressful tasks
(Marcia, 1966), and to be less vulnerable to self-esteem manipulation (Marcia, 1967).
Some data did support the hypothesis that similarity of content could alter the magnitude
of the relationship of attributional style for a particular domain (achievement,
interpersonal) to identity status (ideological, interpersonal). The relationship of
attributional style for interpersonal events to interpersonal identity status was
signiﬁcantly greater than the relationship of interpersonal attributional style to
ideological identity status. However, it is important not to overstate this ﬁnding as
attributions for interpersonal events accounted for only a small portion (1.3%) of the
variance in interpersonal identity status. I
The failure to ﬁnd a direct relationship between attributional style and identity
status raises the question of how attributional style might relate to other aspects of young
adult development, if at all. Given the documented relationship of attributional style to
depression (Peterson & Seligrnan, 1984) and the consistent ﬁnding that Beck Depression
Inventory scores had a signiﬁcant effect on identity status in this sample, it seemed
possible that depression might mediate the relationship of attributional style to identity
status. Correlations were run to assess whether mediation was likely. The failure to ﬁnd a
signiﬁcant relationship between BDI scores and identity status suggested that depression
does not mediate the relationship of attributional style to identity status for any
attributional domain (overall, achievement, interpersonal) or identity status (ideological,
interpersonal). It may be more fruitful for researchers interested in studying attributional

style within a developmental framework to consider other aspects of development.

63

However, it is possible that another measure of identity development might ﬁnd a
relationship to attributional style, especially given the debate over the nature of identity
status (Coté & Levine, 1988a, 1988b; Waterman, 1988). The apparent lack of
relationship between causal attributions and identity status also suggests that young
adults grappling solely with identity concerns would derive little direct beneﬁt ﬁ'om

therapeutic interventions geared toward modifying attributional style

Societal Roles and Norms: Relationship to Attributional Domain and Sex Differences
The ﬁndings support the hypothesis that attributional style is embedded in and

related to the societal context in which a person lives and the cultural norms to which the
individual has been exposed. Through societally assigned roles and cultural norms
(Chodorow, 1978; Parsons & Bales, 1955) and stereotypes (Rosenkrantz etal., 1968),
mainstream US. society gives different messages about achievement and interpersonal
events to men and women. As predicted, these differences are reﬂected in young adult
explanatory styles for each attributional domain as well as in sex differences in
attributional style. As hypothesized, based on societal roles and norms for women and
interpersonal matters, women’s attributions for interpersonal events were more external,
unstable, and speciﬁc than their attributions for achievement events. And, as predicted
based on the norms and roles that society ascribes to men with respect to work, men’s
intemality attributions for achievement events were signiﬁcantly more optimistic than
women’s. Hence, there is support both for the hypothesis that women and men explain
negative events in a manner that facilitates fulﬁlling societally deﬁned roles and norms,
and for the hypothesis that societal factors are related to sex differences in attributional
style.

Two ﬁndings appear to challenge this interpretation of the relationship of
attributions to societal roles and norms. First, men and women both make Signiﬁcantly

more optimistic attributions for interpersonal events than for achievement events.

64

Second, men’s and women’s attributions for interpersonal events do not differ. Although
these ﬁndings initially appear to contradict the socialization hypothesis, they may mean
that interpersonal relations are so central to society that both sexes learn to explain them
in similar ways. They may also mean that there are larger differences with respect to
societal norms and roles for achievement for men and women. Additionally, this pattern
of results could imply that there is an evolutionary beneﬁt for both sexes to explain
interpersonal events in a more optimistic manner. For example, it might provide an
evolutionary advantage for individuals living in family units or in larger communities to
approach interpersonal situations with a mind-set that helps them to view problems as
temporary and ﬁxable.

Although the sex differences noted above are based on only the achievement and
interpersonal items, they appear to contradict Johnston and Page’s (1991) ﬁndings
showing that men’s attributions were more depressogenic than women’s, and other work
ﬁnding no sex differences (Bemdt et al., 1982; Handal et al., 1987; Johnson, 1992;
Whitley et a1, 1991 ). The ﬁnding of sex differences in attributional style for achievement
events and the differences in both sexes’ attributional styles for achievement and
interpersonal events suggest that it is important to consider attributional domain when
examining the ways men and women explain events. Thus it might be good for
researchers to clarify the nature of the relationship of societal factors to attributional
style.

The ﬁndings are generally consistent with Chodorow’s (1978) arguments and
with Rosenkrantz et al.’s (1968) work on stereotypes as well as with data about the labor
market. They suggest that societally deﬁned norms, roles, and stereotypes are affecting
cognition in a manner not usually considered. The ﬁndings also raise the question of
whether large scale societal changes, social programs, or political movements change the
way people think beyond the speciﬁc issues involved. The results of the present study

pertaining to sex differences and to attributional domains also suggest that it is important

65

for attributional theorists and researchers to take societal roles and norms, the existence
of attributional domains, and sex differences into consideration. It appears that what one
might consider a distal inﬂuence on attributional style could have a measurable impact.
Furthermore, these ﬁndings indicate that it would be useful for clinicians to consider
clients’ cultural roles and norms (especially if they differ from the mainstream) and their
relationship to clients’ attributional style. It may also be beneﬁcial for clinicians to attend
to the ways that a client’s attributional style does and does not ﬁt the culturally normative
pattern. For example, one might consider intervening differently with a woman whose
attributions for negative achievement events are more external, speciﬁc, and unstable
than her attributions for interpersonal events (the opposite of the normative pattern in this
sample) or with a man who makes internal, stable, and global attributions about

achievement matters (an atypical pattern in this sample).

