Behaviour Research and Therapy 42 (2004) 293–313

www.elsevier.com/locate/brat

Avoidance and depression: the construction of the

Cognitive–Behavioral Avoidance Scale
Nicole D. Ottenbreit ∗, Keith S. Dobson
Department of Psychology, University of Calgary, 2500 University Drive NW, Calgary, Alberta, Canada T2N 1N4
Received 9 December 2002; received in revised form 14 April 2003; accepted 17 April 2003

Abstract

Previous research has demonstrated a relationship between avoidance and depression, although the nature
of this relationship has been confused by inconsistent definitions of avoidance. Following a review of the
construct of avoidance and an examination of past research, a new multidimensional measure of avoidance
was developed and validated using an undergraduate student sample (245 females and 146 males). Four
reliable factors reflecting combinations of cognitive/behavioral and social/nonsocial dimensions of avoid-
ance were obtained from the factor analysis of the new scale, entitled the “Cognitive–Behavioral Avoidance
Scale” (CBAS). The scale showed the predicted relationships with convergent and divergent measures used
for construct validation. As predicted, subscales of the CBAS, as well as the overall scale score, were
significantly related to the depression and anxiety criterion measures. The findings from this study suggest
that avoidance may be an important construct in the psychosocial conceptualization of depression. In
addition, this integrated measure of avoidance has potential utility for depression researchers. Suggestions
for further scale development and research are provided.
 2003 Elsevier Ltd. All rights reserved.

Keywords: Avoidance; Coping; Depression; Scale construction

1. Introduction

The construct of ‘avoidance’ refers to refraining from, or escaping from, an action, person or
thing. Avoidance has been studied extensively in relation to anxiety and has been established as
a central feature in the description, diagnosis and treatment of anxiety disorders (Barlow, 2002).
In contrast, although Ferster (1973) postulated a central role for avoidance in his functional analy-

∗
Corresponding author. Tel.: +1-403-874-9636; fax: +1-403-220-3697.
E-mail address: [email protected] (N.D. Ottenbreit).

0005-7967/$ - see front matter  2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0005-7967(03)00140-2
294 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313

sis of depression several decades ago, research examining avoidance in the context of depression
has been comparatively scarce. Ferster’s model of depression stated that the depressed person
engages in a high frequency of avoidance and escape from aversive internal and external stimuli
with such behaviors as withdrawing or complaining, and that these efforts preempt positively
reinforced behavior. Ferster indicated that the predominant employment of behaviors serving an
avoidance or escape function leads to a narrowing of the depressed individual’s behavioral reper-
toire and that this repertoire becomes marked by passivity as opposed to initiative or action.
Despite this promising theoretical basis for empirical investigation, it has only been in recent
years that researchers have begun to systematically examine the relationship between the construct
of avoidance and depressive symptoms and disorders. It is possible that the construct of ‘avoid-
ance’, as explored and clarified in this paper, is a permeating factor contributing to the established
relationships between previously studied psychosocial variables and depression. The following
review presents important research on avoidance in the context of depression and an integrated
conceptualization of the avoidance construct which serves as the foundation for the construction
of a multidimensional avoidance scale.

2. Literature review

In general, research examining the relationship between avoidance and depression has examined
avoidance as: (1) a coping strategy, (2) a problem-solving style, and (3) a personality dimension.
Positive associations between various definitions of avoidance subsumed under these facets and
depression have been identified in a number of studies.

2.1. Avoidance as a coping strategy

Most studies employing definitions of avoidance as a coping strategy can be understood in the
context of Moos and Schaefer’s (1993) classification scheme for coping responses. This scheme
posited the following categories underlying coping responses: (1) cognitive versus behavioral, and
(2) approach versus avoidance responses. Within this framework, ‘cognitive avoidance coping’
encompasses responses aimed at denying or minimizing a crisis and/or its consequences or
accepting a situation because of the belief that circumstances cannot be changed. ‘Behavioral
avoidance coping’ includes responses aimed at seeking alternative rewards or escape, or avoiding
behaving in direct response to a stressor. In contrast, approach coping strategies directly address,
through cognitive or behavioral means, the problem and/or its consequences.
A review of the research reveals that most studies have found support for a positive association
between avoidance coping and depression. For example, Blalock and Joiner (2000) examined the
impact of different types of coping, as assessed by the Coping Responses Inventory (Moos, 1988),
and gender on the relationship between life stress and the criterion variables of depression and
anxiety. It was found in this sample of 179 university students that high negative life event scores
were predictive, over a 3-week period, of significant increases in depressive and anxious symp-
toms for females, but not males, who used greater cognitive avoidance coping strategies. The
results also showed that behavioral avoidance coping was unrelated to short-term changes in
depressive or anxious symptoms.
 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313 295

The results attained by Blalock and Joiner (2000) are consistent with recent research on what
have been referred to as the ironic or paradoxical effects of thought suppression (Beevers,
Wenzlaff, & Hayes, 1999; Rassin, Merckelbach, & Muris, 2000). Thought suppression, which
refers to the process of consciously trying to avoid certain thoughts, reflects the functional essence
of cognitive avoidance. The idea reflected in this line of research is that depressed individuals
are likely to engage in thought suppression as a way to promote their well-being, but that attempts
to suppress thoughts may actually lead to increased frequency of these thoughts and a resultant
decrease in well-being.
Kuyken and Brewin (1994) examined the types of coping strategies, as assessed by the Ways
of Coping Questionnaire (Folkman & Lazarus, 1988), employed by 32 clinically depressed and
32 nondepressed women following a recent interpersonal stress event. These investigators found
that depressed women engaged in more escape–avoidance coping strategies (included behavioral
and cognitive avoidance responses) and less planful problem-solving or positive reappraisal, con-
trolling for the stressfulness of the event, than nondepressed female controls. Other studies have
similarly found support for a relationship between avoidance coping and depressive symptoms
with community participants (Folkman & Lazarus, 1986) and psychiatric outpatients (Spurrell &
McFarlane, 1995).
The relationship between avoidance coping and depression has also been examined in longitudi-
nal research. Holahan and Moos (1986) found that a disinclination to use avoidance coping stra-
tegies, in combination with other variables including self-confidence, an easy-going disposition
and family support, was associated with reduced risk for depression and psychosomatic symptoms
over a 1-year period in a large community sample of 245 men and 248 women. Research has
also found avoidance coping at the time of intake to be associated with lack of remission status
over a 1-year period in a sample of 233 women and 172 men presenting for the treatment of
depression (Krantz & Moos, 1988).
The majority of the research in this area indicates that behavioral and cognitive avoidance
coping strategies are associated with higher levels of reported depressive symptoms, both concur-
rently and over time. It should be noted, however, that despite the continuing debate in the coping
literature as to whether coping is situational or more of a general style or trait (Krohne, 1996),
the majority of the studies in this area employ coping measures assessing individuals’ responses
to a specific situation or problem. Future research should operationalize and examine coping
utilizing a trait conceptualization in order to determine if coping styles do in fact show stability
and if any particular styles convey particular vulnerability for depression over time.

