You are on page 1of 20

A Longitudinal Study of the Hopelessness Theuty of

‫„؛؛]؛؛‬: Testing‫ ؟؛‬DiathesisStjess Model wittlin

Paul Kwon
Washington State University

Jean.Philippe Laurenceau
University ‫؛‬oj Miami

A ten-week prospective longitudinal study was conducted to test the


diathesis-stress component 0< the hopelessness theory and to test whether
negative attributional style leads to an increased exposure to stressors.
Participants completed initial measures of attributional style and depres-
sive symptoms tollowed by weekly assessments ot depressive symptoms
and daily hassles. Consistent with the diathesis-stress hypothesis. Hierar-
chical binear Modeling revealed that attributional style moderated ttie impact
of daily hassles on depressive symptoms. Negative attributional style pre-
dieted greater depressive symptom reactivity in response to stress. The
results also indicated that attributional style was not predictive of the
number of subsequent daily hassles. Thus, the data were supportive of a
differential reactivity to stress model, but not supportive of a differential
exposure to stress model. Implications for understanding the mechanisms
involved in the hopelessness tfieory are discussed. © 2002 Wiley Peri-
odicals. Inc. ‫ل‬Cl'in Psychol 58: 1305-1321. 2002.
Keywords: depression: attributional style; diathesis-stress: hopelessness
theory: differential reactivity; differential exposure

Tilts study was supported ill part by a grant from the College of Liberal Arts iit Washington State University
awarded to the first author and a James w. MeLamore Aw’ard from the Office of Researcfi Administration ‫؛‬It the
fJniversity of Miami awarded to the second author. We tliank Kirk Barney. Nathan Causey, Marsliall Clark.
Andrea Devereaux. Pamela Freske. Ryan Sillers. Elizabeth Tiimura, and Jennifer Williams for assistance witli
data collectitm and entry. We also thank Adele Hayes ‫؛‬ind Duncan Campbell for lielpful comments on a previous
draft of the article.
Correspondence concerning this article should be ‫؛‬iddressed to: Paul Kwon. Department of Psycliology. P.O.
Box 64482(). Washington State University. Pullman, Washington 99164-4820; e-m،‫؛‬il: kwonp@wsu.edu;
http://www.wsu.edu/-kwonp.

JOURNAL OF CLINICAL PSYCHOLOGY. Vol. 58(101. 1305-1321 (2002) © 2002 Wiley Periodicals. Inc.
Published online in Wiley InterScience (www.interscience.wiley.cofn). 001: 10 1002/jclp.10043
1.1١ Jt١i،r,٠ol of Clinical Psychology. October 2002

Tlie hopelessness theory of depression (Abramson. Metalsky. & Alloy. 1989: formerly
،lie learned helplessness theory of depression: Abramson, Seligman. & Teasdale. 1978)
posits that a negative attributions! style serves as a dittthesis for depression. Lfnder suf-
fieient levels of life stress, individuals with a tendeney to make global and stitble causal
attributions for negative events itre theoretically iit greater risk for depression. Tliis pro-
posal has generate،! mueli research over the liist severitl decades. Historically, the diatliesis-
stress component of the theory has proven، lifl’icult to test adequately (Abramson. Alloy,
& Metalsky, 1988: Robins & Hayes. 1995). A ،'ter many years ol' mixed ،'indings (.Persons
& Mirandit, 1992: Robins & Hayes, 1995). recent research lias y.ielded more convincing
support for tliis aspect of tlie theory (Alloy et al. ١1999: Alloy & Clements, 1998: Alloy.
Just, & Panzarella, 1997: Metalsky & .loiner, 1992, 1997').
Tlie first goal ،if the present study was to test l'urther the diathesis-stress aspect (i.e.,
differentii،) reactivity model) of tlie htipelessness theory. The secontl goal of this study
was to test a differential exposure to stress Iiitidel. an alternate niotlel that examitied
W'hether attributional style increases an individual's subsequent exposure to stressfirs. To
accomplish tliese goals witli methodological rigtir, we ctillected data using a prospective,
multiwave longitudinal design. A hierarcliiciil linear modeling approach was tlien Lisetl to
capitalize tipon tlie richness ol'tliese data. In examining the diatliesis-stress component of
tlie htipelessness theory, this statistical technique iillowetl US to (a) motlel tlie relatit-in
between stresstirs and depressive sympttmis for each study participant and (.b) investigate
the role til' attributional style as a between-subjects predicttir of tlie relation between
stress atid symptoms. Given that conventional study designs typically omit tlie ftirmer
step in data aiiitlysis, we believe tliat tlie present study design allows for a powerful and
novel test of the study hypotheses.

The Hopelessness Theory of Depression


Consider tlie example of an individual wlio hits been fired from a .job. A negative attri-
bution iti tliis situation miglit be "1 lack tlie ability It) be successl'ul in my career.’' Such an
attribution involves generalizations about tlie cause ،if the negative event, in that a lack o،'
ability is botli stable (i.e. ١persists over time) and global (i.e.. at'fects many situations). Ati
example of a positive attribution, in contrast, woultl be “I was in a company that ditl not
best suit me.” Acctirding to the liopelessness tlieory ،!،' depression (Abramstm et al.,
1989), when face،! with stress, an individual wlio lias a clironic tendency toward making
negative attributions is al greater risk for depression compared with an individual who
tends to make positive attributions.
Early wtirk on attributional models o،' ،lepressitm has consistently demonstrated a
concurrent reliition between attributional style and depression (see reviews by' Biirnett &
(Jotlib. 1988; Peterson &Seligman, 1984: Robins & Hayes. 1995: Sweeney. Anderson. &
Bailey. 1986). In additioti. it lias been ،'ound tliat individuals with a negative, attributional
style have a greiiter likelihtiod ،if having a history o،' past depression compared with
indiv’iduals witli 1‫ ؛‬po.sitive attributional style (Alloy et al.. 1999'). However, this concitr-
rent and retrospective evidence for a relation between attribution،-،! style and depression
does not necessarily support tlie diathesis-stress nature of tlie model. According to a
،liathesis-stress explanation, attributional style is ،،stitble vulnerability factor that is a
predictor of changes in depression. An alterniitive explaniition that also explains the find-
ings is til ،It attributional style is a concomitant ،ir ctinsequence, ratlier than an antecedent,
of depression. In ،'act. eiirlier reviews of the hopelessness theory literiiture coticluded that
tlie bulk of tlie research sujiported tliis altertiative explanation over a diathesis-stress
explanation (Barnett & Gotlib. 1988: Coyne & Gotlib, 1985: Persons & Miranda. 1992').
Hopelessness Theory of Depression 1307

Such an explanation is viable given that an increiise in negative cognitions is a known


symptom (i.e., concomitant or consequence) of depression.
For a study to demonstrate whether atiributional style is a predictor of depression,
tlie study needs to be prospective and longitudinal in design. Tlius, tile diathesis-stress
component of the hopelessness theory would be supported if bitseline levels of attribu-
tional style predicted future changes in depressive symptomatology. More specifically,
initiill attributiotial style should interact with subseque.it stress in accounting for sub-
setjuent' clianges in depression.

