Emotion
Expectancy Bias in Anxious Samples
Cindy M. Cabeleira, Shari A. Steinman, Melissa M. Burgess, Romola S. Bucks, Colin MacLeod,
Wilson Melo, and Bethany A. Teachman
Online First Publication, May 5, 2014. http://dx.doi.org/10.1037/a0035899
CITATION
Cabeleira, C. M., Steinman, S. A., Burgess, M. M., Bucks, R. S., MacLeod, C., Melo, W., &
Teachman, B. A. (2014, May 5). Expectancy Bias in Anxious Samples. Emotion. Advance
online publication. http://dx.doi.org/10.1037/a0035899
Emotion
2014, Vol. 14, No. 1, 000
© 2014 American Psychological Association
1528-3542/14/$12.00 DOI: 10.1037/a0035899
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Expectancy Bias in Anxious Samples
Cindy M. Cabeleira
Shari A. Steinman
University of Western Australia
University of Virginia
Melissa M. Burgess and Romola S. Bucks
Colin MacLeod
University of Western Australia
University of Western Australia and Australian Bushfire
Cooperative Research Centre, Melbourne, Victoria, Australia
Wilson Melo
Bethany A. Teachman
Federal University of Rio Grande do Sul
University of Western Australia and University of Virginia
Although it is well documented that anxious individuals have negative expectations about the future, it
is unclear what cognitive processes give rise to this expectancy bias. Two studies are reported that use
the Expectancy Task, which is designed to assess expectancy bias and illuminate its basis. This task
presents individuals with valenced scenarios (Positive Valence, Negative Valence, or Conflicting
Valence), and then evaluates their tendency to expect subsequent future positive relative to negative
events. The Expectancy Task was used with low and high trait anxious (Study 1: n ⫽ 32) and anxiety
sensitive (Study 2: n ⫽ 138) individuals. Results suggest that in the context of physical concerns, both
high anxious samples display a less positive expectancy bias. In the context of social concerns, high trait
anxious individuals display a negative expectancy bias only when negatively valenced information was
previously presented. Overall, this suggests that anxious individuals display a less positive expectancy
bias, and that the processes that give rise to this bias may vary by type of situation (e.g., social or
physical) or anxiety difficulty.
Keywords: anxiety sensitivity, expectancy bias, extrapolation, trait anxiety
It is well documented that the content of maladaptive cognitions in
anxiety tends to be concerned with the prospect of harmful future
events (Beck & Clark, 1988; Kendall & Ingram, 1989). Those who
are clinically anxious are more likely to have negatively distorted
expectations of the future than are nonanxious individuals (e.g., MacLeod, Tata, Kentish, & Jacobsen, 1997; Miranda & Mennin, 2007).
Consistent with this focus, an anxiety-linked negative expectancy bias
reflects an inflated tendency for anxious individuals to expect an
increased probability of negative relative to positive events. This
anticipation of a wide range of negative events has been demonstrated
not only in individuals who are clinically anxious (e.g., Borkovec,
Alcaine, & Behar, 2004; Dugas et al., 1998), but also in nonclinical
individuals who are highly trait anxious (e.g., MacLeod & Byrne,
1996; Stöber, 2000). Although an anxiety-linked negative expectancy
bias has been documented, current paradigms do not illuminate the
conditions that give rise to this bias.
Cindy M. Cabeleira, Elizabeth Rutherford Memorial Centre for the Advancement of Research on Emotion, School of Psychology, University of
Western Australia; Shari A. Steinman, Department of Psychology, University
of Virginia; Melissa M. Burgess and Romola S. Bucks, Elizabeth Rutherford
Memorial Centre for the Advancement of Research on Emotion, School of
Psychology, University of Western Australia; Colin MacLeod, Elizabeth Rutherford Memorial Centre for the Advancement of Research on Emotion, School
of Psychology, University of Western Australia and Australian Bushfire Cooperative Research Centre, Melbourne, Victoria, Australia; Wilson Melo,
Institute of Psychology, Federal University of Rio Grande do Sul, Porto
Alegre; Bethany A. Teachman, Elizabeth Rutherford Memorial Centre for the
Advancement of Research on Emotion, School of Psychology, University of
Western Australia and Department of Psychology, University of Virginia.
Cabeleira and Steinman contributed equally to this work.
We thank the members of the Program for Anxiety Cognition and
Treatment (PACT) Lab for their insightful comments and suggestions.
We also thank Rae Davidson, Shana Hovitz, Fiona Ritchey, and Lenny
Chan for research assistance. This research was supported in part by
NIA R01AG033033 Grant to Bethany Teachman and a Programmatic
Distinguished Visitor award from the University of Western Australia
awarded to Bethany Teachman. This research was also supported
in part by Australian Research Council Grant DP0879589, and by a
grant from the Romanian National Authority for Scientific Research,
CNCS – UEFISCDI, project number PNII-ID-PCCE-2011–2-0045,
both awarded to Colin MacLeod. It is acknowledged that this
research was conducted while Cindy Cabeleira was an International
Postgraduate Research Scholarship recipient at the University of Western Australia.