Attributional Sgle and Family Factors
Young adult attributions are related to perceptions of parental attributions as

hypothesized. This is true both when considering overall young adult attributional style
(based on the entire Expanded Attributional Style Questionnaire (Peterson & Villanova,
1988)) and when considering more speciﬁc attributional domains (achievement,
interpersonal) for both the young adults and their perceptions of their parents. This result
is consistent with the hypothesis that young adults will use parents as models when trying
to explain events, especially novel ones, and thereby develop a similar attributional style.
This outcome is also consistent with earlier work examining overall attributional style
using these data (Monestere & Thornton, 1993, 1994). Additionally, the present ﬁndings
are congruent with writings (Haines et al., 1999) and research (Seligman et al., 1984) that
suggest a relationship between parental behavior, parental attributions, and children’s
attributional styles.

Although perceptions of mothers’ and fathers’ attributions are related positively

66

to both men’s and women’s attributions, the pattern of relationships differs for men and
women. Women’s attributions are informed by a more diverse set of factors when
considering both the sex of the perceived parent and the attributional domain. For
example, women’s attributions about the stability of events were related to perceptions of
both mother’s and father’s attributions about stability for interpersonal events and
perceptions of father’s stability attributions for the achievement domain. In contrast,
men’s attributions on the stability dimension were related to perceived paternal
attributions for interpersonal events and perceived maternal attributions for achievement
events on the same dimension. Looking across all three attributional dimensions, men’s
attributions appear to draw more equally ﬁ'om perceptions of mother and father.
Perceptions of father’s attributions have a more pervasive effect on young adult
attributions than perceptions of mother’s attributions. In general, when young adult
attributions were regressed on perceptions of both parents’ scores for the same dimension
and domain, perceptions of father’s attributions accounted for signiﬁcantly greater
amounts of variance than perceived maternal attributions. With one exception (intemality
for achievement events), women’s scores followed this pattern on all dimensions in both
attributional domains. This ﬁnding contradicted hypotheses. For men, perceived paternal
attributions only accounted for signiﬁcantly more variance than perceived maternal
attributions on the globality dimension for both attributional domains. Another indication
of the pervasive relationship of perceived paternal attributions to young adult attributions
comes from the structural equation modeling of the Phase I data. This analysis examined
the relationship of perceived maternal and paternal attributions for each attributional
dimension and domain to overall attributions (collapsing achievement, interpersonal, and
other event types) for the young adults. There are 10 signiﬁcant paths linking perceptions
of paternal attributions to young adult attributions but only 5 signiﬁcant paths linking
perceptions of mother’s attributions to young adult attributions. Perceived paternal

attributions appear to have the most pervasive impact on women’s overall attributions;

67

perceived paternal attributions for each attributional dimension and both attributional
domains are signiﬁcantly related to women’s overall attributions.

It was hypothesized that three factors might inﬂuence the magnitude of the
relationship between young adult attributional style and perceptions of parental
attributional style: availability (Bandura, 1977), symbolism (Chodorow, 1978; Parsons &
Bales, 1955), and sameness (Kagan, 1958). In formulating hypotheses about the
magnitude of the relationships between young adult attributions and perceptions of
parental attributions for both attributional domains, each of these factors was given equal
weight. For example, it was predicted that, because of sameness and symbolism, the
relationship of women’s attributions for interpersonal events to perceptions of maternal
attributions for interpersonal events would be signiﬁcantly greater than the same
relationship using perceived paternal attributional style. There was little support for these
hypotheses because they are rooted in the assumption that each factor would account for
an equal amount of variation in young adult attributional style. When considering all
three attributional dimensions in both attributional domains for both sexes, symbolism
(or what a parent represents) appears to be the predominant factor inﬂuencing the
relationship of young adults’ self-reported attributional styles to their conscious
description of parental explanatory styles. From the pattern of perceived paternal
attributions accounting for signiﬁcant variance in both women’s and men’s overall
attributions and their attributions for achievement and interpersonal events, one can infer
that symbolism is the organizing principle. More speciﬁcally, perceptions of paternal
attributions for achievement events most consistently account for a signiﬁcant amount of
variance in young adult overall attributions on all three dimensions. A father’s agentic
role may also explain the ﬁnding that perceived paternal attributional style for
interpersonal events accounts for a signiﬁcant amount of variation in women’s overall
attributional style and in women’s attributions for interpersonal events. Chodorow (1978)

has linked the father’s agentic role to his helping female children separate from the

68

mother and, in this way, the daughter may become more aware of how her father explains
interpersonal events. If sameness governed the relationship of perceived parental
attributions to young adult attributions, then perceptions of maternal attributions would
best account for women’s attributions, and perceptions of paternal attributions would best
account for men’s attributions. Instead, perceptions of father’s attributions are most
frequently related to women’s attributions and a balanced mix of perceived paternal and
maternal attributions are related to men’s attributions. If observational learning and
availability were the primary factors accounting for the relationship of young adult
attributions to perceptions of parental attributions, the pattern of results would show that
perceived maternal attributions best account for young adult attributions.