Status of tlie Diatliesis-Stress Component of tlie Hopelessness Theory


Prospective tlesigns have only recently been used in testing the hopelessness theory.
Preliminary reviews of this work concluded tliat tlie results were botli modest and incon-
sistent in tlieir support for tlie diatliesis-stress model (Persons & Miranda, 1992; Robins
& Hayes, 1995). Tlie initiitl studies witli tlie Jiiost positive results examine،, utidcrgradu-
ates' mood reactions to ptior exaniinati،).! performance. Altliougli several of tliese studies
foutid an interaction lietween attributioniil style atid examitiittion stress in predicting mood
reactions, these fitidings were often transient anti inconsistent. For example, two studies
1'ound tliat tliis interaction pretlicted mood reactions two days after receipt ofexiimination
grades, but not on tlie days before or after tliis point ( Metitlsky, Halberstadt, & Abramson,
1987; Metalsky. Joiner, l-lardin. & Abramson. 1995). Houston (1995) found tliat the Attri-
butional Style X Exatn Stress interaction predicted satne-day reactions in one study, but
not in itntitlier study. Finally, tliree other studies fountl that tliis ititeraclioti did not predict
same-day reactions (Follette & Jacobson, 1987; Hunsley, 1989; Ralpli & Mineka, 1998).
A more recent methodology used in tliis area involves a prospective ،lesign in exatn-
ining an indivitlual's depressive reactitins to aggregate life stress. In investigatifig overall
levels 0،' depressive sytiiptonis ittid stress, tliis design goes beyond tlie investigation of
specific stressors and circumscribed reactions to those stressors. These studies have uti-
lized two-panel Itingitudinal tlesigns assessing attributional sty-le at Time I. and dejires-
sive symptoms at btitli 1'inie I anti Time 2. Also at Time 2, life stress Wits itssesse.d for the
period between Time 1 and Titne 2. Tliere is evidence 1'roni Five stutlies tliat tlie inter-
actitm between attributitinal style anti life stress is predictive of changes in depressive
symptomatology (Alltiy et ill., 1999; Alloy & Clements, 1998; Alloy et ill., 1997; Metal-
sky & Jtiiner, 1992, 1997).
In addition, support for tlie diathesis-stress model liits ctinie from the Temple-
Wisconsin Cognitive Vulnerability to Depression project (Alloy et ill., 1999). This project
is it two-site study thilt uses a prospective behavioral liigli-risk design. Among partici-
[lants witli no history of past depressitm, individuills witli it negittive iittributioniil style
were more likely to experience it subsequent first tinset ol. clinically significant major
depression ctmiparetl with individuals with a ptisitive attributional style ( ٠17 ١'S. 1%, Alloy
et ill., 1999, P.5II). Furthermore, among participants witli it Itislttry tif past depression,
intlivitluitls with a negittive iittributioniil style were nitire likely t( ١experience ii recurretice
ofclinicitlly significant major depression compared witli intlividuills with a positive attri-
butional style (27 vs. 6%. Alloy et al., 1999. p. 511).
Ill summary, an increase in attention lt ١methodological issues lias led tti the recent
use of prospective designs in testing tlie liopelessttess theory of depression. This improve-
nient in methodology lias revealed nitire consistent support for (lie. diathesis-stress coni-
ponent of the model. We ntiw proptise furtlier improvements in tlie methodology !١،such
prospective studies.
1308 Journal of Clinical Psychology. October 2002

Methodologies Issues in Testing a Diathesis-Stress Model


Most of the prospective studies in this area have incorporiited two waves of assessment.
Examining change between two points in titiie is not the most effective and reliable way
of assessifig change and examining correliites of. change over time (Willett. 1997). Spe-
cifically. tlie use of difference scores (i.e., examining pre-lo-post change) to investigate
correlates of. change confounds measurement error with true score error, leading to un-
،lerestimates of the relation between change and predictors of cliange (Willett. 1989).
Methodologists have recotnmended tliat to capture within-subjects change, a longitudi-
nal approach should utilize multiple waves of assessments conducted at sensibly spaced
intervals. Moreover, it has been noted that a drastic increase in statistical power can
result from increasing tlie number of waves of data ctjllection (e.g.. Willett. Singer. &
Martin. 1998).
An additional issue in investigating diathesis-stress models involves selecting the
appropriate level of analysis, specifically, researchers examining sucli model.s are in tlie
position t)f combining data generated t'rom two levels. The within-subjects level of analy-
sis focuses on the impact of multiple stresses on level of depression for a particular
individtial. The between-subjects level ofanitlysis examines whetlier levels ofattribulional
style are related to average levels of depression across individuals. The diathesis-stress
component of the hopelessness theory of depression predicts a significant cross-level
interaction between attributional sty'le (between-subjects) and stress (within-subject) in
explaining depressive symptoms. Typical data analytic approaches, such as regression or
repeated measures ANOVA, cannot simultaneously model the variation tliat takes place at
the within- ،٤nd between-subjects levels. Re.searchers using such traditional analytic
approaches are forced to examine predictions of the hopelessness model of depression by
eliminating one level of analysis, usually via aggregation ol.data from the within-subjects
level of aniilysis. This process leads to the loss of important information, such as the
unique reliition between stress and depressive symptoms for each individual. A specific
class of statistical model, known generally as Ilierarchical or multilevel modeling (Bryk
& Raudenbush, 1992: Goltlstein, 1987; Rogosa, Brandt. & Zimowski, 1982), can be
used by methodologists to analyze data that ctjme from multiple levels of assessment and
contain variation at these multiple levels.
In the otily study in this area usijig a hierarchical linear modeling approach witli
tnultiwave data. Swendsen (1998) utilized an experience sampling method to study par-
ticipants over a one-week period. Participants were asked to record tlieir mood and stress-
ors on a daily basis during the week. Swendsen f.ailed to find a significant interaction
between attribtitional style and stress in pretlicting fiuctuations in depressed mood.
We believe that the ,nultiwave design used by Swendsen is an important advance in
testing the ltopelessness theory. A possible explanation for tlie null findings is the use of
daily assessments, which may have been too short to delect a significant relation between
the Attributional Style X Stress interaction and mood. The selection of spacing between
.measurements in time for assessing tlie nature of change in an outcome variable is an
important consideration (Willett, 1989, 1997). As noted earlier, Metalsky and colleagues
(Metalsky et al. ١1987; Metalsky et al., 1993) found that the Attributional Style X Exam
Stress interaction predicted mood reactions two days after receipt of grades, but not
earlier. Based on tliese findings, tliey proposed that there may be a lag in the mood
reaction to negative events. Thus, the use of weekly assessments may be able to better
capture participants’ reactions to negative events. Weekly assessments also would allow-
for the exatmination of changes in depressive symptoms, rather than changes in depressed
mood, as prescribed by the hopelessness theory of depression.
Hopelessness Theory of Depression 1309

Differential Reactivity and Differential Exposure Models

of the Stress Process


111 a model of studying personality in the stress process. Bolger iind Zuckerman (1995.)
proposed that personality may iiTipact the stress process iti two distinct, hut completnen-
tary. ways. First, tlie differential reactivity model proposes tliat personality may moderate
the effects of stressful events on mood. Second, the differential exposure model proposes
tltitt personality may lead to iitcreased exposure to stressors. Figure 1 depicts tlie expo-
sure ،‫؛‬nd reactivity components of these alternative models for the relations among neg-
ative atlrihutional style, stress, atitl depressive symptomatology. One. hoth. or neitlicr of
these models titay accurately capture the process through which atlrihutional style and
stress can influence mood.
The hopelessness theory of depression is implicitly a differential reactivity mode-1.
Specifically, the hopelessness theory proposes that atlrihutional style moderates tlie
relation between stress attd depressive symptomatology, with negtttive attributions! style
leading t() greater vulnerability to stress. For example, a negative attribution for a failed
romantic date, such as "nobody wants to be with me,” may interact with the stressor to
lead to a depressive reaction. Accortling to the liopelessness theory, negative iittribu-
tional style leads to depression when a threshold of stress lias been exceeded. In other
words, it can be view'ed as a titration model, in that a negative attributional style is
posited to lower the threshold of stress required to produce depression (Abramson
et al., 1989).
The differential reactivity perspective leads US to anotlier possibility: Negative attri-
butional style may lead to. in general, greater depressive reactions to each stressor. This