2.2. Avoidance as a problem-solving style

Within the social problem-solving framework, avoidance strategies are viewed as the result of
ineffective problem-solving, and to inhibit further problem-solving efforts (D’Zurilla & Nezu,
1999). As social problem-solving theory views an active, approach focus in problem-solving as
optimal, it has been shown that a passive, avoidant style of dealing with problems is more typical
of individuals showing elevated levels of depression (D’Zurilla, Chang, Nottingham, & Faccini,
1998).
A longitudinal study by Davila (1993) provides additional support for the role of an avoidant
problem-solving style in depression. This study examined the relationships between attachment
296 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313

cognitions, interpersonal problem-solving ability, stress and depression over a 6-month time frame
in a sample of 94 female high school students. The interpersonal problem-solving measure con-
sisted of an interview-based presentation of four interpersonal scenarios, with responses scored
along a number skill dimensions as well as for level of behavioral avoidance. Although not
intended to be the focus of Davila’s study, behavioral avoidance showed important relationships
with many of the other assessed variables. It was found that an avoidant problem-solving style
was associated with higher levels of insecure attachment cognitions and the generation of more
objectively stressful life events than active problem-solving. The results also showed that avoidant
problem-solving may serve to protect women from depression when dealing with discrete episodic
events, but that these strategies were associated with increased levels of depression when used
to deal with chronic stress in romantic and familial domains.
The above results suggest that a relationship exists between depression and an avoidant prob-
lem-solving style, and that different domains of functioning may be affected differently by the
use of avoidance strategies. It will thus be important to distinguish between avoidance associated
with various domains of functioning. Further research employing longitudinal designs will be
needed to clarify the nature of the relationship between problem-solving styles and depression.

2.3. Avoidance as a personality dimension

Studies defining avoidance in terms of a personality dimension have also found support for the
relationship between avoidance and depression. ‘Harm Avoidance’ is a personality dimension
defined by Cloninger (1987) which refers to the tendency to inhibit behavior to avoid punishment,
novel stimuli and the lack of rewards. Harm Avoidance (HA), along with Novelty Seeking and
Reward Dependence, comprise the three character dimensions assessed by the Tridimensional
Personality Questionnaire (TPQ; Cloninger, Przybeck, Svrakic, & Wetzel, 1994). Research has
shown consistent support for a relationship between HA and depression. For example, Hansenne
et al. (1997) identified a positive relationship between increased serotonergic activity and HA, as
well as an association between severity of depression, as assessed by the Hamilton Depression
Rating Scale (HDRS; Hamilton, 1960, as cited in Hansenne et al., 1997), and HA in a group of
21 depressed inpatients.
Despite the positive association of HA with depression, research examining the stability of HA
has revealed inconsistent patterns over time. Richter, Eisemann, and Richter (2000) found support
for the stability in HA, in that HA scores remained elevated following remission of depressive
symptoms after combined pharmacological and psychotherapeutic treatment. Previous studies,
however, have found that HA scores tend to normalize following successful treatment of
depression (Chien & Dunner, 1996; Hansenne, Pitchot, Gonzalez Moreno, Machurot, & Ansseau,
1998). Thus, although the dimension of HA can be concluded to correlate with depression, the
trait nature of this dimension remains of question.

2.4. Framework for the construction of an ‘avoidance’ scale

Although there clearly appears to be a relationship between avoidance and depression, the
varying definitions of avoidance employed in previous studies render the comparison of results
across studies and the formulation of any definitive conclusions in this area difficult. In addition,
 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313 297

the avoidance measures employed to assess the relationship between avoidance and depression
in these studies typically comprised one or more scales or subscales of other coping or personality
scales. It is likely that these measures fail to capture important dimensions that comprise the true
construct of avoidance. What is needed at this point in time is an integrative and valid measure
of the construct of avoidance which can be employed in future studies examining the nature of
the relationship between avoidance and depression. This measure should utilize and validate a
trait conceptualization of avoidance, given support for the stability of coping (Carver & Scheier,
1994; Carver, Scheier, & Weintraub, 1989) and the value that a trait measure holds in terms of
identifying risk for psychopathology.
It was proposed that several dimensions may be important in the conceptualization of the con-
struct of avoidance. These include cognitive versus behavioral, active versus passive and social
versus nonsocial types of avoidance. The coping literature supports the need to distinguish
between cognitive and behavioral types of avoidance (e.g. Blalock & Joiner, 2000). A review of
the specific types of avoidance assessed in the previous studies also showed variability in the
level of passivity involved in the avoidance strategies employed. It was found that some strategies
involved initiating an action in order to escape a situation, while others involved not doing some-
thing in order to avoid the situation. This active versus passive dimension of avoidance was
thought to hold informational and perhaps discriminative value, given the style of passivity
described in depressed individuals (Ferster, 1973).
Finally, although research has generally failed to distinguish situations or problems being avo-
ided that are of a social versus nonsocial nature, it is likely that the domain in which a problem
or situation arises would offer important information about individuals’ employment of avoidance
strategies and/or the relationship between these strategies and depression. In this regard, it should
be noted that the characteristics of sociotropy and autonomy (Beck, Rush, Shaw, & Emery, 1979;
Clark, Beck, & Alford, 1999) convey differential risk for depression when individuals are con-
fronted with life events in social versus nonsocial domains. Research has shown that sociotropic
individuals (those who place high value on social connection and acceptance) are most at risk
for depression when confronted with interpersonal loss events, while autonomous individuals
(those who place high value on independence and self-control) are at risk for depression when
confronted with loss events threatening independence, control or achievement (e.g. Coyne &
Whiffen, 1995). This research seems to suggest the salience of the domain of a problem or situ-
ation to an individual’s manner of coping with situations and thus, the importance of including
the social versus nonsocial dimension in conceptualizing avoidance.
Definitions of the types of avoidance within this conceptual scheme are presented below. Within
these definitions, a ‘problem’ consists of either an external situation or task or an internal thought,
emotion or experience which demands a response for adaptive functioning (Nezu, Nezu, & Perri,
1989). ‘Active Cognitive Avoidance’ involves avoidance of a problem through denial, minimiz-
ation or cognitive distraction. ‘Passive Cognitive Avoidance’ involves passive acceptance of and
failure to address a problem. ‘Active Behavioral Avoidance’ involves escape from a problem
and/or engagement in alternative/distracting activities, while ‘Passive Behavioral Avoidance’
involves avoidance of a problem or of dealing directly with a problem. These types of avoidance
are further specified by problem domain, which may be either ‘social’, which refers to the involve-
ment of other people (including social contacts and activities) or ‘nonsocial’, reflecting no involve-
ment of other people (including achievement-related and solitary activities).
298 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313

Using this conceptual scheme, a scale was developed to measure an integrated and multidimen-
sional model of avoidance. The primary purposes for the development of this scale were to deter-
mine the important dimensions of avoidance that would emerge in empirical analyses, and to
examine the concurrent relationship between avoidance and depression. It was also hoped that
this study would produce a valid and reliable measure of avoidance which could then be employed
in future investigations to allow for a more in-depth examination of the relationship between
avoidance and depression.
It was hypothesized that multiple reliable factors reflecting the hypothesized dimensions would
emerge from a factor analysis of the avoidance item pool. If all specified dimensions of avoidance
proved to be important to the definition of the construct, eight factors reflecting combinations of
these avoidance dimensions were expected to emerge. The scale was also hypothesized to show
substantive positive correlations with the convergent measures of avoidance described below and
to show nonsubstantive negative correlations with the divergent (approach) measure also described
below. Given that the scale was thought to measure a trait conceptualization of avoidance, it was
predicted that the scale would show good test–retest reliability over a brief interval. Of great
importance to the premise of this project, it was hypothesized that the scale would show substan-
tive positive correlations with the depression and anxiety measures described in the subsequent
section.