Although previous work (Monestere & Thornton, 1993, 1994) demonstrated a
relationship between perceived parental and young adult overall attributional styles, the
present work demonstrates the importance of attending to the sex of the perceived parent
and of the young adult, the attributional domain (achievement, interpersonal), and larger
familial and societal factors (e.g. men’s agentic role in and out of the home). This work is
consistent with the notion that cognitive models are important and can play a role in
understanding young adult attributional style. The results also indicate that societal
norms may be related to young adult attributional style through parents’ performance of
social roles. These ﬁndings suggest that it will be important for researchers studying the
development of attributional style to evaluate the possible roles of parents and of the
children’s perceptions of them. Furthermore, the results suggest that women and men
may incorporate perceptions of parental attributional styles in different ways. It appears
that males draw more evenly upon perceptions of fathers and mothers and that women
draw more on perceptions of paternal than maternal attributional style. Additionally,
women’s attributions have more relationships to perceived parental attributions (e. g.
based on parental sex, attributional dimension, and attributional domain). Perhaps

women, socialized to be aware of others’ thoughts and feelings (Chodorow, 1978;

69

Parsons & Bales, 1955), are better able to incorporate more perceptions of parental
attributional style. The ﬁndings also suggest that therapists working to modify young
adults’ attributions may beneﬁt from asking not only about how their parents would
perceive clients’ situations, but also about each parent individually and how she/he has
responded in Similar achievement or interpersonal situations. The pervasive impact of
perceptions of fathers’ attributions suggests the importance of attending to symbolism
and to societal factors that determine what is valued and how relationships are structured,

as they may ultimately impact young adult thinking.

Relationship of Young Adult an_d Perceived Parental Attributions to Attributions for and
Mood after Failure

One focus of the study was to clarify the relationship of young adult attributional
style and perceived parental attributional style to young adults’ attributions for failure
and their subsequent mood. Another focus was to clarify whether attributional domain
and identity status are related to attributions for and mood after failure. It was
hypothesized that young adult’s attributions for failure would mediate the relationship of
young adult attributional style to mood post failure. It was also argued that perceptions of
parental attributional style would be related to young adult attributions for failure
directly, and indirectly through young adult attributional style. It was proposed that
attributional style for achievement events would be related to attributions for failure. It
was also proposed that identity status would moderate the relationship of attributional
style and perceived parental attributional style to young adults’ attributions for failure.
Identity status, however, was omitted from the structural equation model since it was
unrelated to attributional style.

There was support for an SEM model examining the aforementioned relationships
utilizing perceived parental and young adult generality (stability + globality) scores for

achievement events. The results are consistent with current hypotheses, with Houston’s

70

(1995) work, and with Abramson et al.’s (1989) hypothesis that domain speciﬁc
attributional diatheses are more useful in understanding attributions for events and
subsequent mood.

Perceptions of both parents were hypothesized to relate to young adult
attributional style and to young adult attributions for failure. Both maternal and paternal
generality scores for achievement events were related to young adult generality
attributions for achievement events (in Phase H). However, only perceived maternal
attributions for generality predicted young adult attributions for failure on the anagrams
task. Young adult attributions for failure (a score based on intemality, stability, and
globality attributions) were, as hypothesized, positively related to scores on the visual
analog scale after failure indicating that more internal, stable, and global attributions
were associated with a down mood. Contrary to hypotheses, young adult failure
attributions did not mediate the relationship of young adult generality for achievement to
depressed mood post failure and did not support the causal mediation component of
learned helplessness theory (Abramson, et al., 1978). The lack of mediation contradicts
Metalsky et al.’s (1987) study that found mediation, but is consistent with Houston
(1995) who did not. The failure to ﬁnd mediation must be interpreted cautiously as there
may not have been sufﬁcient power to ﬁnd an effect.

These ﬁndings point to the potential importance of perceptions of parents for
understanding the emotional responses of young adults in failure situations. The pattern
of results suggests that young adults’ self-reported attributional styles reﬂect their
perceptions of both fathers’ and mothers’ attributions, but it is young adults’ perceptions
of how their mothers would explain achievement events that predicts their attributions for
actual failure and thereby subsequent mood. This ﬁnding might reﬂect the greater
availability of the mother as a model and it suggests that interventions to help children
develop optimistic explanatory styles might be strengthened by incorporating a

component addressing the primary caregiver’s attributional style. It also points to a gap

.71

between young adults’ self-reported attributional style using hypothetical situations and
what they think and do in reality. This discrepancy raises the question of why young
adults’ own attributional styles do not predict their behavior. It is possible that there was
inadequate power for the analysis to ﬁnd an effect given the sample size. It is also
possible that young adults in the sample are still separating and individuating from
parents and thus perceived maternal attributional style accounted for their attributions
after failing.