Correspondence concerning this article should be addressed to
Cindy M. Cabeleira, University of Western Australia, School of
Psychology, Faculty of Life and Physical Sciences, M304, 35 Stirling
Hwy, Crawley, WA 6009, Australia. E-mail: cabelc01@student.uwa
.edu.au
1
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2
CABELEIRA ET AL.
In this article, two studies are reported that use a paradigm that
provides individuals with valenced information, and then evaluates
their tendency to expect positive or negative events to occur in the
future. This paradigm is used with both highly trait anxious (HTA)
and highly anxiety sensitive (HAS) samples. Trait anxiety reflects
the propensity to become anxious across many different contexts
(Spielberger, 1983), whereas anxiety sensitivity reflects the fear of
symptoms related to anxiety (e.g., bodily sensations, such as a
racing heart) and the belief that these sensations have negative
physical, social, or psychological consequences (Reiss, 1991; Reiss & McNally, 1985).
Although the anxiety-linked tendency to anticipate negative
future events has been demonstrated using a variety of paradigms,
no methodology has revealed the conditions that give rise to this
expectancy bias in anxious samples. Specifically, it is unclear
whether this bias occurs regardless of the valence of previously
presented information, or whether it is moderated by the valence of
previously presented information. In this article, we consider three
potential hypotheses regarding conditions that may lead to an
expectancy bias.
First, it may be that the expectancy bias seen among anxious
individuals is pervasive, such that it is shown across situations.
Specifically, anxious individuals may have a more negative expectancy bias (relative to nonanxious individuals), regardless of
whether current or recent events are emotionally negative, are
emotionally positive, or are conflicting in emotional valence (i.e.,
containing negative and positive elements). This possibility is
referred to as the Pervasive Expectancy Bias Hypothesis
(Pervasive-EBH). Pervasive in this context refers to the bias occurring irrespective of differently valenced prior information or
preceding events; it does not imply that expectancy biases will
persist regardless of context. If an anxiety-linked tendency to show
a more negative expectancy bias (relative to nonanxious individuals) is found, but it is moderated by the valence of previously
presented information, this would refute the validity of the
Pervasive-EBH. The Expectancy Task allows us to test two such
hypotheses.
The Extrapolation Expectancy Bias Hypothesis (ExtrapolationEBH) suggests that the anxiety-linked elevation in expectations of
negative future events is moderated by the (negative or positive)
valence of previously presented information. Thus, anxious individuals may expect relatively more negative future events because
they exhibit biased extrapolation from current events, relative to
nonanxious individuals. For example, anxious individuals may be
disproportionately inclined to infer that negative current events
will lead to negative future events. Should this be observed, then
the degree to which anxious (compared to nonanxious) participants
inflate the probability of future negative events will be disproportionately greater when the information they are given indicates that
these prior events proceeded in a negative manner.
A third hypothesis, the Emotional Weighting Expectancy Bias
Hypothesis (Emotional Weighting-EBH), refers to the possibility
that the anxiety-linked expectancy bias is moderated by whether or
not previously presented information is unresolved or mixed with
respect to valence. It is hypothesized that, following the presentation of both positive and negative previous information, anxious
individuals may be more likely than nonanxious individuals to
assign more weight to the negative (rather than positive) information, construing the overall event as relatively more negative.
Consequently, following the presentation of both positive and
negative previous information, anxious individuals may be more
likely to expect negative future events to occur. The current
research presents participants with conflicting (negative and positive) information to test this hypothesis. In summary, the three
hypotheses presented here differ in terms of whether an anxietylinked expectancy bias is moderated by the valence of previously
presented information.
Evaluating these hypotheses requires a task that provides information about the manner in which a range of scenarios proceed,
and then assesses participants’ expectancies for alternative possible future events that differ in their emotional valence. By manipulating the valence of information initially presented in each
scenario, it is possible to investigate the circumstances under
which biased expectation for future positive versus negative events
will characterize anxious participants.
In the current article, a paradigm that meets these requirements
is used with two anxious samples: the Expectancy Task (Cabeleira,
Bucks, Teachman, & MacLeod, 2010). Originally introduced and
developed by Cabeleira et al. (2010) and further validated by
Steinman, Smyth, Bucks, MacLeod, and Teachman (2013), the
Expectancy Task presents participants with information about a
range of hypothetical scenarios. The scenarios relate to physical or
social events, which may be processed differently by people with
different types and levels of anxiety. Most importantly, the scenarios vary in valence, and can be negative (including only negative and neutral events), positive (including only positive and
neutral events), or conflicting in valence (including an equal number of positive and negative events). After reading and imagining
themselves in the scenarios, participants are required to rate the
likelihood of three future events occurring next, which can be
negative, positive, or neutral in valence, on a scale of 1 (very
unlikely to happen next) to 4 (very likely to happen next). These
ratings reveal anxiety-related bias in the relative tendency to
expect positive versus negative future events. By examining
whether such expectancy bias is influenced by the valence of the
information provided in the initial scenarios, the three anxietylinked expectancy bias hypotheses described above can be tested.
In summary, the current studies have two key aims: a) to
determine whether anxious individuals (HTA in Study 1, HAS in
Study 2) show an inflated tendency to anticipate relatively more
negative future events relative to nonanxious individuals, which
we term an anxiety-linked negative expectancy bias; and b) to test
the three hypotheses described above by evaluating whether such
a bias is moderated by the valence of previous events.