Figure /. Differential exposure and differential reaetivity models illustrating possible relations among :،ttri-
bution،‫؛‬. style, stress, and depressive symptoms.
/.?/٠ Journal of Clinical Psychology', October 201)2

heightened reactivity to stressors may lead to noticeable depressive symptoms if enough


stressors cumtilate over a period of time, 'riiis mechanism would be consistent with tlte
diathesis-stress perspective. It would however, suggest a specific mechanistn of action
that is not specifietl by the hopelessness theory. ‫ اوا‬otlier words, depressive response to
stress may not be a sutlden, categorical response. Rather, it may be a continuous depres-
sive response that may build to significant depressive symptoms at a certain level of
stress. 1.1 the present study, w'e investigitted wlietlie.r negative attributioniil style would
pretlict depressive responses to stress in tliis matiner.
An additional m١،del proposes tliilt individual differences in altributional style are
related to differential fretjuency t١l ٠exposure to stressors. As noted by Bolger and Zuck-
erman (1995), however, tlie differential expostire model is often overlooked in diiltliesis-
stress theories sucli its the liopelessness theory. Ill other words, the hopelessness theory
does not ilddress tlie issue of whether negiitive attributional style increases one's sub-
sequent exposure to stressors. Investigations of tlie stress generiition process in depres-
sion have revealed an analogous l٠itiding tliilt tlepressed individuals tetid to generate more
stress titan nondepressed individuills (Davila, Hammen, Burge, Paley, & Daley. 1995;
Hiimmen. 1991). It remains to be stutlied W'lietlier tiegative attributional style contribtttes
to tlie stress-generation process in depressed individuals.
From a differential exposure perspective, it could be the case thilt a tiegative attribu-
tiotial style may leild to certain behaviors or itttilutles tliat then result in exposure to
stressors. For instance, a negative attribution about a romantic ،late (e.g.. "nobody wants
to be with me") tiiay induce hopelessness itbout future dates. Tliis liopelessness could
then lead to tiiotivatiotial dil.ficulties, wilieh could in turn lead to less effort spent on
reliitiotisilips. Tliis diniinislied effort iti relating t ١،otliers could tlien lead tti greater inter-
personal stressors in tlie future. Analogous to this example, ititerpersonal tlieorists liave
propt.sed that depressed indivitluals behave interpersonally in a manner that elicits .leg-
ative reactions, wilieh leads to a downward spiral of further depression (see rev'iew' by
Lara & Klein, 1999). As represented iti Figure I, tlie exposure component posits a niedi-
ational liypothesis regarding tlie effects of attributional style oti stress and depressiv.e
symptoms. In other W'ortls. greiiter negative iittribulional style leads to increased stress-
ors, wilieh tlien leads to greater depressive symptomatology.
Comparing tlie differential reactivity model witli the differential exposure model i .1
the context of the liojielessness tlieory of depression raises imporlitnt theoretic ،‫ !؛‬ques-
tions. Does negative attributional style increase one's depressive reactions to stressors‫؟‬
Linder this tlifferenti،‫؛‬. reactivity model, negative attributional style represents a passive
vulnerability tliat increases one’s sensitivity to stress, but does not generate tlie stressful
situations tliat lead to tlepression. Or, tloes negative attributional style le ،‫؛‬d to a generation
of stressful evetits in one's environment‫ ؟‬Under tliis differential exposure model, tiega-
tive atiributioti،‫ !؛‬style rejiresents ati iictive vulnerability fiictor tliat elicits stressful cir-
cumstances that leatl to depression.
Otlier mtidels ofetignitive and personality vulnerabilities have found increased sup-
port for tile differential exposure model. Ftir example, ،‫؛‬n excessive reassurance-seeking
style h،‫؛‬s been fotind to prompt negative interpersonal reactions from (ntlners. initiating
the stress-generation process in depression (Pottlnoff. Holahan, & Joiner, 1995). In ad-
dition, intern،‫ !؛‬cognitive processes can generate stressful circumstances as well. For
insliince, self-critical individuals have been slnown to report greater stress and lower
social support ،nver time (Priel & Shaliar. 2000). One mechanism tli،‫؛‬t may be involved
in tliis l.inding is thilt self-critical individuals make fewet. requests for support 1’rotn
peers, wlio subsequetntly find sucli individuals less expressive ،‫؛‬nd more difficult to get
to know (Mongrain, I99X). In a similar vein, it coultl be tliilt ‫؛‬nil individuill consistently
Hopelessness Theory oj Depression Ull

making negative attributions may set themselves up lor failure. Evaluating both the
differential reaetivity and differential exposure models within tlie same study would
begi.1 addressing tliese important questions regarding the role of attributional style in
the stress-depression process.

Overview and Predictions f ،r tlie Present Study١


Tlie first purpose of this study was to use a multiwave design to test wlietlier attributional
style is predictive of subsequent depressive reactiv'ity to stressors, consistent with tlie
diathesis-stress perspective of tlie hopelessness theory of depression. The second purpose
was to test a differential exposure model in which higher levels ot' negative attributional
style leads to a greater expositre to stressors. Participants initially completed a measure of
attributional style. For tlie next ten weeks, they completed weekly measures of daily
stressors and depressive symptoms.
The dal‫؛‬، collected in lliis study conformed to a multilev.el data structure (Bryk &
Raudenbush. 1992: Goltlstein. 1987: Kenny, Kashy. & Bolger. 1998). Multilevel data
contain a hierarchy t)f observations with lower level (within-subjects) and ujiper level
(between-subjects) data units. In tlie current study, the weekly ratings of depressive
symptoms iind d‫؛‬،ily liassles served ‫؛‬IS the lower level data. '1'lius. tliere were multiple
measures ordered in time witilin each individual in the study, allowing US to estimate
the reliitionsilip betweeji stressors atid depressive symptomatology (i.e., reactivity') for
eacli participant. The upper level data consisted ol. each individual's getider and attribu-
tional style assessed at the beginning of the study. As a result of tlie multilevel nature of
lata analytic strategy and focused،the data, hierarchical linear modeling was used as the
0.1 testing tlie relationsilip between attributional style ‫؛‬،nil individual differences in
reactivity.
Tlie present stutly is novel in tliis area because ot' its use of hierarchical linear mod-
eling in testing tlie lit of the hopelessness tlieory into tlie differenti ‫؛‬،! reactivity/exposure
framework. Given that tliis is ‫؛‬111 initial study in this area, an analogue sample was chosen
'urtlier،as the focus of study. Positive results in tlie current investigation would encourage
in clinical samples.،W'ork
'depression and tlie dil'I'erential reactivity nitidel,،!،Based on tlie hopelessness theory
'the study،!،inset ،(iur first liypotliesis was that negative attributional style assessed at tlie
would be predictive ol' depressive reactions to subsequent stressors. We also testetl tlie
differential exposure nitidel, which pretlicts that negative attributional style measured ill
the onset of the study would prospectively predict ‫ '؛‬.stressors over the،)،1 greater number
course of' tlie study,

Metliod
Participants
l 30 niiile) untlergraduate stutlents enrolled in iin،Piirticipants were 146(116 female an
iurse credit،introductory psychology course at a large university. Participants received c
' the piirticipants was 20.1 yeiirs (SI) = 2.8). A،!،'or tlieir participatitin. Tlie mean age ،
' tlie participants identil'ied themselves as Caucasian،!،majority )81‫؟‬،(. 'rtini par-،Data
! in the analyses if they completed،ticipiints were include ‫؛‬1‫ ؛‬least seven o.'the ten weekly
' an initial pool (if 167 participants who completed tlie initial ques-،juestionnaires. 0،
)t differ sign،litl not meet tliis requirement. Excluded participants did n،lionnilire, 21 ‫؛‬،'-
'rom tlie rest ol' tlie Siimple on eitlier tlie Measure of Attributional Style (MAS،icantly :
I312 2002 Clinical Psychology, October ‫ آال‬1‫اال‬،‫اا‬١،‫ا‬

Kwon & Whisnian. 1.65( = 165(, /)992‫ا‬, I1.S. ١or the Beck Repression Inventory-!
(BR1-IL Beck, Steer, & Brown, 1996).11( = .165‫ )ر‬. n.s. Among tlie llnnl sample of 146
participants, the vast majority (125.) completed all questionnaires. Tilirteen participants
missed one of the weekly, questionnaires, 5 missed two questionnaires, and 3 missed
three questionnaires.