The discrepancy between the young adults’ attributional styles and their actual
attributions for failure suggests that it may be important to observe individuals in
multiple situations that they are then asked to explain. This might help to clarify whether
young adults’ self-reported attributional styles reﬂect the way they interpret situations as
they happen. It may also help to clarify the extent to which attributional style is a
diathesis for depression. ,

It is important to note that many ﬁndings were not signiﬁcant. Given the large
sample size in Phase I and the moderate reliability of the measures, it is unlikely that the
lack of ﬁndings was due to measurement difficulties. However, it is possible that the
limited sample size and the moderate reliability of the measures in Phase H, could
account for nonsigniﬁcant paths.

There are a number of limitations to consider when interpreting the results of this
research. First, there is the question of whether having only a small portion of the
individuals from Phase I complete Phase H limits the ability to generalize the results to
the larger sample and to other similar young adults. Several factors indicate that one can
generalize from the Phase H ﬁndings. Analyses suggest that the Phase H sample is
representative of the Phase I group; the groups did not differ on nearly all demographic
features. Additionally, the smaller sample appears to be due to factors that seem unlikely
to produce a sample that is atypical for college students. The exclusionary criteria

instituted to protect students eliminated some volunteers. The original design called for

72

25 individuals from the achievement or moratorium identity statuses and 25 individuals
from the diffusion or foreclosure identity statuses to be in each failure condition.
Additionally, each failure condition was to have nearly even numbers of men and
women. This meant that once a category in a cell (c. g. women in the achievement or
moratorium statuses) was ﬁlled, some volunteers could not be asked to participate.
Furthermore, the ﬁrst phase of the research was worth many research credits which
meant that many Phase I participants completed their research participation requirement
and did-not need to volunteer for Phase H. The second limitation is the reliance on
perceptions of parental attributions. There is a clear need to do research using actual
parental attributions. However, the present approach is a viable ﬁrst step and uses a
methodology (asking young adults to report on their parents and their experiences
growing up) that is common in developmental literature. A third limitation on the
generalizability of the ﬁndings reﬂects the passage of time. The data set is 1.0 years old
and it is possible that cultural norms and societal roles could be shifting somewhat and
limit the applicability of the ﬁndings. Finally, the use of college students limited the
developmental variation and range of depression exhibited in the sample.

In summary, the present study questions the assumptiOn that attributional theory
can be studied without considering broader contextual factors. The study suggests that
young adult explanatory style is related to perceptions of parental attributional style and
that the attributional domain affects the relationship. The study has identiﬁed sex
differences in attributional style and in the relationship of young adult attributional style
to perceived parental attributional style. The research suggests that societal roles and
norms affect young adult attributional style. Additionally, the study suggests that
perceptions of parental attributional style, especially perceptions of mothers, may play an
important role in understanding young adult response to failure.

Future work needs to investigate the relationship of actual parental attributional

style to young adult attributional style and to young adults’ perceptions of parental

73

attributional styles and how these relationships change over time. Studying the
relationship of actual parental attributional style to young adults’ perceptions of parental
attributional style will clarify the ways in which young adults’ perceptions or mental
models do and do not correspond to parents’ self-described attributional style.
Furthermore, it may clarify the value of perceptions, as opposed to actual parental
attributional style, for predicting young adult attributional style. Examining these
relationships over time as well as attributions for actual events may help determine
whether there is a developmental shift from interpreting events based on perceptions of
what one believes a parent would think to explaining events based on one’s own thoughts
about causality.

Given the limited sample size in Phase H, it will be important fOr ﬁrture work to
examine the relationship of young adult attributional style and perceived parental
attributional style to young adult attributions for actual failures. This future research
could examine sex differences and it could also examine developmental changes in what
. predicts an individual’s attributions for events. In this way, researchers could come to
understand what predicts response to failure and improve interventions designed to help
individuals cope with failure.

Future work might also attempt to clarify the relationship of attributional style to
societal norms and roles. More speciﬁcally, it might be fruitful to examine this
relationship by measureing the young adult’s adherence to traditional sex roles instead of
using sex differences. It might also be beneﬁcial to evaluate the extent to which the
dominant social norms were represented in the individual’s environment while growing
up. Articulating these relationships could enhance our understanding of the ways that

societies shape individuals.

74

APPENDIX A
Mernpging Missing Identiry Status Data
The classiﬁcation rules outlined in Adams et al. (1979) were used to determine
individual’s identity statuses. Cases with missing data were handled in the following
manner.

1) Cases missing total scores on more than one of the four identity status scales within a
realm (ideological, interpersonal) were omitted from analyses. Remaining cases were
treated in the manner described in step three.

2) Cases missing scOres on one identity status scale in each realm (ideological,
interpersonal) were allowed to remain and were considered in step three.

3) For each remaining case, the 8 items comprising the one scale from a realm that
could not be totaled were examined. Cases in which the scale was missing more than
one item were omitted from the analyses. For the remaining cases, the mean of the 7
answered items was computed and substituted for the missing item. Scale scores were

then recomputed and the individual’s identity status was determined.

75

APPENDIX B

Fi es

Sex

 

 

 

 

 

 

 

 

W v / [Amﬂ/

 

 

 

 

 

 

 

 

 

is“

Hypothesis: A and B should be larger than either C or D.