3. Method

3.1. Participants

The sample consisted of 391 undergraduate students (245 females, 146 males) recruited through
the University of Calgary, Department of Psychology Bonus Credit Program. The mean age of
the participants was 21.70 years (SD = 4.59; range 17–51). The racial makeup of the sample was
predominantly Caucasian (62%), followed by Asian (28%), East Indian (6%), and other groups
(4%).

3.2. Measures

3.2.1. Coping Responses Inventory (CRI; Moos, 1988)

The CRI is a self-report measure of eight different types of coping responses to stressful life
situations as reflected in eight subscales. Logical Analysis and Positive Reappraisal comprise the
cognitive approach coping scales. Seeking Guidance and Problem-Solving comprise the behavioral
approach coping scales. The cognitive avoidance coping scales include Cognitive Avoidance and
Acceptance and Resignation and the behavioral avoidance scales include Alternative Rewards and
Emotional Discharge. All eight subscales are composed of six items. Each item asks respondents
to indicate to what extent, from “not at all” to “fairly often”, they employed this strategy to deal
with the most stressful situation they encountered in the last year. The scale shows fair internal
consistency (a = 0.61–0.72 across scales for males and 0.58–0.71 across scales for females) and
stability (r = 0.45 for males and 0.43 for females averaged across scales) over a 12-month period.
This scale shows good convergent validity with prior established coping scales. The avoidance
 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313 299

coping subscales were used to assess the convergent validity, whereas the approach coping subs-
cales were used to assess the divergent validity, of the new avoidance measure.

3.2.2. Ways of Coping Questionnaire—Escape–Avoidance Scale (WCQ; Folkman & Lazarus,

1988)
The Escape–Avoidance Scale of the WCQ is an eight-item scale designed to assess coping
responses reflecting wishful thinking and behavioral efforts to escape or avoid the problem.
Respondents are asked to indicate to what extent they employed these strategies (on a 0–3 scale)
in coping with the most stressful situation they experienced in the past week. This scale shows
adequate internal consistency (a = 0.72), which the authors indicate is high in comparison to
other coping measures. This measure was used to assess the convergent validity of the new scale.

3.2.3. Tridimensional Personality Questionnaire—Harm Avoidance (HA) Dimension (TPQ;

Cloninger, Przybeck, & Svrakic, 1991; Cloninger et al., 1994)
The HA dimension of the TPQ is a 34-item scale designed to assess an individual’s tendency
to inhibit behavior to avoid punishment, novel stimuli and the lack of rewards. This dimension
is composed of four subscales: Anticipatory Worry (10 items), Fear of Uncertainty (seven items),
Fatigability (seven items) and Shyness with Strangers (10 items). Respondents are asked to indi-
cate in true/false response format whether statements pertaining to HA describe them. This scale
dimension has been found to show good internal validity (a = 0.85 for Caucasian males and
females, and 0.77 and 0.80 for Black males and females, respectively, in a national probability
sample) and test–retest reliability (r = 0.79 and 0.51 in national probability and depressed out-
patient samples, respectively). This dispositional measure of avoidance was also employed to
assess the convergent validity of the constructed avoidance measure.

3.2.4. Beck Depression Inventory—II (BDI-II; Beck, Steer, & Brown, 1996)
The BDI-II is a 21-item self-report inventory that assesses the presence and severity of depress-
ive symptoms over a 2-week period. Each item is rated on a scale of 0–3, with total scores ranging
from 0 to 63. The BDI-II shows high internal consistency (a = 0.93 for college students and 0.92
for outpatients) and adequate content, factorial, convergent and divergent validity. This measure
has also been shown to discriminate between individuals with and without a depressive disorder
diagnosis. The BDI-II was employed to assess the primary criterion relationship of interest in this
study between the new measure of avoidance and depression.

3.2.5. Depression Adjective Checklist—‘General’ Form (DACL; Lubin, 1967)

The DACL is a 34-item checklist designed to measure self-reported depressive mood. The score
is computed by summing the depressive mood items that are endorsed and the positive affect
items that are not endorsed. This instrument (Form E) shows good internal consistency (a =
0.83 for males and 0.88 for females) and convergent validity. The ‘general’ form of the DACL,
which asks participants to respond according to how they ‘generally’ feel, as opposed to how
they feel that day, was employed in order to attempt to measure participants’ more enduring mood
states. The DACL was used as a supplementary depression criterion measure.
300 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313

3.2.6. State–Trait Anxiety Inventory—Trait Version (STAI; Spielberger, Gursuch, & Lushene,
1970)
The STAI—Trait version is a 20-item self-report scale designed to measure trait anxiety or a
more general anxiety-proneness. Items are rated on a four-point scale reflecting the frequency
with which respondents feel what is reflected in the item from “Almost Never” to “Almost
Always”. This scale shows good internal consistency (a = 0.86–0.92 across samples of high
school and undergraduate male and female students) and test–retest reliability (r = 0.76 for female
and 0.86 for male undergraduate students over a 20 day test–retest interval). The STAI—Trait
version shows good convergent validity with other trait measures of anxiety. This scale was
employed to assess the criterion relationships between the new measure of avoidance and trait anx-
iety.

3.3. Procedure

3.3.1. Scale design and development

Items designed to sample the hypothesized dimensions of avoidance were developed through
review and adaptation of items from other avoidance measures as well as theoretically based
writing of new items. Eight positively keyed items, written as unidimensional self-descriptor
items, were developed to reflect each of the eight possible dimension combinations, for a total
of 64 items. The instructions stated, “Different people use different strategies to deal with prob-
lems and situations in their lives. Below are a number of strategies that people may use to deal
with situations and problems”, and directed participants to indicate how each statement applied
to them “in general” in order to obtain a trait measure of avoidance. The response format entailed
a five-point Likert-type scale reflecting similarity ratings (1 = not at all true for me, 2 =
somewhat true for me, 3 = moderately true for me, 4 = very much true for me, and 5 =
extremely true for me).
Content validation of the written items was conducted through backward coding of item content
dimensions by an expert who was unaware of the writer’s coding of the content under examin-
ation. The items for which dimensional ratings were discrepant between the writer and the expert
were modified and the item pool was recoded by the expert. Following this modification, item
rating agreement for the cognitive versus behavioral, social versus nonsocial, and active versus
passive dimensions were 100%, 99% and 92.5%, respectively. The readability level of the final
item pool and instructions, using the Flesch–Kincaid method in Microsoft Word 2000, was esti-
mated at a Grade 6.5 level.

3.3.2. Scale evaluation

After obtaining informed consent from participants, the avoidance items and convergent, diver-
gent and criterion-related measures were administered to participants in a group format. In order
to control for order effects, the questionnaires were administered in a counterbalanced order and
order effects were tested. Basic demographic information, including participants’ gender, age and
race, was collected for the purposes of sample description and supplementary analyses.
The construct validity of the new avoidance scale was evaluated through examination of the
factor structure and internal consistency of the scale as well as through examination of the relation-
ships between the new scale and the convergent and divergent measures used in the study. The
 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313 301

criterion-related (concurrent) validity of the constructed avoidance measure was tested through
examination of the relationships between the new avoidance measure and the depression and
anxiety measures. Although the primary criterion relationship of interest was that between avoid-
ance and depression, the strong relationship and high rates of comorbidity between anxiety and
depression (Mazer & Cloninger, 1990), as well as the opportunity to compare the criterion
relationship of interest to that of an established criterion relationship between avoidance and
anxiety (Barlow, 2002), supported the utility of the inclusion of the anxiety measure.