Measures

The MAS (Kwon & w'hisman. 1992) presents piirticipants witli 26 hypothetical negative
life events (e.g., "You liave been looking for a job unsuccessfully for several weeks"). For
each situation, tlie participant is asked to vividly itiiagine the scetiario and tlien choose
the most likely cause of tlie situation atuong four alternatives, W'liicli are crossed 2x2 oti
dimensions of internality and generality. To illustrate, tlie clioice “You tlo not have much
to offer as a job applicant" would be high on botli dimensions whereas tlie clioice "Btisi-
fiess lias beeti unusually slow in your town tlie past few weeks” would be lo ١١on botli
dimensions. Tlie generality dimension incorporates both globalily and stability, which
both appear tti be measures of catastrophizing (Peterson & Seligman, 1985), iire liighly
correlated (Peterson & Villanova. 1988), itnd have been grouped together in tlie hope-
lessness theory (Abramson et al. ١1989.). A score for the generality' dimension is derived
by counting tlie number tif selected items tliitt incorporate the dimension (range = 0-26).‫ا‬
A previous study has indicated tliilt the MAS Iiiity be nitire predictive of longitudiiiiil
changes in depressive symptoms thill! the Expanded Attributional Style Questionnaire
(EASQ: Peterson & Villanova, 1988). Kwtin atid Whisnian (1992) ctinducted a prospcc-
tive three-month stutly of. 80 students using botli the MAS and EASQ. The EASQ failed
to predict longitudinal clianges in depressive symptoms tiver tlie course of tlie sttidy
eitlier as a main el'fect or interaction with life stress. In contrast, the MAS significantly
interiicted witli life stress to predict chitnges in depressive symptoms. Tlie MAS ،‫؛‬Iso has
been successfully tisetl in two other studies involving attributional style (Kwon. ‫؛‬1999
Kwon & Lemon, 2000). Kwon anti Whisnian (1992) found a tliree-tiiontli test-retest reli-
ability of .82 for tlie MAS generality dimension, ،‫؛‬ldicating excellent stability over time.
In tlie present study, coef'ficient alpli ،‫؛‬was .67.
Tlie BDI II (Beck. Steer. & Brown, 1996) is a 21-itetu, self-report instrument tif
depressive symptomatology. Each item is sctired 1'rtim 0 to 3, with a liiglier score intli-
eating greater sytuptotu severity. Given thilt assesstnents took pl ،٤ce on ،‫؛‬weekly basis,
the instructions were tuodified so that participants reported <١n their synipto.us during the
past week rather tlian two weeks. Tliis second edition ol' the instrument provides com-
preliensive coverage of DSM-IV (American Psychiatric Association, 1994) diagnostic
criteria. The BD1-11 inclutles new items iissessing incretises in iippetite, weiglit, and sleep‫؛‬
the first edition (Beck. Ward, Mendelsoji. Mtick. & Erbaugh, 1961) assessed (Itily decreases
in these symptoms. The BDI-I1 lias been shown to be a reliable and valid nietisure ol.
depressive symptomatology (Rozois, Rtibson. & Alinberg, 1998). 1.1 tlie present study,
coefficient alph ،‫؛‬was calculated ،‫؛‬t each titue period, with values ranging from .84 to .90.
Tlie Hassles and Uplifts Scale (.ReLojigis, Folkinan, & Lazarus. 1988) is a tjuestion-
Iiaire of. daily stressors. Subjects were given ،‫؛‬list of 53 aspects ol. living (e.g., meeting
'The hopelessness theory of depression no longer ineorporates tlie intern ،‫ !؛‬dimension in Its conceptualization
ofa depressogenic attributional style (Abramson el 1989‫؛‬،!.. ), consistent witli studies intlicaling tliat an internal
attribution to negative events can at limes increase, rather than decrease, one's coping ability (e.g.. Dwcck &
l.icht. 1980: Janoff Butman, 1979.). Tlius. the internal dimension of attributional style will not be discussed in
this article.
Hopelessness Theory of Depression 1313

deadlines or goals on tlie job) and were asked lo indicate whether eaeli item was a hassle
during tlie previous week, and if so, to what degree using a rati.ig of () (none or not
applicable), I (somewhat), 2 (quite a bit), or ٠١(a great deal). Tlie uplif'ts section oftfie
questionnaire was not administered. A Jiieasure tifthe number of daily liassles (DHS) was
derived by counting tlie number of items intlicated as a liiissle (rated 1 or higher). Sub-
.jects’ weights of the items, wliich may be confounded witli depressive symptoms, were
not incorporated in the L)H score 1(1 reduce subjectivity in tliis measure. Itejiis indicating
symptotiis of' psychological or physical disorders were excluded by tlie scale authors
(DeLongis et al. ١1988). In tlie present study, coefficient alplia was calctilated iit eacli
time period, with values riinging from .81 to .91.
Means and standard deviations r١f study questionnaires iire presented in Table I.

Procedure

Participants came in for iin initial session in groups of five to eiglit. A reseitrcli assistiint
administered an initial questionnaire tliat included demographic questions iind the MAS.
Tlie particijnants also were given a weekly questionnaire and ten ‫؛‬uiswer sheets. Tlie
weekly questionnaire consisted of the B 1)1-11 anti the DHS. Tlie research assistant instructed

fable I
Means and Standard Deviations
0.1 Study Questionnaires
Variable Mean o Range

MAS
Baseline 606 3.51
BD.-II
Week ‫ا‬ 1(1.70 7.46 0-28
Week 2 8.43 6.91 0-42
Week 2 10.07 8.57 0-44
Week .1 ٦.2.٦ 6.98 0-30
Week 4 7.24 7.01 0-37
Week 5 7.18 6.52 0-30
Week 6 0.26 6 42 0-32
Week 7 0.30 5.83 (1-25
Week 8 5.83 6.80 (.1-58
Week 9 5.35 4.98 0-24
Week 10 4.97 5.44 0-28
DHS
Week 1 18.95 (1.78 1-40
Week ‫و‬ 13.61 8.36 0-40
Week 4 12.79 8.31 0-39
Week 5 12.02 8.35 0-38
Week 6 12.02 8.06 1-39
Week 7 11.57 8.07 0-37
Week 8 11.37 8.42 0-40
Week 9 10.55 7.75 0-31
Week It) 10.63 8.47 0-37

Noll•. BDI II = Beck impression Inveniury-II: MAS = Measure of


Attribution^ Style( generality dimension,: lilts = Hassles and Uplifts
Seales (number of daily hassles).
4‫اال‬ Journal of Clinical Psychology. October 2002

the participants to complete the weekly questionnaire for tlie next ten weeks oil tlie same
weeknight. For example, if it participant Ciime in for an initial session on a Tuesday, the
weekly questionnaires were to he completed every Tuesday night for the next ten weeks.
Participants were asked to keep tlieir questionnaire and record all answ-ers on tlie answer
sheets. Answer sheets were returned to the experimenter's office the day at'ter cotnple-
lion. Receipt ،)..extra credit was contitigent on timely return of questionnaires. Ill addi-
tion, participants received one lottery ticket for the timc-ly return of each weekly answer
sheet, and received ten bonus lottery tickets for tlie timely return ol.all teti answer slieets.
At the completion of tlie study, five $10() prizes were itwartled based on a random draw-
ing of lottery tickets.