Key: I/E Internal / External
S/U Stable / Unstable
G/S Global / Speciﬁc
ACH Achievement Attributional Style
IN P Interpersonal Attributional Style
ID-ID Ideological Identity Status
IP-H) Interpersonal Identity Status

Figure 1

Latent variable multiple groups model showing relationships among attributional styles
and identity statuses varying by sex

76

IP-ID 1‘

 

 

Sex

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

iS/U

 

 

 

 

 

 

 

 

 

ID-H)
it I... ACH
S/U }
G/ S
PPAS-ACH PPAS-TNP
S/U G/ S

 

 

 

 

 

 

“+5
__)

Figure 2

 

. V

 

 

 

 

I/ElS/U

 

 

 

 

 

 

Latent variable multiple groups model showing relationships among perceived parental
attributional styles, young adult attributional styles and identity statuses by sex

77

 

 

PMAS-ACH

 

 

 

 

V

\ YAS-ACH

 

 

 

 

 

PPAS-ACH

 

 

 

 

 

A

Attribution

 

 

 

\ Visual
Analog

 

 

 

\

BDI

 

 

 

 

 

 

 

Figure 3

Model testing failure attribution as predicted by young adult and perceived parental

attributions for achievement

78

 

5803 was :08 Sm 3033028 5: $8333 bacoE
98 803333 :33 was?» 5826 8 mucusage 3:83 058% 59:26 33083 «c mama—mucus? 05 me .688 Bﬁmoﬁonmm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

e 2:5
9-9 a»:

bzﬁﬁm . £3885

£1366 <> <>
Sr

€320 £320 £38 3:58 £3825 £3835
mgmga 5422; 42332; 5432,. 2-32m :U<-m<2._

£366 £320 £58 £58 £38.85 Eases

Eim<mm IU<im<mm maimaodm IU<im<mm maimﬁmm EO<im<mm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

79

from 3:03:25? 3:23 333qu oEoonéREoc can moron garnets.“ :8ch 958 8m 6on 35m

 

 

 

   
  
 

bane—O
<>

 

 

 

N

€320
eeaﬁa

 

 

£386
:U<-m<2m

m 2%:

 

 

 

 

      

£39m
<>

 

 

 

 

 

 

 

‘

ﬁance—O
Evermedm

 

 

‘

£33m
:U<-m<2m

azssm

 

 

 

 

 

 

 

 

maimedm

 

 

 

 

 

‘

geese
5.3.4.8

 

 

 

 

£3825
E-m<2m

 

 

80

815 383353 3:23 338qu oEooamiﬁeEow use 33% 7803333 _Ego @5895 8m 3on 35m
A e 28E

 

 

 

 

 

 

 

 

 

 

 

 

 

 

         

 

 

 

 

 

 

 

 

 

h'
5:320 $.33 Eases
<> <> <>
2. L. ,N. am.
2.

£320 £820 £33 £53 £362.:
£320 £820 :3. :o< :3. ,2. £38 gases :3.
255,. 2-3%. -mSé as: -9»: -32.. 2.3.: 2.3.5 .9»:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

81

 

 

PMAS-ACH PPAS-ACH
Generality Generality

A

 

 

 

Y

YAS-ACH

t
D \ Generality

A

 

 

 

 

 

 

 

 

. Failure
l D Attribution

AA

 

 

 

DACL

 

 

 

 

 

 

 

 

 

Figure 7 '
Model testing failure attribution as predicted by young adult and perceived parental
generality attributions for achievement events ‘

82

 

 

PPAS-ACH PMAS-ACH

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Generality Generality
.53 .28
n.s.
.29
YAS-ACH
1
D \ Generality
n.s.
Failure
D Attribution
.29 n. s.
1 Visual t DACL
D Analog D \

 

 

 

 

 

Figure 8
Final model testing failure attributions as predicted by young adult and perceived parental
generality attributions for the achievement events

83

REFERENCES

Abramson, L.Y., Metalsky, G.I., & Alloy, LB. (1989). Hopelessness depression:
A theory-based subtype of depression. Psychological Review. 96. (2), 358-372.

Abramson, L.Y., Seligrnan, M.E.P., & Teasdale, JD. (1978). Learned
helplessness in humans: Critiques and reformulation. Journal of Abnormal Psychology,
81, (1), 49-74.

Abramson, L.Y., Alloy, L.B., & Metalsky, 6.1. (1986). The cognitive diathesis-
stress theories of depression: toward an adequate evaluation of the theories' validities. In
L.B. Alloy (Ed.), Cogpitive processes in depression. NY: Guilford Press.

Adams, G.R., & Fitch, SA. (1982). Ego state and identity status development: A
cross-sequential analysis. Journal of Personality and Sgial Psychology, 43, (3), 574-583.

Adams, G.R., Shea, J., & Fitch, SA. (1979). Toward the development of an
objective assessment of ego-identity status. Journal of Youth and Adolescence, 8, (2),
223-23 7.

Bandura, A. (1977). Social Learning Theory. Englewood Cliffs, NJ : Prentice-
Hall.

Baumgardner, A.H., & Arkin, RM. (1988). Affective state mediates causal
attributions for success and failure. Motivation and Emotion 12 (2), 99-111.

Beck, A.T., & Steer, RA. (1987). BDI: Beck Depression Inventog Mgpﬂ. New
York: The Psychological Corporation Harcourt Brace Jovanovich, Inc.