4. Results

All ANOVAs conducted to test for questionnaire order effects on the new avoidance scale,
depression and anxiety scores were nonsignificant. The section that follows describes the item
reduction methods used to construct the new scale, and the final factor solution, reliability, conver-
gent, divergent and criterion-related validity of the scale. The convergent, divergent and criterion-
related analyses were conducted by gender and are reported as such. In addition, ancillary gender
and racial analyses are presented.

4.1. Factor analysis

Item frequencies, means and standard deviations were examined prior to factor analysis to
ensure adequate discriminability of items. No items were removed on the basis of the removal
criteria of a standard deviation less than 0.75 or 75% endorsement of a single Likert response.
Exploratory factor analysis using maximum likelihood extraction and Varimax rotation1 was
employed to determine the factor structure of the 64 avoidance items, and to direct item analysis
and reduction. The results, based on the criteria of eigenvalues greater than 1 and Cattell’s scree
analysis, suggested a four factor solution. Through an iterative process, items that did not load
substantively (below 0.30) or discriminantly (difference in substantive item loadings across factors
less than 0.05) on the interpretable factors, or that showed low (below 0.30) item total correlations,
either with the total scale or respective subscale, were dropped and further analyses were conduc-
ted.
After 11 iterations, the final analysis producing the best solution was comprised of 31 items.
Four interpretable factors emerged, with eigenvalues of 8.54, 2.44, 1.64 and 1.32. Together, these
factors accounted for a total of 44.95% of the variance in item responses (Factor 1 = 27.54%,
Factor 2 = 7.87%, Factor 3 = 5.30% and Factor 4 = 4.24%). The four factors that emerged were
labeled as follows: Factor 1 (eight items)—Behavioral Social, Factor 2 (10 items)—Cognitive
Nonsocial, Factor 3 (seven items)—Cognitive Social and Factor 4 (six items)—Behavioral Nonso-

1
Prior to Varimax rotation, an oblique rotation was conducted such that the correlations among factors in the original item pool
could be examined. The magnitudes of the correlations between factors, with the exception of one correlation (0.33), were all below
0.30, suggesting that Varimax rotation was appropriate for this analysis.
302 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313

cial (see Table 1 for factor loadings table). The resulting overall scale, reflecting the factors that
emerged, was entitled the Cognitive–Behavioral Avoidance Scale (CBAS).
Examples of item content and their factor loadings for the four factors are presented below.
Items loading highly on the Behavioral Social factor included “I tend to make up excuses to get
out of social activities” (0.73) and “I avoid attending social activities” (0.65). The Cognitive
Nonsocial factor included items such as “While I know that I have to make some important
decisions about school/work, I just do not get down to it” (0.63) and “I distract myself when I

Table 1
Factor loadings for CBAS items

Item number and condensed description Behavioral Cognitive Cognitive Behavioral

Social Nonsocial Social Nonsocial

1. (4) Avoid attending social activities 0.65 0.15 – 0.15

2. (12) Uncertain future, do not think about what want – 0.37 0.22 0.31
3. (13) Like to achieve, but have to accept limits – – – 0.34
4 (14) Fail to follow through with achievement goals – 0.63 0.18 0.26
5. (18) Avoid disappointment, not serious work/school – 0.39 0.17 0.12
6. (19) Rather than try new activities, do what know 0.29 0.17 – 0.54
7. (25) Turn down opportunities education/career 0.16 0.35 – –
8. (27) Do not answer phone in case social invitation 0.46 – 0.25 0.19
9. (28) Quit activities that challenge too much 0.25 0.13 0.15 0.54
10. (30) Try not think problems in relationships 0.16 – 0.41 –
11. (33) Think not able to complete challenging tasks 0.13 0.32 0.14 0.49
12. (34) Rather than getting out, sit home watch TV 0.32 0.23 0.16 0.41
13. (35) Avoid trying activities potential for failure 0.27 0.16 0.17 0.64
14. (39) Do not go to events when people do not know 0.60 – 0.14 0.29
15. (40) Instead thinking, tell self prefer to be alone 0.58 0.21 0.30 –
16. (4l) Fail to address tension in friendship 0.16 0.21 0.45 0.15
17. (42) Want to leave social gatherings 0.72 0.12 – 0.13
18. (45) Try not think about how improve performance – 0.35 0.16 0.14
19. (47) Try not think about future/what do with life 0.10 0.42 – –
20. (48) Wait tension in relationships, hope go away 0.17 – 0.80 0.15
21. (49) Make excuses to get out of social activities 0.73 0.11 0.18 0.30
22. (50) Nothing can do improve relationship problems 0.24 0.29 0.35 –
23. (51) Turn down socializing with opposite sex 0.55 0.22 0.21 0.16
24. (52) Remain to self during social gatherings 0.52 0.10 0.16 0.18
25. (53) Avoid making decisions about future – 0.57 0.18 0.18
26. (55) Confusion in relationships, not try figure out 0.20 0.24 0.47 0.11
27. (57) Not think how solve family problems—useless 0.18 0.23 0.37 0.13
28. (58) Should make decisions relationships, let go on 0.22 0.20 0.63 0.16
29. (59) Distract self when think about performance 0.15 0.44 0.20 0.23
30. (60) Decisions work/school, do not get down to it – 0.63 0.14 –
31. (63) Avoid tasks that are really important 0.12 0.54 0.12 0.14

Eigenvalue 8.54 2.44 1.64 1.32

% Variance 27.54 7.87 5.30 4.24

Note. Items loading substantively on factors are presented in bold-face type. Loadings less than 0.10 were suppressed.
 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313 303

start to think about my work/school performance” (0.44). Representative items from the Cognitive
Social factor consisted of items such as “I just wait out tension in my relationships hoping that
it will go away” (0.80) and “I try not to think about problems in my personal relationships”
(0.41). Items loading highly on the Behavioral Nonsocial factor included the following: “I avoid
trying new activities that hold the potential for failure” (0.64) and “I quit activities that challenge
me too much” (0.54).

4.2. Internal consistency

Factor/subscale intercorrelations and correlations with the total scale, coefficient alphas and
item total correlations were computed to assess the internal consistency of the scale. Moderate
intercorrelations, ranging from 0.39 to 0.57, were found between subscales, suggesting that the
subscales are measuring distinct, but related, constructs (see Table 2). High correlations, ranging
from 0.78 to 0.80, were found between subscale scores and the total scale score. Taken together,
these findings indicate that a composite avoidance score is appropriate. Coefficient alphas for the
subscales were all adequate (0.86, 0.80, 0.78 and 0.75 for Factors 1–4, respectively) and coefficient
alpha for the total scale was 0.91. Item total correlations for items with their respective subscale
and with the total scale were typically in the 0.40–0.60 range. With the exception of one item,
all items showed item total correlations, with both subscale and total scale, greater than 0.30.
These results suggest that the CBAS has good internal consistency.

4.3. Test–retest reliability

Three-week test–retest reliabilities (correlations) were computed to determine if CBAS scores

showed stability over time. The 3-week test–retest reliability coefficient for the total scale was
0.92. The reliability coefficients for the Cognitive Nonsocial, Behavioral Nonsocial, Behavioral
Social and Cognitive Social subscales were 0.94, 0.88, 0.86 and 0.58, respectively, reflecting high
stability in all but the Cognitive Social subscale and moderate stability for this subscale. These
findings provide support for the trait conceptualization of the construct of avoidance espoused by
the CBAS.

Table 2
Subscale intercorrelations for the CBAS (total sample)

Subscale Behavioral Cognitive Cognitive Behavioral

Social Nonsocial Social Nonsocial

Behavioral Social –
Cognitive Nonsocial 0.39 –
Cognitive Social 0.53 0.53 –
Behavioral Nonsocial 0.57 0.52 0.45 –

Note. All correlations are significant at the p ⬍ 0.01 level.