Results
Overview of Data Analytic Strategy
Hierarchic‫؛‬،! Linear Modeling (HLM: Bryk & Raudenbush. 1992‫ ؛‬Bryk. Riiudenbusli. &
Congdon. 1990.) w‫؛‬،! to analyze these dat،s use‫؛‬،. HL.M is an ideal clioice for tlie atiiilysis
ol.tliese data because it estimates within-subject (lower level or Level 1 data) and between-
sub.ject (upper level or Level 2 datii) variation simultaneously, thus allowing for the
if variatitin while considering the stiitistical characteristics of،if eacli source ،tnodeling
the other. The HLM strategy als،١ accommodates missing data (e.g.. Empiric‫؛‬،! Biiycs
estimates of Level 1 coefficients), an appealitig characteristic given tlie multiwave meth-
odology used in this stLitly. Tlie hierarchical linear models used in tlie current study
)t،t Level 1, but n،permitted missing data i ‫؛‬،t Level 2.

the Relation ،،،The Role of Negative Attributional Style '


between Daily Hassles and Depressive Symptoms
Tlie tliatliesis-stress component of tlie liopelessness theory (i.e., differential reactivity
model) predicts that the relation between daily liassles atid depressive symptomatology
is moderated by negative attributional style. To test tliis hypothesis, the within-sub.jects
model depicted eacli participant's level of depressive symptomatology as predicted by
nmiiber of daily Intssles. We observed a gener‫؛‬،! decreiising lineiir trentl in depressiv.e
if،symptomatology over time for many participants when inspecting tlie individual plots
B191-11 scores across the ten-week period, suggesting a significant ni‫؛‬،i ef.fect of time.،i
.depressive symptomatology, as indicated in،1،'riius. we first included time as a pre.dictor
the f.ollowing Level 1 model:

BDIjj = fioj + 01 >( Weekj) + ru (!)


wliere 0,„ represents participant y's predicted BDI-1I value when Week equals zero١, 01,
represents tlie linear effect of time iicross occasions. Week, represents the value of time
frtini Week () to Week 9, and fjj is tlie residual term ‫؛‬،ssociatetl witli observation / for
person ‫؛ 'ر‬،.id is ‫؛‬،ssunied to be independent and norniiilly distriliuted across p‫؛‬،rt‫؛‬cip‫؛‬،nts.
lull model as:،'Lhe Level 2 model represented a
00/ = Too + ،/)!/ )2(
01 j = 7،(‫ا‬ ‫ا‬ a 1/ )3(
where Too represents tlie predictetl group averiige BD1-I1 intercept. Toi represents tlie
)up average line،gr‫؛‬،،i the B191-II across tlie ten time nieiisurenients. iind Ho, and -،decrease i
Hopelessness Theory of Depression ISIS

//// are between-subjects residua! terms. Results continued a significant lineal, decrease in
BDI-11 scores over time. 0‫ل‬00,‫لم‬,>.9,86( = -145, ،)45‫ — = را‬..This finding is c.nsistent
with Ollier research that lias shown tliat participants in diary studies demonstrate a liabit-
uation effect in reporting mood symptoms (Gleason, Bolger. & Shrout. 2()()1).
To test tlie diathesis-stress liypotliesis. we first included daily hassles as a Level I lime-
varying covariate (Bryk & Raudenbush, 1992), resulting in tlie following Level I model:
BDI,, = (3„; + ($]j(Weekj) + pij(DHij) + r٧ (4)

wliere 02/ represents tlie change in depressive symptomatology for participant 7 for a
corresponding cliiinge in daily liiissles, DHjj is tlie number (if reported daily hassles for
occasion and participant 7. anti rn is tlie resitlual term associated witli observation /' for
person‫ر‬. Results ol'tflis l.,eve! 1 model and a corresponditig null Level 2 model ‫؛‬.idicated
that, in ildditioti tti significant intercept and linear chiinge coel'ficients. daily hassles emerged
as it significant predictor. 0٦, - .27. ،(145) = 9.56, p < .0()1. This result intliciited a
positive covariation between daily liitssles and depressive symptomatology above the
ef'fects of a significant lineiir trend over time. Ill addititm. variance components indicated
significttnl individual variation ،‫؛‬round tliese estiniiited average Level 1 coefficients, sug-
gesting tliilt predictors may be inclttded in tlie Level 2 model.
To examine tlie role ol. negative attribulional style ،‫؛‬s a cognitive diatlnesis ،.or depres-
sive reiictivity. tine following level 2 model was estiniiited:
00, = 7)<0 ‫ ب‬70‫اسا‬5١‫ (ر‬+ ،،><; ).‫(؟‬

01/ : 7 ‫ب‬0‫ ا‬JniMASj) ‫ ب‬U \ j (6)

02, = 720 ‫ ؛‬I(AUSj) + Uy٩7 (7)

where y٠)| is tine cross-level interactio،! term representing the effect of negiitive attribu-
tional style on initiill levels of depressive symptomatology. Til is the cross-level inter-
action term representing the effect of negiitive attributional style on linear clnange in
depressive symptomatology tnver time, )21 is tlie cross-level interaction term representing
tlie effects of negative attributioiiiil style on the link between daily liassles and depressive
symptomatology, and ‫ردء‬،،, ،،,;. iind ،،_,' iire between-subjects residuill terms. Tine effect of
MAS on initial 1319.-11 levels showed a trend towiird significance, y,)| = 0.25. ،(144) =
1.93, p = .053, suggesting that tlnose with greater levels ol. negative attributional style
tended to liiive higlnet- initial levels of depressive symptomatology. The effect of MAS on
lineat. decrease in BDI-11 across the ten weekly occasions was not signil'iciint, yii = 0.02.
،( 144) = 1.07, n.s. The effect o،. MAS on the relation between daily hassles iind BDI-11
was significiint. 72.07( = 144, ،)0.02‫ = ات‬, p < .05, intlicating that greater levels 0،.
negative attributional style were associated witli greiiter levels of depressive reactivity in
response to dilily liassles across tine ten weekly measurement occasions. Using a riltio of
the standard deviations for the MAS and riintlonn effect of (S2j. tlnis association translates
into ii standardized effect size of .39, ،'ailing in tlie range of a moderate effect. It should be
empliasizetl tliilt the decrease in BDI-11 scores over lime does not affect the interpretiition
ol' these ،.‫؛‬ndings. Tine effect of week numbei" on B191-11 scores was already accounted for
in the model. Moreover, tlnis pilttern of results remained when controlling ،'or the signif-
icant effects of the previous week's BDI-11 scores, suggesting that carryover effects in
depressive symptomatology from week to week did not affect tliese ،.‫؛‬ndings. Fiiiiilly. sex
ol' piirticipant and an interaction term ol. Sex X MAS also were entered in tine Level 2
nundels, but W'ere not significiint predictors of variability in Level 1 components.
1M6 Journal ‫ آذ‬Clinical Psychology, October 2002

Negative Attributioniil Style end Exposure to Stressors

Tlte differential exposure model predicts that individuals with greater negative attribu-
tional style would report more daily hassles. To evaluate this prediction, we first esti-
milled a model W'liere an individuitl’s frequency of daily ltassles on a particular occasion
is determined by iin average frequency of daily hassles iind i'،n individual deviation from
tlte group average. The Level I model is represented by tlie following equation:

+ ( 8)
where l)H,, is participant ‫ الر‬nuniber of daily hassles for tlte /th fjccasion. Poj is partici-
pant ,’s av'erage number of daily ltassles across all ten occasions, and is the residual
term associated with observation ‫ ز‬for person .‫'ر‬.
Tlte Level 2. or between-sub.jects, model evaluates whether tlte number of daily
hassles is a function ol' negative attributional style. At tliis level of analysis, the average
number of daily ltassles. 0(),, is hypothesized to be a ،'unction of atlributiitnal style. To
examine tliis hypotliesis, there must be intlividual variability in the average number ol'
daily hassles across all participants in the sample. Thus, we first determined whether
tltere was significant variability in average daily ltassles across the ten occasions ،'or each
participant by testing the magnitude of ،lie variance citmponent associated witli 0Oj using
the ،-'ollowittg Level 2 equation:

007‫ ر‬: <X) + U()j (9)


where tlte ujtper level estimate, Too, represettts tlte. grand mean of eaclt participant's
average daily hassles across tlte ten occasions and ‫ل‬0‫ اأ‬is a between-subjects residuill term.
The results based on estijitatittg the previously mentioned Level 1 and Level 2 mod-
els indicate tliat the predicted grand ntean of daily lta.ssles in the satttple (i.e., Too) was
13.26. In atldition. tlte grand Ittean of daily ltassles evinced significant individual varia-
tion. VAR itioj) = 49.77, ٢٦( 145) = 3.445.37. /> < .001. The presettce of significant
individual variation allow.s us to determine whether level of daily ltassles is a ،'unction of
attributional style. Tit examine this hypothesis, the MAS was entered as a Level 2 pre-
dictor of variability in Level 1 individual average daily hassles and is represented by the
،'ollowing Level 2 equation:
00/ = Ton ٠‫ا‬0‫(و ب‬MASj) + Uoj (10)

where 7(H) represents the grand ntean 0،'eaclt participant’s average daily hassles across the
te.n occasions when the V'alue 0،' MAS is 0, 7( ١‫ا‬represents tlte predictive effect of tlte MAS
on tlte average nuniber of daily hassles ،'or each participant; and u,,j is a between-subjects
residual term.
Results indicated tliat tlte MAS was not a signil'icant determinant of the dentitn-
strated individual variability in average daily ltassles. ■yoi = 0.14, t(144) = .80, n.s. Tliis
represents an e،'،'ect size o،' .08 and falls in the range of a small eft'ect siz.e. Moreover, sex
of participant and an interaction term of Sex X NJAS also were entered in the Level 2
model, but were not significatit predictors 0،' variability in Level 1 components. Thus,
tliese results are nitt supportive of the liypothesis tliat particijtants with negative attribu-
tional styles are expitsed to greater numbers of daily hassles.?

Based ‫؛‬in alpha of .05. ‫؛‬1 medium effeet size, ‫؛‬ind 145 parlieipanls. a conservative estimate of power given
;
on
the use of a repeateil Itteasures design is above .90. Moreover, tlte zero-order correlation between the MAS
and
-II is،initial BD .26)/‫ >ل‬.001). suggesting that there is sufficient ritnge of scores in the MAS to determine
Hopelessness Theory of Depression ISIS

statistically significant l'indings.


Hopelessness Theory of Depression 1317

Discussion
The findings from this study support tlte diathesis-stress component of the hopelessness
theory of depression: attributions ،style moderates tlie impact of stressors on depressive
symptomatology. A major contribution of tlte present study is that tlie findings shed
insight into a specit.ic mechitnism through which tlte ltopelessness the-ory Ittity operate.
The l'indings suggest tltat negative attributions ،style leads to greater depressive reactiv-
ity to stressttrs over little. For individuals with a negative attributions, style, tltcre is ،‫؛‬
greater correspondence between levels of depressive symptoms and tlie severity' of stress-
ors experienced at a particular ptiint in lime. Tltis fintling stood even itfter controlling for
tlie previous week's level of depressive symptoms, indiciiling that the result was not due
to carryover effects in depressive symptomatology. In addition, tliere was no evidence for
a model ol' differential exposure to stressors as a patli toward increasetl depressive
symptomatology.

Different،،‫ !؛‬Reactivity U S ‫ا‬،Framework for Understanding


the Diathesis-Stress Mechanism
Traditionally, negative attributional style has been viewed as a vulnerability that leads to
depression given a certain threshold of life stress (i.e.. titration Jttodel). Ill addition, diathesis-
stress models of depression suclt as tlte liojtelessness theory assttnte that predispositions
to depression are liitent entities that do not become activated until tlte intlividuill experi-
ences life stressors (Monroe & Sintotts. 1991). Tlte Jtreseni findings suggest that individ.
uals witli a negative attributional .style have greater depressive reactions to eaclt life
stressor. These negative reactions iire not [tretlicaled on exceeding a certain threshold of
stress. Thus, tltis vulnerability is important across itll levels of life stress. Clinically', tltis
suggests til ill clients witli a negative attributional style may benefit ،'rojtt reevaluating
tlieii" attributions regarding every life stressor, regardless of its severity. These clients
may not require a severe level of life stress foj- depressive reactittns to occur.
Researchers liave found thitt other forms of. increase،! reactivity lead to a greater
propensity toward depression or depressive symptotnatology. For example, self-esteetit
variability anti reactivity to lil'e stressttrs lias been prospectively relilted to depressive
symptomatology (Roberts & Gotlib. 1997). Thus, individuals wltose sense of worth vac-
illiites accordittg t ١،theii. life circujttstatices are at greater risk for depressive symptoms.
Cognitive reactivity t،) a negative mo١،d induction also has been linketl witli dejtression
(Miranda & Persons. 1988) and is believed to increase the risk of relapses in clinical
depressioti (Segal. Genial.. & Williams. 1999). Moreover, hyperactivity ol.tlie neurochetii-
ical and hormonal systetiis in reaction to lil'e stressors lias been hypothesized to be a
causal mechanism in depression (Gold. Goodwin, & Chrousos, 1988).
Another contribution of tlie current study was the investigation of whether attribu-
tional style plays a role in differential exposure to stressors. Tlie results indiciited tliat
attributional style does not pretlict t.uture ،)ccurrcnces of stressors. This is a potentially
key finding in clarifying the depressogenic mechanism involved in tlie hopelessness theory
ol" depression. Unlike tntlier vulnerabilities, such as self-criticism, it appeiirs that a negji-
tive attributiotial style represents an internal, passive process thitt does not generate iitldi-
tionitl stress, 'riius, it appears at this point tliat tlie hopelessness theory of depression is
best conceptualized solely as a differential reactivity model.

Study Limitations mid Future Directions


Our fintlings suggest that stressors, in tlie form of daily liassles. are linked to depressive
symptomatology anti ‫؛‬Jiteriict with individual dil'l'erences in negative attributional style to
/.?/>‫ ؟‬2‫(ااا‬2, )(‫ا‬-‫ا‬،٠‫ا‬١‫اا‬٢ 1 PsvcholoRy‫ا؛‬١‫الا؛‬61 ،٠١ r٠١al،١٠،J