Study 1
Method
Participants. Thirty-two first year psychology undergraduates from the University of Western Australia were recruited based
on their score on the Trait form of the State–Trait Anxiety Inventory (STAI-T; Spielberger, 1983). The mean trait anxiety score for
college students (Spielberger) was used to determine cut-off scores
for inclusion in HTA and LTA groups. To be included in the HTA
group, participants had to score at least one standard deviation
(SD ⫽ 9.67) above this mean trait anxiety score (M ⫽ 39.35) for
college students, thus scoring 50 or above on the STAI-T. To be
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
EXPECTANCY BIAS AND ANXIETY
included in the LTA group, participants had to score at least one
standard deviation below this mean trait anxiety score for college
students, thus scoring 29 or below on the STAI-T. There were 16
LTA participants (50% female), and 16 HTA participants (50%
female). The mean age of the sample was 17.6 years (SD ⫽ 1.16;
range 17–22 years), and race was reported as follows: 68.8%
White, 21.9% Asian, 6.3% Black/African, and 3.1% “other.” The
University of Western Australia’s Human Research Ethics Committee (HREC) approved this study.
Materials.1
Anxiety symptoms. The 20-item STAI (Spielberger, 1983)
includes one scale to assess state anxiety (STAI-S) and one scale
to assess dispositional trait anxiety (STAI-T). The reliability
(Barnes, Harp, & Jung, 2002) and validity (Spielberger) of the
scales are well established, and Cronbach’s alpha was .96 for the
STAI-T and .91 for the STAI-S in the current study.
Expectancy task. The Expectancy Task (Cabeleira et al.,
2010) is a computerized reading judgment task designed to evaluate an individual’s tendency to anticipate positive or negative
events to occur (labeled “expectancy bias”). The Expectancy Task
involves presenting 64 scenarios (in 16 blocks of four) that vary in
the extent to which positive, neutral, or negative events occur, then
asking participants to judge the likelihood of future valenced
events occurring in each of these scenarios. The Expectancy Task
includes a Scenario Presentation Component and an Expectancy
Rating Component.
In the Scenario Presentation Component, participants were
asked to read and imagine themselves in a number of scenarios,
each described by six statements: a Title, an Orienting Sentence,
and four events (see Appendix for examples). The Title remained
in the center of the computer screen for the duration of the scenario
presentation, while the other five statements appeared directly
below the Title, and each remained on the screen only until the
participant pressed the spacebar, signaling that he or she had read
the statement and was ready for the next statement to be shown.
The four events in a scenario could be shown in any of three
Passage Valence Conditions. In the Positive Valence condition,
two positive and two neutral events were presented. In the Negative Valence condition, two negative and two neutral events were
presented. In the Conflicting Valence condition, two negative and
two positive events were presented. The neutral events were included to control for amount of information presented in each
combination (i.e., such that each scenario consistently included
four events, and each valence was consistently represented by two
events). Order of valenced events (e.g., positive vs. neutral) within
a scenario was counterbalanced. A graphical depiction of the
Scenario Presentation Component of the Expectancy Task is presented in Figure 1, which provides an illustrative example using a
scenario relating to physical concerns delivered in the Conflicting
Valence condition.
In the Expectancy Rating Component of the task, participants
were asked to think about the likelihood of different specified
candidate future events for each of the scenarios they had previously read and imagined themselves in. On each trial, participants
read the Title and Orienting Sentence from one of the previously
seen scenarios (and the four events previously presented as part of
that scenario were represented as lines of stars below the orienting
sentence), which remained on screen while participants were asked
to rate their beliefs concerning the likelihood that each of the three
3
Going to the Doctor
Going to the Doctor
You go to the doctor’s rooms
Going to the Doctor
************************
You find out you need a biopsy done
Figure 1. A graphical depiction of the Scenario Presentation Component
of the Expectancy Task which provides an illustrative example using a
scenario relating to physical concerns delivered in the Conflicting Valence
condition. (Minor visual modifications were made in Study 2.)
specific events would happen to them within the scenario they
imagined themselves experiencing. These candidate future events
included one positive event, one negative event, and one neutral
event, presented in a random order. These three candidate future
events were displayed in the middle of the screen and participants
were instructed to use a scale ranging from 1 (very unlikely to
happen next) to 4 (very likely to happen next) to rate the subjective
likelihood of each event. A graphical depiction of the Expectancy
Rating Component of the Expectancy Task is presented in Figure
2, which provides an illustrative example using a scenario relating
to physical concerns delivered in the Conflicting Valence condition. Scenarios were presented in blocks of four so that the load on
1
Additional self-report measures were used in this study and are reported elsewhere. For a complete list of measures, please contact Cindy M.
Cabeleira.
CABELEIRA ET AL.
4
Going to the Doctor
************************
***********************************
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
The doctor prescribes you medication that can have bad side effects
(see Cabeleira, 2010). Specifically, all Scenario Event Sets used in
the current study were judged to be relevant to either physical or
social concerns, and valenced ratings for events were consistent
with the intended valence of each event (i.e., positive, negative, or
neutral). Additionally, the positive and negative events were rated
to have equivalent valence intensity.