 

Beck, A.T., Steer, R.A., & Garbin, MG. (1987). Psychometric properties of the
Beck Depression Inventory: Twenty-ﬁve years of evaluation. Clim’cal Psychology
Review 8 (1), 77-100.

 

Beck, A.T., Ward, C.H., Mendelson, M., Mock, J., & Erbaugh, J. (1961). An
inventory measuring depression. Archives of Qeneral Psychiatry, 4, (6), 53-63.

Bentler, RM. (1988). Comparative ﬁt indexes in structural models. Psychological
Bulletin, (1990), 107, 23 8-246.

Bentler, P. M. (1995). EQS structural equations prom manual. Encino CA:
Multivariate Software, Inc.

Bentler, P.M., & Bonnet, D. G. (1980). Signiﬁcance tests and goodness of ﬁt in
the analysis of covariance structures. Psychological Bulletin, 88, 588-606.

Bentler, P.M. & Bonnett, D. G. (1987). This week’s citation classic (comments on
Bentler & Bonnet, 1980). Current Contents A & H, 9 (#37), 16.

Bemdt, S.M., Berndt, D.J., & Kaiser, CF. (1982). Attributional styles for
helplessness and depression: The importance of sex and situational context. Sex Roles 8
(4), 433-441.

 

Bilsker, D., Schiedel, D. & Marcia, J. (1988). Sex differences in identity status.
Sex Roles 18 (3/4), 231-236.

 

84

Boggiano, A.K. (1998). Maladaptive achievement patterns: A test of the

diathesis-stress analysis of helplessness. Jopm_al of Personaliiy and Social Psychology,
l4,_(6), 1681-1695.

Broverman, I.K., Broverman, D.M., Clarkson, F .E., Rosenkrantz, P.S., & Vogel,
SR. (1970). Sex-role stereotypes and clinical judgments of mental health. Journal of

Consulting and Clinical Psychology, 34, (1), 1-7.

Brown, J. (1984). Effects of induced mood on causal attributions for success and
failure. Motivation and Emotion 8 (4), 343-353.

Chodorow, N. (1978). The reprpgiuction of mothering: Psychoanalysis and the
sociology of gender. Berkeley, CA: University of California Press.

Cook, T. D., & Campbell, D. T. (1979). ngi-Exmrimentation: Desigp &
analysis issues for ﬁeld settings. Boston, MA: Houghton Mifﬂin Co.

 

Coté, J.E., & Levine C. (1988a). A Critical examination of the ego identity status
paradigm. Developmental Review, 8, 147-184.

Coté, J .E., & Levine, C. (1988b). On critiquing the identity status paradigm: A
rejoinder to Waterman. Developmental Review. 8. 209- 218.

Coyne, J.C., & Gotlib, I.H. (1983). The role of cognition in depression: A critical
appraisal. Pachological Bulletm,’ 94, 3, 472-505.

Crocker, J. (1981). Judgment of covariation by social perceivers. Psychological
Bulletin 90 (2), 272-292. .

 

Dixon, J.F., & Ahrens, AH. (1992). Stress and attributional style as predictors of
self-reported depression in children. o 'tive era and Researc l6 (6), 623-634.

Dweck, CS. (1975). The role of expectations and attribution in the alleviation of
learned helplessness. Journal of Personalig gpd Speial Psychology, 31, 674-685.

Erikson, EH. (1963). Childhood and sociery (2nd ed.). New York: W.W. Norton
& Company.

Feather, N.T., & Simon, J .G. (1971). Causal attributions for success and failure in
relation to expectations of success based upon selective or manipulative control. Jppnrai

of Personalig, 39, 527-541.

F emberger, SW. (1948). Persistence of stereotypes concerning sex differences.
Journal ef Abnormal and Social Psychology, 97-101.

F ollette, V.M., & Jacobson, NS. (1987). Importance of attributions as a predictor
of how people cope with failure. Journal of Personalig and Social Psycholpgy, 52, (6),
1205-1211.

 

Frank, S.J., Pirsch, L.A., & Wright, V.C. (1991). Late adolescents' perceptions of
their relationships with their parents: Relationships among deidealization, autonomy,
relatedness, and insecurity, and implications for adolescent adjustment and ego identity
status. Joumal of Youth and Adolescence 19 571-588.

 

85

Frank, S., Pirsch, L.A., & Wright, V.C., & Pedigo, T. (1990). Developmental and

Adaptational Implications of Young Adults’ Experiences of Autonomy from their
Parents. Unpublished manuscript.

 

Grotevant, H.D., & Adams, GR. (1984). Development of an objective measure to
assess ego identity in adolescence: Validation and replication. Journal of Youth and
Adolescence 13 (5), 419-438.

 

Haines, B., Metalsky, G., Cardamone, A., & Joiner, T. (1999). Interpersonal and
Cognitive Pathways into the Origins of Attributional Style: A Developmental
Perspective. In T. Joiner and J. Coyne (Eds), The Interactional Nature of Depression:
Advances in Intermrsonal Approaches (pp. 65-92). Washington, DC: American
Psychological Association.

Handal, P.J., Gist, D.C., & Wiener, R.L. (1987). The differential relationship
between attribution and depression for male and female college students. Sex Roles 16
(1/2), 83-88.

 

Hollingshead, AB. (1957). The two-factor index of social position. Unpublished
manuscript, Yale University, New Haven.