predict greater increases in depressive reactivity. Although our findings point in this
direction, it is an open empirical question whether tlie utility of the differential reactivity
model will lioltl in samples containing more cliniciilly iicute forms of depression.
A replication with a clinical sample wotild he important in addressing tliis issue.
However, it should he noted that re.search with an analogue population can be potentially
informative i.1 developing testable hypotheses regarding clinical disorders (e.g.. Tennen.
Eberhardt, & Affleck, 1999). It also should be noted tliat tlie curre.it study assessed stress
in terms of daily liassles. Daily hassles are iin importiint stress variable in considering
variations in miltl tlepression over short intervals of. time (DeLongis et al., 1988). How'-
ever, clinically significant depression is ol'ten triggered by major life eve.nts. Future inves-
ligations using a prospective, multiwiive approach in tliis area could assess ma.jor life
events thougli the use of. interviews. It slitiuld be noted, however, that given tlie low'
frequency of major events, tlie time interval usetl in such studies may need to be longer
(e.g.. one-montli intervals). 1.1 addition, it may lielp to preselect high-risk and low-risk
individuiils biised till tliei.- attributiojiiil sty'le, using a prospective behiivioral high-risk
tlesign. Tliis design, i،!ong with tlie use of ،‫؛‬large sample, would increase the odds that
participants in tlie study would experience clinically significant lev.els of depression in
response to .najor life events. In addition, sucli a design would allow a ،'urther test ol'tlie
differential exposure model.
In I'uture studies, incltiditig tlitise involving clinical depression, it may be lielpful to
test both tlie traditional titration model itnd tlie differential reactivity model presented
liere. It may be that botli meclianisnis are possible in producing clinical depression or it
may be that one Jiiodel is more accurate in capturing the mechiinism of action in the
hopelessness theory.
It also would be informative to replicate tlie currefit findings with Ollier measures ()،-
attributional style, sucli as tlie EASQ. The MAS has not been as widely used as the EASQ
i.1 invesligatitms of tlie liopelessness tlieory of depression.
It is possible that a differential exposure el'fect was present in our study, hut was too
sniiill to be detected with our sa.nple size. In addition, the low-scale reliability ol' the
MAS in tlie curretit study (a = .67) reduced the power in tlie current study to detect such
an effect. Finally, the use (if life-stress interviews, rather tlian a self-report questionnaire,
may liave bee.1 a .more powerful test ol.the differential exposure model. Continued work
1-uli.ig out tlie dit'ferential exposure model regarding attributiojiiil style would be useful.
To furtlier explore tlie role of attributional style in therapy for depression, l.uture
investigations could incorporiite a prospective niultiwave longitudinal design investigat-
ing responses to therapy among depressed clients. One could investigate W'hether attri-
liutiojial style .ueasured at tlie onset of tlie study is predictive o،' clianges in depressive
symptomatology during psychotherapy. Such ii study could investigate not only an Attri-
liulional Style X Stress interaction in predicting increases in depressive symptoms but
also an Attributional Style X Tlierapy Intervention interaction in pretlicting decreases in
depressive symptoms. 1.1 addition, stutlying cliniciil populatiojis i.1 future st'utlies would
extend the generalizability of the current ،'indings.
Conducting tests of. alternative hypotheses regarding tlie ..elation between attribu-
tional style and depression also would be beneficial. For instance, support for tlie liope-
lessness tlieory of depression would be strengthened it. research coultl show that attributional
style does not cov-ary with tlie ticcurrence of stress or in response to clianges in depres-
sion. Such fintlings would rule out tlie possibility that attributio.ial style is a consequence
of depression anti would strengthen its status as a vulnerability factor to depression. To do
this, future longitudinal studies could assess attributiontil style iis a repealed measure
along witli stress anti depression.
Hopelessness Theory of Depression /.?/y

References

Abramson. L.Y., Alloy. L.B.. & Metalsky, G.I. (1988). The cognitive diathesis-stress theories ol'
depression: Toward an adequate evitlualion of tlie theories' validities. In L.B. Alloy (Ed.),
Cognitive processes in depression (pp. 3-30). New York: Guilford Press.
Abramson, L.Y.. Metalsky, (3.1., & Alloy, L.B. (1089). Hopelessness tlepression: A theory-based
subtype of depression. Psychological Review. 96. 338-372.
Abramson, L.Y.. Seligman. M.E.P., & Teasdale. J.D. (1978). Learned ftelplessness in ltumans: Cri-
titjue anti reformulation. lountal of Abnormal Behavior, 87, 49-74.
Alloy, L.B., Abramson. L.Y., Whilehouse. W.G., Hogan. M.E., Tashman. N.A.. Steinberg, D.L..
Rose. D.T., & Donovan, p. (1999). Depressogenic cognitive styles: Predictive valitlity. infor-
mation processing and personality characteristics, anti developmental origins. Behaviour
Research anti Therapy, 37, 303-331.
Allfiy, l-'.B., & Clements. C.M. (1998). Hopelessness theory of depression: Tests of the symptom
component. Cognitive Therttpy and Resettrch. 22. 303-333.
Alloy, L.B., Just. N., & Panzarella. c. (1997). Attrihutional style, daily life events, anti hopeless-
ness depression: Subtype validation by prospective variability iind specificity of symptoms.
Cognitive Therapy and Researcli. 21. 321-344.
American Psychiatric Association. (1994,). Diagnostic and statistical manual of mental disorders
(4th etl.). Washington. DC: Author.
Btirnett. P.A., & Gotlib. I.H. (1988.). Psychosocial functioning and depressil.n: Distinguisliing among
،‫؛‬ntecedents, concomitants, itnd consequences. Psychological Bulletin. 104, 97-126.
Beck. A.T., Steer. R.A., & Brown. G.K. (1996). Beck Depression Inventory manual (2nd etl.). San
Antonif). TX: Psyclioltigical Corporatitin.
Beck, A.T., Ward. C.H.. Mentlelst.n. M., Mock, j.. & Erbaugh, j. (1961). An inventory for meitsur-
ing depression. Archives of General Psycliiatry. 4. 361-671.
Bt.lger. N., & Zuckerman. A. (.1995). A framework ('or study'ing personality in tlie stress process.
Journal of Personality :،nd Sociitl Psycliology, 69, 890-902.
Bryk, A.S., & Raudenbush, S.W. (.1992). Hierarchical linear models: Applications and datit itnalysis
methods. Newbury Park, CA: Sage.
Bryk, A.S.. Raudenbush. S.W.. & Ct.ngdon. R.T. (1996). Hierarchical linear modeling witli tlie
HLM/2L and HLM/3L programs. Cliicagt,: Scientific Software.
Coyne. J.C.. ‫ه‬Gotlib. 1.1-1. (1983'). The role of cognition in depression: A critical appraisal. Psy-
chological Bulletin, 94. 472-505.
Davila, j.. Hanimen. c.. Bui'ge. D.. Paley. B.. & Daley. S.E. (1995). Poor interpersonal problem-
solving as a mechiinism ol. stress generation in depression among adolescent women. Journal
of Abnormal Psychology. 104.592-600.
DeLongis, A.. Folkman. s.. & Lazarus. R.S. (1988 ). Tlie impact of daily stress tin liealtli and mood:
Psychological ،‫؛‬nd social resources as mediattirs. Journal of Personality and Sociitl Psychol-
ogy. 54. 486-495.
Dozois, D..I.A., Dobson, K.S.. & Alinberg. J.L. (1998). A psychtimetric evaluatitin tif tlie Beck
Depression Inventory-II. Psyclioltigical Assessment. 10. 83-89.
Diveck. C.S.. & Liclit. B.G. (1980). !.earned helplessness and intellectUill achievement. In j. Garber
& M.E.P. Seligman (Etls.). IluniiUi lielplessness: Tlieory and ‫؛‬-،pplication ('pp. 197-221). New
York: Acitdemic Press.
Follette, V.M., & Jacobson. N.S. (1987). Tlie importance tif attributitms as it predictor of how
people ctipe witli failure. Journiil of Personality and Soci،‫ !؛‬Psychology, 52, 1205-1211.
Gleason, M., Btilger, N., & Shrout, p. (2001. February). The effects 1-if researcli design tin reports of
mood: Comparing ditily diary, panel, and cross-sectitinal design. Paper presented at tlie iinnuiil
meeting of the Sticiety t٠t١r Personality and Social Psychology-. Sill! Antonio. TX.
Goltl. P.W., Gtiodwin. F.K.. & Chrousos. G.P. (1988). ('linical ،‫؛‬nil biticliemical manifestations ،if
1320 Jourittl ٠«‫ ا‬Clinical Psychology. October 2002