Procedure. Participants were informed that the experiment
was designed to illuminate how people differ in terms of their
understandings of hypothetical scenarios. Participants began the
session by completing the STAI-S. Participants then completed
eight practice scenarios of the Expectancy Task, followed by the
full Expectancy Task. At the end of the session, participants
received course credit for their participation and were debriefed.
Going to the Doctor
************************
***********************************
******************************************************
The doctor informs you that you are at a healthy weight
Going to the Doctor
************************
***********************************
******************************************************
*********************************************
The doctor says she is happy with your exercise regime
Results
Descriptive statistics. Chi-square tests revealed that the LTA
and HTA groups did not differ by gender (2 ⫽ .00, p ⫽ 1.000) or
race (2 ⫽ 4.06, p ⫽ .541), and an independent samples t test
demonstrated there was no significant difference between the LTA
and HTA groups in terms of age, t(30) ⫽ .91, p ⫽ .37, d ⫽ 0.32.
An independent samples t test was used to compare both trait and
state anxiety scores at baseline between groups (LTA vs. HTA).
As expected, this revealed a significant group difference in trait
anxiety, t(30) ⫽ 27.79, p ⬍ .001, d ⫽ 9.82, such that HTA
participants reported higher trait anxiety than LTA participants.
Unsurprisingly, a significant group difference was also observed
for state anxiety, t(30) ⫽ 5.16, p ⬍ .001, d ⫽ 1.82, such that HTA
participants reported higher state anxiety than LTA participants.
Descriptive statistics for age, trait, and state anxiety scores for each
anxiety group are presented in Table 1.
Going to the Doctor
Figure 1 (continued).
You go to the doctor’s rooms
***********************************
******************************************************
memory would be minimal, with each block being followed by the
expectancy ratings for the future events for the four scenarios. In
total, 16 blocks of four scenarios were presented in this manner,
totaling 64 scenarios.
Scenario event sets. Each of the 64 scenarios presented in the
study was derived from its own Scenario Event Set. Each Scenario
Event Set represented a hypothetical scenario related to either a
physical or social concern, and included 11 items: a Title, an
Orienting Sentence, and nine candidate events. Of the nine candidate events, three were positive, three were negative, and three
were neutral (see Appendix). The four events actually presented in
the Expectancy Task for any scenario were selected from its
Scenario Event Set, in a manner that took account of the Passage
Valence Condition for that scenario. Two of the three events of
each valence to be presented in the scenario were randomly selected for display in the Scenario Presentation Component of the
task, whereas the third event of each valence was shown in the
Expectancy Rating Component of the task. All Scenario Event Sets
were previously validated by an independent sample of 16 raters
*********************************************
*********************************************
RATING?
The doctor warns you all your family is at risk of diabetes
The doctor says your heart sounds very healthy
The telephone rings
1 – Very UNLIKELY to happen next
2 – Somewhat UNLIKELY to happen next
3 – Somewhat LIKELY to happen next
4 – Very LIKELY to happen next
Figure 2. A graphical depiction of the Expectancy Rating Component of
the Expectancy Task which provides an illustrative example using a
scenario relating to physical concerns delivered in the Conflicting Valence
condition. (Note that the word “RATING” moves to subsequent statements
once a rating has been entered. Minor visual modifications were made in
Study 2.)
EXPECTANCY BIAS AND ANXIETY
Table 1
Study 1: Descriptive Statistics for Low and High Trait
Anxious Groups
Measures
Age
STAI-T
STAI-S
Low Trait Anxious (LTA)
n ⫽ 16
M ⫾ SD
High Trait Anxious (HTA)
n ⫽ 16
M ⫾ SD
17.75 ⫾ 1.39
25.88 ⫾ 2.63
27.19 ⫾ 5.86
17.38 ⫾ 0.89
54.81 ⫾ 3.23
40.06 ⫾ 8.08
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Note. STAI-T ⫽ State-Trait Anxiety Inventory – Trait Score; STAI-S ⫽
State-Trait Anxiety Inventory – State Score.
Evidence for expectancy effects.2 The Expectancy Task was
scored by subtracting participants’ average likelihood ratings for
negative future events from their average likelihood ratings for
positive future events to yield an expectancy bias index. The
decision to create this relative, rather than absolute, bias index
derived from a desire to simultaneously consider the valuing of
positive and negative choices, given the external validity of needing to weigh multiple options simultaneously when predicting the
future. A score of zero on this index would indicate that there was
no difference between participants’ ratings for the likelihood of
negative and positive future events. A score on this index that is
greater for Participant A than for Participant B would indicate that
the former participant exhibited a relatively greater tendency to
expect positive events, while the latter participant demonstrated a
relatively greater tendency to expect negative events. A summary
of the mean probability ratings for negative and positive future
event statements is presented in Table 2, with mean ratings organized by Passage Valence condition (Negative Valence, Positive
Valence, Conflicting Valence), Scenario domain (Physical, Social), and Anxiety group (LTA, HTA).