304 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313

4.4. Convergent and divergent validity

The construct validity of the scale was further evaluated through the examination of correlations
between the CBAS and convergent avoidance and divergent approach coping measures. The
CBAS showed moderate correlations with other avoidance measures (r = 0.34 for females and
r = 0.30 for males on the CRI Total Avoidance Scale, r = 0.41 for females and r = 0.50 for
males on the Escape–Avoidance Scale of the WCQ, and r = 0.56 for females and r = 0.63 for
males on the HA dimension of the TPQ) (see Table 3). Of note, the CBAS showed higher corre-
lations with the dispositional measure of avoidance (HA dimension of the TPQ) than the situ-
ational avoidance measures (CRI Total Avoidance Scale, and the Escape–Avoidance Scale of the
WCQ). While it seems intuitive that trait measures would correlate more highly with one another
than with a situational coping measure, it appears that the poor convergent validity for the
behavioral avoidance scales of the CBAS with the behavioral avoidance scales of the CRI attenu-
ated the overall convergent relationship between these measures. Given the varied definitions of

Table 3
Convergent and divergent correlations

Behavioral Cognitive Cognitive Behavioral Total

Social Nonsocial Social Nonsocial CBAS

Convergent
Females
CRI (Avoidance) 0.22 0.34 0.29 0.28 0.34
(Cognitive avoidance) 0.22 0.41 0.37 0.37 0.41
(Behavioral avoidance) 0.12∗ 0.10∗∗ 0.07∗∗ 0.05∗∗ 0.11∗
Escape/Avoidance Scale 0.32 0.34 0.32 0.37 0.41
Harm Avoidance Scale 0.56 0.38 0.34 0.55 0.56
Males
CRI (Avoidance) 0.12∗∗ 0.29 0.41 0.06∗∗ 0.30
(Cognitive avoidance) 0.13∗∗ 0.31 0.47 0.13∗∗ 0.35
(Behavioral avoidance) 0.05∗∗ 0.14∗ 0.15∗ 0.05∗∗ 0.11∗∗
Escape/Avoidance Scale 0.31 0.32 0.49 0.41 0.50
Harm Avoidance Scale 0.62 0.32 0.38 0.56 0.63

Divergent
Females
CRI (Approach) ⫺0.05∗∗ ⫺0.13∗∗ ⫺0.17 ⫺0.16 ⫺0.15
(Cognitive approach) ⫺0.01∗∗ ⫺0.02∗∗ ⫺0.08∗∗ ⫺0.07∗∗ ⫺0.05∗∗
(Behavioral approach) ⫺0.08∗∗ ⫺0.21 ⫺0.22 ⫺0.21 ⫺0.21
Males
CRI (Approach) ⫺0.15∗ ⫺0.18∗ ⫺0.21 ⫺0.12∗∗ ⫺0.22
(Cognitive approach) ⫺0.04∗∗ ⫺0.06∗∗ ⫺0.08∗∗ ⫺0.08∗∗ ⫺0.08∗∗
(Behavioral approach) ⫺0.22 ⫺0.24 ⫺0.28 ⫺0.14∗ ⫺0.30

Note. Unmarked correlations are significant at the p ⬍ 0.01 level.

∗
Correlation was not significant at the p ⬍ 0.01 level.
∗∗
Correlation was not significant at the p ⬍ 0.05 level.
 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313 305

the CBAS behavioral avoidance scales and the CRI behavioral avoidance scales (the latter con-
sisting of Seeking Alternative Rewards and Emotional Discharge) and the convergent relationships
shown with other avoidance measures, this finding does not present any considerable challenge
to the convergent validity of the CBAS behavioral avoidance scales. The CBAS was found to
show consistently negative correlations of low magnitude with the approach coping measure of
the CRI (r = ⫺0.15 for females and r = ⫺0.22 for males on the CRI Total Approach scale).
These findings indicate that the CBAS shows good convergent and divergent validity.

4.5. Criterion correlations

Correlations with criterion measures were conducted to determine if avoidance, as measured

by the CBAS, was related to depression and anxiety. As expected, the CBAS showed moderate
correlations with depression (r = 0.48 for both females and males on the BDI-II) (see Table 4).
All subscales of the CBAS showed substantive relationships with BDI-II scores, with the Cogni-
tive Nonsocial subscale showing the strongest relationship with depression (r = 0.45 for females
and r = 0.41 for males). The correlation between the Behavioral Nonsocial subscale and BDI-II
measured depression was found to be higher, although not significantly, for females (r = 0.41)
than males (r = 0.34). Correlations between the CBAS and the DACL were also significant, but
somewhat lower than the correlations between the CBAS and the BDI-II.
As expected, the CBAS was found to show moderate to strong correlations with anxiety (r =
0.58 for females and r = 0.59 for males on the STAI). All subscales were found to show at least
moderate correlations with the STAI. The results indicated that the relationship between
Behavioral Nonsocial avoidance and anxiety was stronger, albeit not significantly, for females
(r = 0.53) than males (r = 0.45). Overall, it is noteworthy that the magnitude of the criterion-
related correlations between the CBAS and depression approached those of the correlations
between the CBAS and anxiety.

Table 4
Criterion correlations

Behavioral Cognitive Cognitive Behavioral Total

Social Nonsocial Social Nonsocial CBAS

Females
BDI-II 0.36 0.45 0.36 0.41 0.48
DACL 0.33 0.35 0.24 0.38 0.40
STAI 0.48 0.49 0.40 0.53 0.58
Males
BDI-II 0.30 0.41 0.39 0.34 0.48
DACL 0.43 0.31 0.26 0.29 0.44
STAI 0.46 0.46 0.38 0.45 0.59

Note. All correlations are significant at the p ⬍ 0.01.

306 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313

4.6. Ancillary analyses

Ancillary analyses employing one-way ANOVAs to examine the use of avoidance strategies,
as assessed by the CBAS, across gender and race were conducted. In addition, the correlations
between the CBAS and the criterion-related measures were examined separately across sizeable
racial groups.
It was found that males obtained higher total CBAS scores than females (F(1,389) = 6.26, p
= 0.01, ḣ2 = 0.02). Males were also found to score significantly higher than females across most
of the subscales with the exception of the Behavioral Nonsocial subscale, which showed no sig-
nificant differences across gender.
The one-way ANOVA conducted to examine differences across Caucasian (N = 242), Asian
(N = 110) and East Indian (N = 22) groups on the total CBAS was significant (F(2,371) = 7.69,
p ⬍ 0.001, ḣ2 = 0.04). Follow-up analyses indicated that Asian individuals reported significantly
greater use of avoidance strategies than Caucasian individuals (t(350) = 3.87, p ⬍ 0.001). The
analysis of group differences across subscales showed that there were significant differences across
racial groups on the Behavioral Social (F(2,371) = 6.59, p = 0.002, ḣ2 = 0.03) and Behavioral
Nonsocial (F(2,371) = 9.37, p ⬍ 0.001, ḣ2 = 0.05) subscales. Follow-up testing showed that Asian
individuals scored significantly higher than Caucasian individuals on the Behavioral Social
(t(350) = 3.61, p ⬍ 0.001) and Behavioral Nonsocial (t(350) = 4.27, p ⬍ 0.001) subscales. No
significant racial group differences were found across the Cognitive Social and Cognitive Nonso-
cial subscales.
Important correlations between CBAS scores and the criterion measures for the sizeable racial
groups are presented below. It was found that, for Caucasian individuals, the CBAS showed
moderate correlations with depression, as assessed by the BDI-II, and anxiety (r = 0.46 and 0.58,
respectively) and that the criterion relationships were consistent in general magnitude across the
CBAS subscales. For Asian individuals, it was found that CBAS scores related significantly with
depression (r = 0.49) and anxiety scores (r = 0.55) and that the Cognitive Nonsocial subscale
showed a stronger relationship with depression (r = 0.52) than the remaining subscales (which
ranged from r = 0.29 to 0.38). This finding suggests that Cognitive Nonsocial avoidance had a
particularly strong association with depression for Asian individuals.
In contrast to the other racial groups, it was found that the correlation between CBAS scores
and depression (r = 0.47) was higher than the correlation between CBAS scores and anxiety (r
= 0.24) for East Indian individuals, In addition, it was found that the Cognitive Nonsocial and
Cognitive Social avoidance subscales showed substantive correlations with depression (r = 0.73
and 0.55, respectively). In contrast, the relationships between Behavioral Social and Behavioral
Nonsocial avoidance with depression were nonsignificant for this group. These findings suggest
that for East Indian persons, avoidance strategies related more specifically to depression than
anxiety and that cognitive avoidance in particular was associated with depression.