House, WC. (1976). Effect of locus of control, expectancy conﬁrrnation-
disconﬁrmation, and type of goal on causal attributions of failure. Journal of Research in

Personalig, 10, 279-292.

Houston, D.M. (1995). Surviving a failure: efficacy and a laboratory based test of

the hopelessness model of depression. Euroman Journal of Social Psychology, 25, 545-
558.

Hunter, S.C. (1986). The dictionary of anagrams (2nd ed.). New York: Routledge
and Kegan Paul.

Johnson, JG. (1992). Gender and mood as mediators of the relationship between
attributional style, daily events, depression symptoms, and hopelessness. Cogpitive

Therapy and Research, 16, (6), 687-697.

Johnston, M.A. & Page, S. (1991). Subject age and gender as predictors of life

stress, attributional style, and personal adjustment. Canadian Journal of Behavipural
Science 23 (4), 475-478.

 

Joiner, Jr., T.E., & Rudd, M..D. (1996). Toward a categorization of depression-
related psychological constructs. Cogm'tive Thery and Research, 20, (1), 51-68.

Kagan, J. (1958). The Concept of Identiﬁcation. Psychological Review 65, (5),
297-305.

Kagan, J. (1971). Personalng Development. (1. Janis, Ed.). New York: Harcourt-
Brace-Jovanovich.

Kaslow, N.J., Rehm, L.P., Pollack, S.L., & Siegel, A.W. (1988). Attributional
style and self-control behavior in depressed and nondepressed children and their parents.

Jomal of Abnormal Child £sychology, l6, (2), 163-175.

86

Kaslow, N.J., Rehm, L.P., & Siegel, A.W. (1984). Social-cognitive correlates of
depression in children. Journal of Abnormal _C_hild Psychology, 12, 605-620.

Kelley, H.H. (1967). Attribution theory in social psychology. In D. Levine (Ed.),

Nebraska Smmsium on Motivation (Volume 15). Lincoln, NE: University of Nebraska
Press.

Kitay, RM. (1940). A comparison of the sexes in their attitudes and beliefs about
women: A study of prestige groups. Sociomeg, 3, (4), 399-407.

Komarovsky, M. (1950). Functional analysis of sex roles. American Sociological
Review 15 (4), 508-516

 

Lazarus, RS. (1991). Emotion and Adaptation. NY: Oxford University Press.

Lazarus, RS. (1993). From psychological stress to the emotions: A history of
changing outlooks. Annual Review of Psychology, 44, 1-21.

Lewinsohn, P.M., Rohde, P., Seeley, J .R., Klein, D.N., & Gotlib, I.H. (2000).
Natmal course of adolescent major depressive disorder in a community sample:

Predictors of recurrence in young adults. American Journal of Psychiatry, 157, (10),
1 584-1 591 . ’

Loehlin, J .C. (1992). latent variable models. (2"d Ed.) Hillsdale, NJ: Erlbaum.

Lubin, B. (1965). Adjective checklists for the measurement of depression.
Archives of General chhiag, 12, 57-62.

Lubin, B. (1967). Manuel for the depression ecljective checklists. San Diego, CA:
Education and Industrial Testing Service.

Marcia, J .E. (1966). Development and validation of ego-identity status. low of
Personality and Socig Psychology, 3, (5), 551-558.

Marcia, J.E. (1967). Ego identity status: Relationship to change in self-esteem,
"general maladjustment", and authoritarianism. Journal of Personaliry, 35, (1), 118-133.

Marcoulides, G.A., & Schumacker, R.E. (Eds). (2001). New developments and
techniques in structural equation modeling. New Jersey: Lawrence Erlbaum Associates,
Inc..

Metalsky, G.I., Halberstadt, L.J., & Abramson, L.Y. (1987). Vulnerability to
depressive mood reactions: Toward a more powerful test of the diathesis-stress and
causal mediation components of the reformulated theory of depression. Journal of

Personality and Social Psychology, 52, (2), 386-191.

Metalsky, G.I., Joiner, Jr., T.E., Hardin, T.S., & Abramson, L.Y. (1993).
Depressive reactions to failure in a naturalistic setting: A test of the hopelessness and
self-esteem theories of depression. Journal of Abnormal Pyohplogy, 102, 101-109.

Mischel, W. (1970). Sex-Typing and Socialization. In Mussen, P., (Ed.),
Carrpri'cﬁel’s Manual of Child Psychqlogy (3rd ed.) (pp. 3-72). New York: John Wiley &

Sons.

87

Monestere, CM. (1993). The perceptions of parental attributional styles
questionnaire: Monestere’s modiﬁcation of Peterson & Villanova’s (1988) expanded
attributional style questionnaire. Unpublished measure for research purposes only.

Monestere, C.M., & Thornton, D.W. (1993). The Relationship of Identiry Status

and Perceptions of Parental Explanatog Sryles to Young Adult Attributional Srer and
A_.ttr1_°butions for Failure. Poster session presented at the annual meeting of the American

Psychological Association, Toronto.

 

Monestere, C. M., & Thornton, D.W. (1994, August). The Relationship of Family

Factors and Perceptions of Parental Attribptional Sgle to Young Adults’ Attributional
Sgles and Attributions for Failure. Poster session presented at the annual conference of

the American Psychological Association, Los Angeles.