A repeated measures analysis of variance (ANOVA) with one
between-subjects factor of Anxiety Group (LTA, HTA), and two
within-subjects factors of Passage Valence Condition (Positive
Valence, Negative Valence, Conflicting Valence) and Scenario
Domain (Physical, Social), was conducted to examine whether an
anxiety-linked difference in negative expectancy bias was evident,
and, if so, under which experimental conditions it was observed.
The results of the ANOVA revealed a significant main effect of
Anxiety Group, F(1, 30) ⫽ 22.71, p ⬍ .001, p2 ⫽ .43, reflecting
the anticipated lower positive expectancy bias index scores for the
HTA group (M ⫽ 0.86, SD ⫽ 0.44) compared with the LTA group
(M ⫽ 1.01, SD ⫽ 0.64).
There was also a main effect of Passage Valence Condition, F(2,
29) ⫽ 22.74, p ⬍ .001, p2 ⫽ .61. Follow-up analyses showed that
ratings of future events differed significantly across the three
passage valence types in the expected direction (all p ⬍ .001), such
that Negative Valence passages (M ⫽ ⫺0.01, SD ⫽ 1.14) yielded
a relatively less positive expectancy index compared to the Conflicting Valence passages (M ⫽ 0.50, SD ⫽ 0.73), which in turn
yielded a less positive expectancy index relative to the Positive
Valence passages (M ⫽ 1.16, SD ⫽ 0.68). This suggests that the
Expectancy Task is sufficiently sensitive to show expectancies are
influenced by the valence of prior information.
Importantly, there was a significant 2-way interaction between
Anxiety Group and Passage Valence Condition, F(2, 29) ⫽ 6.46
5
p ⫽ .005, p2 ⫽ .31. This was further embedded in a significant
3-way interaction of Anxiety Group, Passage Valence Condition,
and Scenario Domain, F(2, 29) ⫽ 3.92 p ⫽ .031, p2 ⫽ .21; see
Figure 3. No other significant effects emerged from this analysis.
To understand the 3-way interaction, between group differences in
the expectancy rating data were analyzed separately for each
Scenario Domain.
For scenarios related to physical concerns, there was a main
effect of Anxiety Group, F(1, 17.19) ⫽ 19.94, p ⬍ .001, p2, ⫽ .40,
such that HTA individuals had a less positive expectancy bias
relative to LTA individuals (HTA M ⫽ 0.13, SD ⫽ 0.42; LTA
M ⫽ 0.98, SD ⫽ 0.63). No significant 2-way interaction of
Passage Valence Condition and Anxiety Group, F(2, 29) ⫽ 2.14
p ⫽ .137, p2 ⫽ .13, was observed. In other words, both HTA and
LTA individuals’ expectancies were similarly affected by the
valence of initial scenarios, indicating that extrapolation from
valence of initial events did not vary by anxiety group, though the
HTA group expected relatively less positive future events in general.
For scenarios related to social concerns, there was a main effect
of Anxiety Group, F(1, 24.13) ⫽ 20.88, p ⬍ .001, p2 ⫽ .41, that
was subsumed by a significant 2-way interaction of Passage Valence Condition and Anxiety Group, F(2, 29) ⫽ 11.73, p ⬍ .001,
p2 ⫽ .45. Independent samples t tests were conducted to compare
the Anxiety Groups’ expectancy index ratings for each Passage
Valence type for the social scenarios. Results revealed that HTA
participants showed less positive expectancy bias than LTA participants when scenarios were initially presented in the Negative
Valence condition, t(30) ⫽ 5.96, p ⬍ .001, d ⫽ 2.11; HTA
M ⫽ ⫺0.98, SD ⫽ 0.67, LTA M ⫽ 0.84, SD ⫽ 1.02, or in the
Conflicting Valence condition, t(30) ⫽ 4.38, p ⬍ .001, d ⫽ 1.55;
HTA M ⫽ 0.00, SD ⫽ 0.57, LTA M ⫽ 0.97, SD ⫽ 0.68. There was
no significant Anxiety Group difference in expectancy bias scores
when scenarios were initially presented in the Positive Valence
condition, t(30) ⫽ 0.79, p ⫽ .435; HTA M ⫽ 1.09, SD ⫽ 0.85,
LTA M ⫽ 1.32, SD ⫽ 0.77, with both groups similarly rating
positive future events as more likely to occur than negative future
events. Thus, whenever the initial social scenario contained negative events (i.e., in both the Negative Valence and Conflicting
Valence conditions), the HTA participants showed lower expectancy bias for future positive events, compared to the LTA participants. However, this was not the case when the initial scenario did
not contain negative events (i.e., in the Positive Valence condition).
To further understand the significant 3-way interaction of Anxiety Group, Passage Valence Condition, and Scenario Domain,
within anxiety group differences were examined next. For the LTA
group, there was the expected main effect of Passage Valence
Condition, F(1.43, 21.44) ⫽ 4.94, p ⫽ .026, p2 ⫽ .25. Least
Significant Difference (LSD) comparisons revealed relatively
greater positive expectancy when the preceding information was
Positive (M ⫽ 1.34, SD ⫽ .65) relative to that containing negative
information (i.e., Negative M ⫽ 0.77, SD ⫽ .99; p ⫽ .031, and
2
An ANOVA revealed that the order of events (e.g., negative-neutral vs.
neutral-negative) presented in the Scenario Presentation Component of the
task did not influence the ratings made in the Expectancy Rating Component of the task (all p ⬎ .05). Thus, order of events was not included as a
factor in the analyses presented.