5. Discussion

The primary goal of this study was to develop and validate a multidimensional scale of avoid-
ance. In line with hypotheses, it was found that cognitive versus behavioral and social versus
 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313 307

nonsocial dimensions emerged in empirical analysis as important dimensions of avoidance. How-

ever, the hypothesized active versus passive dimension of avoidance did not emerge in this analy-
sis. The final resulting factors/subscales, which included Behavioral Social, Cognitive Nonsocial,
Cognitive Social and Behavioral Nonsocial avoidance, and the total Cognitive–Behavioral Avoid-
ance Scale (CBAS) all showed adequate internal consistency and 3-week test–retest reliability.
The CBAS total scale and subscales showed good convergent and divergent validity. It was
found that the CBAS correlated more strongly with the convergent measure of avoidance which
reflected a trait measure, as opposed to the convergent measures assessing situational coping. The
CBAS also showed moderate correlations with the depression and anxiety criterion measures.
Although not a primary focus of this study, gender and racial differences were found to emerge
in the use of avoidance strategies and the strength and/or pattern of the relationships between
CBAS scores and the criterion depression and anxiety measures.

5.1. Dimensions of avoidance

The finding of the emergence of the cognitive versus behavioral dimension as an important
dimension of the construct of avoidance is consistent with the findings of Blalock and Joiner
(2000). These authors employed confirmatory factor analysis to test their hypothesis that the
factors of cognitive and behavioral avoidance coping would better capture variability in university
students’ responses to the four heterogeneous avoidance scales of the CRI, which included Cogni-
tive Avoidance, Acceptance/Resignation, Seeking Alternative Rewards and Emotional Discharge,
than would a single avoidance factor. Consistent with their hypothesis, the authors found, using
a sample of 72 male and 107 female university students, that this two-factor model of avoidance
provided a superior fit to the data compared to the single avoidance factor model.
These studies clearly indicate that cognitive and behavioral methods of avoiding situations and
problems delineate the construct of avoidance. Avoidance then encompasses not just what one
does behaviorally to refrain from or escape from situations, but also cognitive measures one may
take to attempt to avoid or escape thinking about situations or problems. This finding implies that
measures assessing the broad domain of avoidance must attempt to quantify cognitive avoidance
in addition to behavioral avoidance in order to sample the domain of the avoidance construct.
The empirical emergence of the social versus nonsocial dimension in characterizing the avoid-
ance construct conveys special significance, given that coping research up to this point has gener-
ally failed to distinguish between coping which occurs within a social versus nonsocial domain.
This finding indicates that the domain in which a problem or situation presents to an individual
is important in determining the nature of the individual’s coping response. These results are in
line with the literature on the salience of social versus achievement domains characterizing life
events in the prediction of the interaction between the personality vulnerability characteristics of
sociotropy and autonomy and life events in depression onset and relapse (Coyne & Whiffen,
1995). The present study suggests that measures of avoidance should specify the social or nonso-
cial nature of the situation or problem with which individuals cope in order to convey important
information about the individual’s manner of dealing with situations and problems and the impact
of these different strategies on adjustment.
Contrary to hypotheses, it was found that the active versus passive dimension did not emerge
as an important dimension of avoidance in empirical analysis. It is conceivable that the active
308 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313

versus passive dimension was not distinguished in this analysis given that avoidance is generally
a passive strategy whereby a person fails to invoke a course of action in which the problem is
dealt with directly. Thus, it may be that this level of passivity characterizing avoidance makes it
difficult to distinguish between passive and active avoidance strategies at the item level.

5.2. Stability of the CBAS

The stability of CBAS scores over a 3-week period suggests support for the hypothesized trait
conceptualization of avoidance. This finding is consistent with research in the coping literature
that has typically shown significant, low to moderate correlations between dispositional coping
styles and situational coping responses (Carver & Scheier, 1994; Carver et al., 1989). It should
be noted that the demonstration of the stability of CBAS scores over a longer time frame, perhaps
in the range of 6–12 months, would have provided more convincing support for a trait concep-
tualization of avoidance. This finding is also important in the sense that employment of a dispo-
sitional measure of avoidance conveys particular advantage over a situational measure for
researchers who may desire to examine avoidance in a longitudinal manner as a potential risk
factor for depression. Further research is needed to examine the longer-term temporal stability of
the CBAS, and its relationship to other constructs over time.

5.3. Criterion relationships of avoidance with depression and anxiety

Of primary importance in this study are the findings of moderate correlations between avoidance
strategies assessed by the CBAS and depression. These results indicate that the employment of
higher levels of both behavioral and cognitive avoidance is associated with higher levels of
depressive symptoms. The finding that cognitive nonsocial avoidance showed the highest corre-
lation with BDI-II scores across gender is in line with previous research finding cognitive avoid-
ance coping specifically to be related to distress (Blalock & Joiner, 2000) and the literature on the
potential detrimental effects of thought suppression on well-being previously discussed (Beevers et
al., 1999; Rassin et al., 2000). The cross-sectional nature of this study, however, does not allow
for any inferences as to the causal nature of the relationship between avoidance and depression.
However, the magnitude of the correlations shown between various forms of avoidance and
depression suggests that longitudinal research to examine the nature of the relationship between
avoidance and depression is warranted.
It should also be noted that the magnitude of the relationship between avoidance and depression
approaches that of the relationship between avoidance and anxiety in this study. This finding is
significant for two reasons. First of all, given the prominent role that avoidance occupies in the
conceptualization of anxiety disorders (Barlow, 2002), this finding suggests that the construct of
avoidance may deserve more consideration in the conceptualization of depressive disorders. Sec-
ondly, the strong relationship between avoidance and both depression and anxiety suggests the
possibility that avoidance may be a common factor accounting for the high level of covariation
in depressive and anxious symptoms and comorbidity at the disorder level (Mazer & Cloninger,
1990). These theories reflect mere speculation at this point. However, the potential implications
of these theories, should they be supported in future research, provide the rationale for their inves-
tigation.
 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313 309