Muthen, L. K., & Muthen, B. O. (1998). Mplus user’s gg'de. Los Angeles, CA:
Muthen & Muthen.

Nelson, L.J., & Cooper, J. (1997). Gender differences in children’s reactions to
success and failure with computers. Computers in Human Behavior, 13, (2), 247-267.

Parsons, T., & Bales, R. (with Olds, J ., Zelditch, M., & Slater, P.). (1955).
F ami y, Socialization and Interaction Brpcess. New York: The Free Press.

Peterson, C., Colvin, D., & Lin, EH. (1992). Explanatory style and helplessness.
Social Behavior and Pemnaliry, 20, (1), 1-14.

Peterson, C., & Seligrnan, M.E.P. (1984). Causal explanations as a risk factor for
depression: Theory and evidence. Pachological Review. 9L (3), 347-374.

Peterson, C.,.Seligman, M.E.P., & Valliant, GE. (1988). Pessimistic explanatory
style is a risk factor for physical illness: A thirty-ﬁve-year longitudinal study. Journal of

Personality and Social Psyehology, 55, (1), 23-27.

Peterson, C., & Villanova, P. (1988). An expanded attributional style
questionnaire. Journal of Abnormal Peychology, 97, (1), 87-89.

Rosenkrantz, P., Vogel, S., Bee, H., Broverman, I., & Broverman, D.M. (1968).
Sex-role stereotypes and self-concepts in college students. Journal of Cpnsulting and

Clinical Psychology, 32, (3), 287-295.

Rotter, J.B. (1966). Generalized expectancies for internal versus external control
of reinforcement. Psychological Monogmphs, 80, (1, Whole No.609).

Sarason, I.G., Johnson, I.H., & Siegel, J .M. (1978). Assessing the impact of life
changes: Development of the Life Experiences Survey. Journal of Consulting and

Clipical Psychology, 46, (5), 932-946.
SAS, Inc. (1999). SAS Software, Version 8.1. SAS Institute Inc., Cary, NC

SAS, Inc. (2000). SAS Software. SAS Institute Inc., Cary, NC.
Seligman, M.E.P. (1975). Helplessness: On depressiop, developmenr, and death.

San Francisco: W.H. Freeman.

88

Seligrnan, M.E.P. (1990). Learned optimism: How to change your mind and your
life. New York: Pocket Books.

Seligrnan, M.E.P., Abramson, L.Y., Semmel, A., & von Baeyer, C. (1979).
Depressive attributional style. Journ_al of Abnormal Psychology, 88, 242-247.

Seligrnan, M.E.P., Peterson, C., Kaslow, N.J., Tanenbarun, R.L., Alloy, L.B., &
Abramson, L.Y. (1984). Attributional style and depressive symptoms among children.

Journal of Abnormal Psychology, 93, (2), 235-238.

Steer, R.A., Beck, A.T., Brown, G., & Berchick, R.J. (1987). Self-reported
depressive symptoms differentiating major depression from dysthymic disorders. Journal

of Clinical Psychology, 43, 246-250.

Stevens, J. (1992) Applied multivariate statistics for the social sciences. Hillsdale
NJ: Erlbaum.

US. Census Bureau. (2000). Statistical abstract of the United Sgtes: 2000. (120th
ed.). Washington, DC: Author.

Ward, L.G., Friedlander, M.L., & Silverman, W.K. (1987). Children’s depressive
symptoms, negative self-statements, and causal attributions for success and failure.

Cogm'tive Theogr and Research 11, (2), 215-227.

Waterman, AS. (1982). Identity development from adolescence to adulthood: An
extension of theory and a review of research. Developmental Psycholpgy, 18, 341-358.

Waterman, AS. (1988). Identity status theory and Erickson’s theory:
communalities [sic] and differences. Developmental Review, 8, 185-208.

Waterman, A. (1999). Issues of Identity Formation Revisited: United States and
The Netherlands. Developmental Review. 12 463-479.

Waterman, A.S., Geary, P.S., & Waterman, CK. (1974). Longitudinal Study of
Changes in Ego Identity Status from the Freshman to the Senior Year at College.

Developmental Psychology, 10, (3), 387-392.

Weiner, B., Frieze, I.H., Kukla, A., Reed, L., Rest, S., & Rosenbaum, RM.
(1971). Perceiving the ca_u.ses of success and failure. Morristown, N.J.: General Learning
Press.

Weiner, B., Heckhausen, H., Meyer, W.U., & Cook, RC. (1972). Causal
ascriptions and achievement behavior: The conceptual analysis of effort. Journal of

Personaliry and Social P§ychology, 21, 239-248.

Weiner, B., Russell, D., & Lerman, D. (1979). The cognition-emotion process in

achievement-related contexts. Journal of Personalip; and Social Psychology, 37, 1211-
1220.

Whitley, Jr., B.E., Michael, S.T., & Tremont, G. (1991). Testing for sex
differences in the relationship between attributional style and depression. Sex Roles 24
(11/12), 753-758.

 

Zuckerman, M., & Allison, SN. (1976). An objective measure of fear of success:

89

Construction and validation. Journal of Persorralip/ Assessment. 40. (4), 422-430.

 

9O