CABELEIRA ET AL.
6
Table 2
Study 1: Summary of Mean Probability Ratings for Negative and Positive Future Event
Statements, With Mean Ratings Organized by Passage Valence Condition (Negative Valence,
Positive Valence, Conflicting Valence), Scenario Domain (Physical, Social), and Anxiety Group
(LTA, HTA)
Low Trait Anxious (LTA) n ⫽ 16
Passage valence
Negative valence
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Positive valence
Conflicting valence
High Trait Anxious (HTA) n ⫽ 16
Future
valence
Social scenarios
M ⫾ SD
Physical scenarios
M ⫾ SD
Social scenarios
M ⫾ SD
Physical scenarios
M ⫾ SD
Negative
Positive
Negative
Positive
Negative
Positive
2.1 ⫾ 0.6
2.9 ⫾ 0.5
1.8 ⫾ 0.5
3.1 ⫾ 0.4
2.0 ⫾ 0.5
3.0 ⫾ 0.4
2.2 ⫾ 0.8
2.9 ⫾ 0.5
1.9 ⫾ 0.5
3.2 ⫾ 0.5
2.1 ⫾ 0.4
2.9 ⫾ 0.4
2.9 ⫾ 0.4
2.0 ⫾ 0.4
2.0 ⫾ 0.6
3.1 ⫾ 0.4
2.6 ⫾ 0.4
2.6 ⫾ 0.3
2.8 ⫾ 0.5
2.2 ⫾ 0.4
2.2 ⫾ 0.4
3.0 ⫾ 0.4
2.6 ⫾ 0.5
2.8 ⫾ 0.4
Conflicting Valence passages M ⫽ 0.92, SD ⫽ .62; p ⫽ .012,
respectively). There was no significant difference between the
observed positive expectancy when preceding information was of
a Negative Valence compared to a Conflicting Valence (p ⫽ .370).
There was no significant 2-way Passage Valence Condition by
Scenario domain interaction, F(2, 30) ⫽ 0.53, p ⫽ .592, p2 ⫽ .03.
Notably, the absolute value of the expectancy bias was positive
across all three Passage Valence Conditions (i.e., t tests indicated
the bias value was significantly greater than zero; all p ⬍ .05).
For the HTA group, there was again a main effect of Passage
Valence Condition, F(2, 30) ⫽ 39.67, p ⫽ .000, p2 ⫽ .73, which
was subsumed by a significant 2-way Passage Valence Condition
by Scenario domain interaction, F(2, 30) ⫽ 5.95, p ⫽ .007, p2 ⫽
.28. For social scenarios, LSD comparisons revealed relatively
greater positive expectancy when the preceding information was of
a Positive Valence (M ⫽ 1.09, SD ⫽ .85) compared with Conflicting Valence (M ⫽ 0.00. SD ⫽ .57; p ⬍ .001), which in turn
resulted in greater positive expectancy than when the preceding
information was of a Negative Valence (M ⫽ ⫺0.98, SD ⫽ .67;
p ⬍ .001). Further, the HTA group displayed a relatively negative
expectancy bias (i.e., less than zero) in social scenarios when the
preceding information was of a Negative Valence, t(15) ⫽ 5.83,
p ⬍ .001, d ⫽ 1.46, no expectancy bias (i.e., bias is not significantly different from zero, t(15) ⫽ 0.00, p ⫽ 1.000) when preceding information was of a Conflicting Valence (i.e., contains negative and positive information), and a positive expectancy bias
when preceding information was of a Positive Valence, t(15) ⫽
5.15, p ⬍ .001, d ⫽ 1.28.
For physical scenarios, LSD comparisons revealed a more positive expectancy bias when the preceding information was of a
Positive Valence (M ⫽ 0.85, SD ⫽ .65), rather than being of a
Conflicting Valence (M ⫽ 0.14, SD ⫽ .67; p ⫽ .004), which in
turn resulted in greater positive expectancy than when the preceding information was of a Negative Valence (M ⫽ ⫺0.59, SD ⫽
.70; p ⫽ .002). Further, participants displayed a negative expectancy bias in physical scenarios when the preceding information
was of a Negative Valence, t(15) ⫽ 3.37, p ⫽ .004, d ⫽ 0.84, and
no expectancy bias was observed when preceding information was
of a Conflicting Valence (i.e., contains negative and positive
information; t(15) ⫽ 0.84, p ⫽ .416), and a positive expectancy
bias when preceding information was of a Positive Valence,
t(15) ⫽ 5.27, p ⬍ .001, d ⫽ 1.31.
Taken together, these results confirm the presence of an anxietylinked expectancy bias. Interestingly, when examining the
between-groups effects, the cognitive mechanisms underpinning
this anxiety-linked negative expectancy bias appear to differ for
scenarios relating to physical concerns versus social concerns.