depression: Relation to llie neurobiology ol stress. New England Journal ol Medieine. 319,
413-42(1.
Goldstein. H. (1987). Multilevel models in educational and social research. London: Oxford flni-
versity Press,
Hammen. c. (1991). The generation ol stress in tlie ,course of unipolar depression. Journal of
Abnormal Psychology, 1(1(1. 535-301.
Houston. D.M. (1995). Vulnerability t،> depressive mood reactions: Retesting the hopelessness
model of depression. Britisfi Journal of Social Psychology. 34. 293-3(12.
Hunsley. j. (1989). Vulnerability t١، depressive mood: An examination ol. the temporal consistency
of the reformulated learned helplessness model. Cognitive Therapy and Research. 13.599-0( IX.
Janoff-Bulman. R. (1979). Charactcrological versus behavioral self-blame: Inquiries in،() depres-
sion iind rape, .lournal of Personality and Social Psychology, 37. 1798-1809.
Kenny, D.A., Kashy, D.A.. & Bolger, N. (1998,1. Data analysis in social psychology. In D.T. Gil-
bert. S.T. Fiske. & G. Lindzey (Eds.), Handbook of social psychology (4 ‫ا‬1‫ ا‬ed.. Vol. 2, (ip.
233-205). Boston: McGraw-Hill.
Kwon, p. (1999,). Attributions! style and psychodynamic defetise mechanisms: Toward an inlegra-
tive model of depression. Journal of Personality. 07, 045-058.
Kwort. p.. & Lemon. K. (2(1(10). Attributiottal style and defense meclitinisms: A synthesis of cogni-
tive ‫؛‬ind psychodynamic factors in depression. Journal of Clinical Psychology. 50. 723-735.
Kwon. p.. & Whisman, M.A. (1992. November). A longitudinal stutly of tlie hopelessness theory of
depression: New measures for assessing depressogenic attributional style. Paper presented itt
the annual nteelittg of tlie Association f١،r Advancement of Behavior Tlierapy. Boston.
Lara. M.F:., & Klein. D.N. (1999). Psychosocial processes Ultderlying the maintenance and persis-
tence ،)٠، depression: Implications for understitnding chronic depression. Clinicttl Psychology
Review. 19, 553-57(1.
Metalsky. G.I., Halberstadt. l..J.. & Abramson. L.Y. (1987). Vulnerability to depressive m،a١d reitc-
tions: Toward ،‫؛‬more powerful test of tlie diilthesis-stress and Citusal mediation components of
the reformulated theory' of depression. Journal of Personality and Social Psychology, 52,
386-393.
Metalsky. G.I., & Joint.r. T.E.. Jr. (1992). Vulnerability to depressive symptomatology: A prospec-
tive test of tlie diathesis-stress iuid Citusal mediation components of the hopelessness theory ol.
depression. Journal of Persrmality ،٤n،l Social Psycliology. 63. 667-675.
Metalsky, G.I., & .!،liner. T.E., .lr. (1997). The !!،.ipelessness Depression Symptom Questionnaire.
Cognitive 'fherapy and Researcli. 21. 359-384.
Metalskv. G.I.. Joine.-. T.E.. Jr.. Hitrdin. T.S.. & Abramson. L.Y. (1993). Depressive reactions to
failure in ،‫؛‬naturalistic setting: A test of the hopelessness ،‫؛‬nd self-esteem theories of tlepres-
sion. Journal of Abnormal Psychology, 102, 101-109.
Miranda.& Persons. J.B. (1988,). Dysfunctional ‫؛‬Ittitudes ‫؛‬ire mood-state dependent. Journal of
Abnormal Psychology, 97, 76-79.
Mongrain. M. (1998). Parental representations and support-seeking behaviors related to depen-
dency and self-criticism. Journal of Personality, 66. 151-173.
Monroe. S.M., & Simons, A.D. (1991). Diathesis-stress theories in tlie context of life stress research:
Implications for tlie depressive disorders. Psychological Bulletin, 1 IB. 406-425.
Persons, J.B.. & MiriUlda. j. (1992). Cognitive theories of vulnerability to depression: Reconciling
negative evidence. Cognitive Therapy anti Researcli. 16, 485-502.
Peterson, c.. & Seligman. M.E.P. (1984). Citusal explanations its it risk fiictor for depression:
Theory and evidence. Psychtilogiciil Review. 91.347-374.
Peterson, c.. & Seligmttn. M.E.P. (1985). Tlie leiirned helplessness mtidel of depression: Current
status oftlietiry iind research. In F.F.. Beckliam & W.R. Leber (Eds.). Handbook of depression:
Treatment, assessment, and research (pp. 914-939). Homewood. IL: Dorsey.
Hopelessness Theory of Depression 1321

Peterst.li, c.. & Villanova. p. (1988). All Expanded Attributional Style Questionnaire. Journal of
Abnormal Psychology. 97. 87-89.
Potthoff, J.G.. Holahan. C.J., & Joiner, T.E. (1995'). Reassurance-seeking, stress generiition. iind
depressive symptoms: An integriitive mtidel. Jtiurnal of Personality and Social Psychology,
‫؛‬18. 664-670.
Priel, B., & Slialiar, G. (2000). Dependency, self-criticism, sociill context, and distress: Ctimparing
moderating and mediating models. Personality & Individual Differences. 28. 515-525.
Ralph. J.A., & Mineka. S. (1998). Attrihutional style anti sel!'-esteem: The prediction ol'emotional
tlistress following a midterm exam. Journal of Abnormal Psychology, 107. 202-215.
Roberts, J.E., & Gotlib. 1.11. (1997). Temporal varialiility in global self-esteem and specific sell'-
evaluation as prospective pretlictors of emotional distress: Specificity in predicttirs iind out-
come. Journal of Abnormal Psycliology. 106. 521-529.
Robins, C.J.. & Hayes, A.M. (1995). The role ofcaustil attributions in die pretliction of depression.
In G.M. Buchanan & M.E.P. Seligman (Eds.), Explanatory style (pp. 71-97). Hillsdale. NJ:
Erlbaum.
Rogosa. 17.R., Brandt, 17.. & Zimowski. M. (, 1982). A growth curve approach to the meiisurement of
change. Psycliological Bulletin. 90. 726-748.
Segiil. Z.V., Gemar. M.. & Williams. s. (1999). Differential cognitive response to a mood challenge
following successful cognitive tlierapy or pharmacotherapy for unipolar depression. Journal of
Abnormal Psychology, 108. 2-10.
Sweeney. P.D.. Anderson. K.. & Bililey. s. (1986). Attrihutional style in depression: A meta-
analytic review. Journal ol' Personality iind Social Psychology, 50. 974-991.
Swendsen, J.17. (1998). Tlie helplessness-hopelessness tlieory anil daily mooil experience: An idio-
grapilic and cross-situational perspective, .lournal ol' Personality and Social Psychology. 74.
1298-1408.
Tennen, Eberhardt, T.l..١ & Aflleck. G. (1999). Depression research methodologies at the social-
clinical interfiice: Still liazy after ill! tliese years. Journal of Social and Clinical Psychology,
18, 121-159.
Willett, J.B. (1989). Some results on reliability ،'or the longitudinal measurement ol'change: Impli-
cations for the design of studies of individual growtli. Educationtil and Psychological Mea-
surenient. 49, 587-602.
Willett. J.B. (1997). Measuring chiinge: w'liat individual growtli modeling buys you. Ill E. Anistead
& K.A. Renninger ٠Ells.). Cliange anil development: Issues of theory, metliod. anil application
(jip. 212-242). Mahwah, NJ: Lawrence Erlbaum.
Willett, J.B.. Singer, J.D.. & Martin. N.C. (1998). The design and analysis of longituilinal studies of
development and psychopathology in context: Statistical models anil metliodologic،‫ !؛‬reconi-
mendations. Development and Psychopathology, 10. 295-426.
Copyright of Journal of Clinical Psychology Is the propefy of John Wiley & Sons Inc.
and Its content may not be copied or emailed to multiple sites or posted to a listserv
without the copyright holder's express written permission. However, users may print,
download, or email articles for Individual use.

You might also like