5.4. Gender and racial differences in avoidance employment and relationships

Ancillary analyses showed that gender and racial differences were found in the employment
of avoidance strategies and the relationship between avoidance and criterion measures of
depression and anxiety. It was found that males generally employed higher levels of avoidance
strategies, as assessed by the CBAS, than females. This finding is not in line with socialization
theory, which would postulate greater use of avoidance-type strategies for females than males
given socialization practices reinforcing passive and emotion-focused methods of coping for
females, compared to socialization practices reinforcing active, problem-focused coping for males
(Ben-Zur & Zeidner, 1996; Pearlin & Schooler, 1978), However, other research has failed to
provide either clear support or challenge for this theory, as findings of gender differences in the
employment of avoidance strategies have generally been mixed (Ben-Zur & Zeidner, 1996; Carver
et al., 1989). It was also found that the employment of behavioral nonsocial avoidance strategies
was more strongly associated with depression and anxiety for females compared to males,
although the magnitudes of these differences were not significant. Thus, it seems that this type
of avoidance may be slightly more maladaptive for females than males.
The results also showed that Asian individuals employed avoidance strategies to a greater extent
than Caucasian individuals. Other research has produced similar findings (Bjorck, Cuthbertson,
Thurman, & Soon Lee, 2001). Different patterns of associations between types of avoidance and
depression and anxiety were also found across racial groups. However, given that level of accultur-
ation is an important variable to be considered in the examination of racial or ethnic differences
across variables of interest (Dana, 1996) and that this variable was not assessed, results pertaining
to racial differences were presented in descriptive form only and are not interpreted. The patterns
of racial differences in the employment of avoidance strategies and relationships between types
of avoidance and depression found in this study, however, suggest that further research in this
area, considering level of acculturation, should be conducted.

5.5. Limitations and directions for future research

The primary limitations of this research include the employment of a student sample, as opposed
to a sample of clinically depressed individuals, to examine important phenomena pertaining to
depression and the exclusive use of self-report measures to assess variables of interest. The use
of student samples to inform depression theory has been criticized by Coyne (1994) on the basis
that depression seen in student samples is qualitatively different than depression seen in clinical
samples, that self-reported distress and clinical depression have different psychosocial correlates
and that the utilization of student samples to draw conclusions about depression trivializes the
nature of clinical depression. Subsequent research, however, examining student and clinical
depression in a comparative fashion (Cox, Enns, Borger, & Parker, 1999; Flett, Vredenburg, &
Krames, 1997) has shown that more evidence exists to support, rather than disconfirm, that differ-
ences between student and clinical depression are more quantitative than qualitative in nature.
The use of self-report measures to assess all variables of interest in this study reflects a limi-
tation in that common-method variance may be affecting the relationships between variables.
However, it should be noted that the nature of a number of the variables under investigation in
this study complicates the use of alternative behavioral or other-reported measures to quantify
310 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313

these variables of interest. In line with claims of Lonigan, Hooe, David and Kistner (1999),
affective states, such as depression or anxiety, are unique internal states accessible only to the
individual and, therefore, best communicated through self-report. In addition, while a behavioral
task could be used to obtain a measure of behavioral avoidance, cognitive avoidance reflects a
covert event and thus cannot be observed or reported by others. However, it should be noted that
new methodology assessing the concreteness of individuals’ reported problem elaborations has
been suggested to reflect a valid non-self-report measure of cognitive avoidance in the anxiety
literature (Stöber & Borkovec, 2002; Stöber, Tepperwien, & Staak, 2000).
An additional limitation reflects the cross-sectional design of this study. As is the case with all
cross-sectional research, temporal precedence as a necessary condition to infer causality cannot
be established. Thus, no conclusions about the existence or direction of causality between avoid-
ance and depression can be discerned.
In consideration of the above limitations, a number of directions for future research are sug-
gested. One of the first logical areas of research extension would be the cross-validation of the
CBAS using a community and/or clinical sample. Cross-validation of the scale using community
and/or clinical participants would allow for the examination of whether the factor structure and
reliability and validity of the CBAS is such that this new instrument can be used appropriately
with these populations. Another logical extension of this research would involve the longitudinal
examination of the relationship between avoidance and depression. This type of research would
allow the examination of the nature of the relationship between avoidance and depression, such
that it could be determined if avoidance seems to function as a risk factor for depression or a
consequence or mere correlate of depression. It would be useful in future research to examine
how avoidance relates to other variables frequently employed in depression research, such as the
cognitive vulnerability characteristics of sociotropy and autonomy, rumination and distraction
response styles to depression, and self-efficacy. The CBAS might also be employed in research
on generalized anxiety given the recent attention on cognitive avoidance in this area (Stöber &
Borkovec, 2002; Stöber et al., 2000) as well as to examine whether avoidance may be a factor
accounting for comorbidity between depression and anxiety.
In the effort of construct validation, a number of directions for future research are indicated.
The CBAS should be compared to other measures of avoidance existing in the literature such as
the Avoidance Style Scale of the Social Problem-Solving Inventory—Revised (SPSI-R; D’Zurilla,
Nezu, & Maydeu-Olivares, 2002) and the Approach-Avoidance Scale of the Problem-Solving
Inventory (PSI; Heppner, 1988). In addition, it would be useful to devise a behavioral avoidance
task that could be compared to individuals’ self-report of avoidance in order to determine if self-
reported avoidance corresponds to the in vivo employment of avoidance strategies. It might also
be useful to determine if less concrete elaborations of problems by individuals, as is interpreted
to reflect cognitive avoidance in the anxiety literature (Stöber & Borkovec, 2002; Stöber et al.,
2000), is related to self-reported cognitive avoidance.
In conclusion, this study involved the construction of a multidimensional avoidance scale called
the ‘Cognitive–Behavioral Avoidance Scale’ which has potential utility for researchers in the area
of depression. It was found that the following types of avoidance emerged in empirical analysis:
Behavioral Social, Cognitive Nonsocial, Cognitive Social and Behavioral Nonsocial avoidance.
All of these subscales, along with the total CBAS, showed moderate relationships with depression.
It will thus be important to pursue longitudinal investigation of the relationship between avoidance
 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313 311

and depression for the purpose of determining if avoidance is a risk factor for depression. Should
avoidance be found to be a risk factor for depression, it would follow that the construct of avoid-
ance would warrant address in the psychosocial conceptualization and treatment of depression.

Acknowledgements

This research was supported through grants awarded to N.D. Ottenbreit from the Alberta Heri-
tage Foundation for Medical Research and the Alberta Heritage Scholarship Fund. The assistance
of Ms. Catherine Munns with data collection and scoring is also gratefully acknowledged.