With respect to physical scenarios, there was no evidence that
valence of the prior information provided in the scenarios affected
the magnitude of this anxiety-linked effect, thus providing no basis
for refuting the Pervasive-EBH. With respect to social concerns,
the HTA participants displayed a more negative/less positive expectancy bias only when negative events had already initially
occurred within these scenarios. These results for the HTA participants appear attributable to greater (negative) extrapolation from
Figure 3. Study 1: Expectancy bias by scenario type, valence condition,
and level of trait anxiety. Expectancy bias was scored by subtracting
average ratings of negative future events from average ratings of positive
future events, so higher scores indicate a more positive bias.
EXPECTANCY BIAS AND ANXIETY
previous negative events, consistent with the Extrapolation-EBH.
When examining the within-group effects, the LTA were pervasively more positive in their expectations of future events, while
the expectancies of the HTA group were more consistent with the
valence of previously presented information.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Discussion
This study had two aims: a) to determine whether HTA individuals do indeed show an inflated tendency to expect negative
future events relative to LTA individuals, and b) to investigate
whether such an anxiety-linked bias is moderated by the emotional
tone of previously presented information. The findings confirmed
that, compared with LTA participants, HTA participants showed a
more negative/less positive expectancy bias, consistent with previous research (e.g., Miranda & Mennin, 2007). Furthermore, the
present study also sheds light on the nature of this bias, revealing
that it differs between high and low anxiety groups depending on
the type of concerns (physical or social) depicted in events.
In scenarios relating to physical concerns, the observed anxietylinked expectancy bias was not moderated by the emotional tone of
previous events. Of course the absence of an interaction here
cannot be interpreted as absolute evidence for the Pervasive-EBH,
given that a conclusion cannot be sustained on the basis of confirming predicted null results. Perhaps with greater power (e.g., a
larger sample size), valence may have moderated the anxietylinked expectancy bias in physical scenarios, and there are of
course numerous other factors not investigated here that could
have moderated the anxiety-linked effect. Notwithstanding, with
respect to the factors investigated here, the Pervasive-EBH appears
to most parsimoniously accommodate the results.
When participants were required to judge the likelihood of
future events in scenarios related to social concerns, the anxious
participants only demonstrated a more negative/less positive expectancy bias than nonanxious participants when negative events
had already occurred within the initial description of these scenarios (i.e., in the Negative Valence and Conflicting Valence conditions). This pattern of results is consistent with greater negative
extrapolation, whereby anxious individuals showed an elevated
tendency to infer that future events will continue to be negative
when previous events have been negative (matching the
Extrapolation-EBH).
These different findings for the physical versus social scenarios
were unexpected. As mentioned earlier, sample sizes were small in
this study, thus limiting power to detect significant effects. It is
possible that a pattern of results consistent with the ExtrapolationEBH may have occurred for the physical scenarios if the study was
conducted with a larger sample. Another challenge in interpreting
this difference is that social concerns may be more personally
relevant for this sample of young, HTA individuals than physical
concerns, given the automatic “attention-grabbing power of negative social information” (Pratto & John, 1991, p. 380). Thus,
perhaps these individuals were more likely to make greater use of
previously presented, valenced information in forming their future
expectations about socially related matters.
In terms of the within anxiety group effects observed, the LTA
appeared to have a pervasively positive expectancy bias regardless
of scenario type (i.e., physical or social) or prior valence (though
degree of positivity clearly differed by prior valence), whereas the
7
expectancies of the HTA were more consistent with the valence of
preceding information (i.e., positive bias in Positive Valence passages, negative bias in Negative Valence passages, and no significant bias in Conflicting Valence passages) for scenarios relating
to both social and physical concerns.
To address issues of power and relevance of scenario concerns,
and replicate and extend the findings of Study 1, Study 2 uses a
much larger, alternate anxious sample for which physical concerns
tend to be more personally relevant.
Study 2
Although the presence of an expectancy bias in trait anxious
samples has been previously explored, it is unclear whether this
phenomenon is consistent across different types of anxious samples, or whether anxiety subtypes differ in their expectancies and
the conditions under which negative expectancy biases arise. Anxiety sensitivity involves the fear of anxiety-related symptoms,
including various bodily sensations (Cox, Parker, & Swinson,
1996) and the belief that these sensations have negative physical,
social, or psychological consequences (Reiss, 1991; Reiss & McNally, 1985). Study 2 aims to explore whether an anxiety-linked
expectancy bias in scenarios relating to physical concerns will
once again not be moderated by the valence of previous events and
linked to a Pervasive Expectancy Bias (in line with the Study 1
findings), or whether an Extrapolation Expectancy Bias will
emerge (akin to findings for the social scenarios in Study 1). Given
high anxiety sensitive (HAS) individuals are known to have concerns about the physical consequences of anxiety, the physical
scenarios may be more relevant than they were for the HTA group
in Study 1. No research to the authors’ knowledge has investigated
expectancy bias in HAS individuals in a manner that can test
whether results are consistent with a Pervasive-EBH,
Extrapolation-EBH, or Emotional Weighting-EBH account.