References

Barlow, D. A. (2002). Anxiety and its disorders: The nature and treatment of anxiety and panic (2nd ed.). New York:
Guilford Press.
Beck, A. T., Rush, A. J., Shaw, B. F., & Emery, G. (1979). Cognitive therapy of depression. New York: Guilford Press.
Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Beck Depression Inventory—II manual. San Antonio, TX: The
Psychological Corporation.
Beevers, C. G., Wenzlaff, R. M., & Hayes, A. M. (1999). Depression and the ironic effects of thought suppression:
therapeutic strategies for improving mental control. Clinical Psychology: Science and Practice, 6, 133–148.
Ben-Zur, H., & Zeidner, M. (1996). Gender differences in coping reactions under community and daily routine con-
ditions. Personality and Individual Differences, 20, 331–340.
Bjorck, J. P., Cuthbertson, W., Thurman, J. W., & Soon Lee, Y. (2001). Ethnicity, coping, and distress among Korean
Americans, Filipino Americans and Caucasian Americans. The Journal of Social Psychology, 141, 421–442.
Blalock, J. A., & Joiner, T. E. (2000). Interaction of cognitive avoidance coping and stress in predicting
depression/anxiety. Cognitive Therapy and Research, 24, 47–65.
Carver, C. S., & Scheier, M. F. (1994). Situational coping and coping dispositions in a stressful transaction. Journal
of Personality and Social Psychology, 66, 184–195.
Carver, C. S., Scheier, M. F., & Weintraub, J. K. (1989). Assessing coping strategies: a theoretically-based approach.
Journal of Personality and Social Psychology, 56, 267–283.
Chien, A. J., & Dunner, D. L. (1996). The Tridimensional Personality Questionnaire in depression: state versus trait
issues. Journal of Psychiatric Research, 30, 21–27.
Clark, D. A., Beck, A. T., & Alford, B. A. (1999). Scientific foundations of cognitive theory and therapy of depression.
New York: John Wiley and Sons, Inc.
Cloninger, C. R. (1987). A systematic method for clinical description and classification of personality variants: a
proposal. Archives of General Psychiatry, 44, 573–588.
Cloninger, C. R., Przybeck, T. R., & Svrakic, D. M. (1991). The Tridimensional Personality Questionnaire: US norma-
tive data. Psychological Reports, 69, 1047–1057.
Cloninger, C. R., Przybeck, T. R., Svrakic, D. M., & Wetzel, R. D. (1994). The Temperament and Character Inventory
(TCI): A guide to its development and use. St. Louis, MO: Center for Psychobiology of Personality.
Cox, B. J., Enns, M. W., Borger, S. C., & Parker, J. D. (1999). The nature of the depressive experience in analogue
and clinically depressed samples. Behaviour Research and Therapy, 37, 15–24.
Coyne, J. C. (1994). Self-reported distress: analog or ersatz depression? Psychological Bulletin, 116, 29–45.
Coyne, J. C., & Whiffen, V. E. (1995). Issues in personality as diathesis for depression: the case of sociotropy–
dependency and autonomy–self-criticism. Psychological Bulletin, 118, 358–378.
Dana, R. H. (1996). Culturally competent assessment practice in the United States. Journal of Personality Assessment,
66, 472–487.
Davila, J. (1993). Attachment, interpersonal problem-solving, and stress in depression: An integrated developmental
approach. Doctoral dissertation, University of California, 1993. Dissertation Abstracts International, 5383.
312 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313

D’Zurilla, T. J., Chang, E. C., Nottingham, E. J., & Faccini, L. (1998). Social problem-solving deficits and hopelessness,
depression, and suicidal risk in college students and psychiatric patients. Journal of Clinical Psychology, 54,
1091–1107.
D’Zurilla, T. J., & Nezu, A. M. (1999). Problem-solving therapy: A social competence approach to clinical intervention.
(2nd ed.). New York: Springer Publishing Co.
D’Zurilla, T. J., Nezu, A. M., & Maydeu-Olivares, A. (2002). Manual for the Social Problem-Solving Inventory—
Revised. North Tonawanda, NY: Multi-Health Systems.
Ferster, C. B. (1973). A functional analysis of depression. American Psychologist, 28, 857–870.
Flett, G. L., Vredenburg, K., & Krames, L. (1997). The continuity of depression in clinical and nonclinical samples.
Psychological Bulletin, 121, 395–416.
Folkman, S., & Lazarus, R. S. (1986). Stress processes and depressive symptomatology. Journal of Abnormal Psy-
chology, 95, 107–113.
Folkman, S., & Lazarus, R. S. (1988). Ways of Coping Questionnaire Manual. Redwood City, CA: Consulting Psychol-
ogists Press.
Hansenne, M., Pitchot, W., Gonzalez Moreno, A., Machurot, P. -Y., & Ansseau, M. (1998). The Tridimensional Person-
ality Questionnaire (TPQ) and depression. European Psychiatry, 13, 101–103.
Hansenne, M., Pitchot, W., Gonzalez Moreno, A., Reggers, I., Machurot, P. -Y., & Ansseau, M. (1997). Harm avoidance
dimension of the Tridimensional Personality Questionnaire and serotonin-1A activity in depressed patients. Biologi-
cal Psychiatry, 42, 959–961.
Heppner, P. P. (1988). The Problem-Solving Inventory. Palo Alto, CA: Consulting Psychologists Press.
Holahan, C. J., & Moos, R. H. (1986). Personality, coping, and family resources in stress resistance: a longitudinal
analysis. Journal of Personality and Social Psychology, 51, 389–395.
Krantz, S. E., & Moos, R. H. (1988). Risk factors at intake predict nonremission among depressed patients. Journal
of Consulting and Clinical Psychology, 56, 863–869.
Krohne, H. W. (1996). Individual differences in coping. In M. Zeidner, & N. S. Endler (Eds.), Handbook of coping:
Theory, research, and applications (pp. 381–409). New York: John Wiley and Sons, Inc.
Kuyken, W., & Brewin, C. R. (1994). Stress and coping in depressed women. Cognitive Therapy and Research, 18,
403–412.
Lonigan, C. S., Hooe, E. S., David, C. F., & Kistner, J. A. (1999). Positive and negative affectivity in children:
confirmatory factor analysis of a two-factor model and its relation to symptoms of anxiety and depression. Journal
of Consulting and Clinical Psychology, 67, 374–386.
Lubin, B. (1967). Manual for the Depression Adjective Checklist. San Diego, CA: Educational and Industrial Test-
ing Service.
Mazer, J. D., & Cloninger, R. E. (1990). Comorbidity of mood and anxiety disorders. Washington, DC: American
Psychiatric Press.
Moos, R. H. (1988). Coping Responses Inventory Manual. Odessa, FL: Psychological Assessment Resources, Inc.
Moos, R. H., & Schaefer, J. A. (1993). Coping resources and processes: current concepts and measures. In L. Gold-
berger, & S. Breznitz (Eds.), Handbook of stress: Theoretical and clinical aspects (2nd ed., pp. 234–257). New
York: Free Press.
Nezu, A. M., Nezu, C. M., & Perri, M. G. (1989). Problem-solving therapy for depression: Theory, research and
clinical guidelines. New York: John Wiley & Sons.
Pearlin, L. I., & Schooler, C. (1978). The structure of coping. Journal of Health and Social Behavior, 19, 2–21.
Rassin, E., Merckelbach, H., & Muris, P. (2000). Paradoxical and less paradoxical effects of thought suppression: a
critical review. Clinical Psychology Review, 20, 973–995.
Richter, J., Eisemann, M., & Richter, G. (2000). Temperament and character during the course of unipolar depression
among inpatients. European Archives of Psychiatry & Clinical Neuroscience, 250, 40–47.
Spielberger, C. D., Gursuch, R. L., & Lushene, R. E. (1970). Manual for the State–Trait Anxiety Inventory (Self
Evaluation Questionnaire). Palo Alto, CA: Consulting Psychologists Press.
Spurrell, M. T., & McFarlane, A. C. (1995). Life-events and psychiatric symptoms in a general psychiatry clinic: the
role of intrusion and avoidance. British Journal of Medical Psychology, 68, 333–340.
Stöber, J., & Borkovec, T. D. (2002). Reduced concreteness of worry in generalized anxiety disorder: findings from a
therapy study. Cognitive Therapy and Research, 26, 89–96.
 N.D. Ottenbreit, K.S. Dobson / Behaviour Research and Therapy 42 (2004) 293–313 313

Stöber, J., Tepperwien, S., & Staak, M. (2000). Worry leads to reduced concreteness of problem elaborations: evidence
for the avoidance theory of worry. Anxiety, Stress and Coping, 13, 217–227.