In Study 2, the Expectancy Task was used with low anxiety
sensitive (LAS) and HAS samples. Only scenarios related to
physical events (e.g., going to the doctor, or exercising) were
included in this version of the Expectancy task given our interest
in better understanding the nature of expectancies for this material
in a sample known to have concerns about physical sensations
(Clark, 1986). Note that the scenarios included did not all perfectly
align with fears of bodily sensations. Rather, the scenarios included a broad range of physical concerns. Given the relationship
between health anxiety and anxiety sensitivity (e.g., Wheaton,
Berman, & Abramowitz, 2010), it is probable that many of the
scenarios were personally relevant for participants with HAS, and
we wanted to sample the physical domain broadly. Of note, we
used a much larger sample in Study 2, which addresses power
concerns in Study 1.
Additionally, in Study 2, we consider another potential moderator of expectancies—the role of priming concerns tied to the
physical scenarios to make those concerns salient before forming
expectancies. The inclusion of this moderator follows mixed results in the field about the role of such primes in the expression of
cognitive biases in anxious samples. For instance, priming concerns related to specific fears have led to enhancement of recall
biases in a spider-fearful sample (e.g., Smith-Janik & Teachman,
2008), but also reduction of attention biases in a snake-fearful
sample (e.g., Mathews & Sebastian, 1993). On the contrary, prim-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
8
CABELEIRA ET AL.
ing concerns related to specific fears had no effect on memory
biases in another spider phobic sample (see Study 1 in Watts &
Dalgeish, 1991). Thus, we are interested in how an analogous
prime tied to physical concerns will alter the expression of expectancy bias. It is hypothesized that priming physical concerns will
increase state anxiety for the HAS group, but not the LAS group.
In response to this prime, it is tentatively hypothesized that the
difference in expectancy bias between LAS and HAS individuals
will be magnified following priming of physical concerns because
the prime will make those concerns salient, but given prior mixed
results, this evaluation is somewhat exploratory.
Finally, to test whether expectancy bias is related to markers of
anxiety beyond only questionnaire measures of anxious symptoms,
a measure of anxiety experienced during an anxiety sensitivityrelevant stressor was included. It is predicted that expectancy bias
will be related to anxiety experienced during an anxiety-relevant
trigger, speaking to the predictive validity of expectancy bias.
Method
Participants. Participants were recruited through the University of Virginia’s psychology department participant pool, based
on their responses to the Anxiety Sensitivity Index (ASI; Reiss,
Peterson, Gursky, & McNally, 1986). Students who scored 14 or
below on the total ASI (so they were at least .5 standard deviations
below ASI college student norms; Peterson & Reiss, 1992) were
invited to participate in the LAS group. Students who scored 23 or
greater on the total ASI (so they were at least .5 standard deviations above ASI college students norms; Peterson & Reiss) were
invited to participate in the HAS group.3 One hundred and thirtynine students participated in the study. One participant was excluded from analyses, as a result of being an outlier in age (8.16
years above the rest of the sample’s mean). Sixty-eight LAS
students (63.2% female) and 70 HAS students (65.7% female)
were included in analyses. The mean age was 18.84 (SD ⫽ .93,
range ⫽ 17 – 22 years). Seventy-one percent of participants
reported their race as White, 14.5% as Asian, 8.0% as Black/
African American, 5.1% as Bi- or multiracial, and 1.4% as “other.”
The University of Virginia’s Institutional Review Board (IRB)
approved this study.
Materials.4
Anxiety symptoms. The Anxiety Sensitivity Index (ASI; Reiss
et al., 1986) is a 16-item questionnaire that measures an individual’s concern over symptoms associated with anxiety (e.g., It
scares me when my heart beats rapidly). The ASI has good
reliability and validity (Peterson & Reiss, 1993), and includes
items relevant to physical, social, and mental incapacitation concerns. Cronbach’s alpha for the ASI was .95 in the current study.
The Positive and Negative Affect Schedule-Fear Subscale
(PANAS-FS; Watson & Clark, 1994) is a widely used self-report
measure of affect based on adjective ratings. The PANAS has good
reliability and validity (Watson & Clark). In the current study, only
the 6-item fear subscale was administered to determine if the
physical sensations prime affected state fear. Across administrations, the average Cronbach’s alpha was .82 (range ⫽ .74 –.87).
Physical sensation prime manipulation. Participants were
randomly assigned to either a physical sensation Prime or No
Prime condition to test the impact of a prime on the expression of
expectancies. To prime physical sensations relevant to anxiety
(and, in turn, presumably alter state fear for the HAS group),
participants in the Prime condition were asked to complete the
Candle-Blowing task. This task was derived from the widely used
Panic Control Treatment manual (Barlow & Craske, 1994) and has
been used in previous studies examining anxiety reactions (e.g.,
Gordon & Teachman, 2008; Steinman & Teachman, 2010). In the
Candle-Blowing task, participants were asked to imagine that their
index finger was a candle that they must blow out repeatedly for 45
seconds. To standardize the tempo of breathing, participants were
asked to blow with the beat of a metronome set to 100 beats per
minute. Although this task is harmless, it produces temporary
physical sensations, such as sweating, numbness, dizziness, hot
flashes, and tingling.
In the No Prime condition, participants were asked to work on
a Word Search Task, which was related to animals. This task was
designed to match the conditions for overall time, but not to prime
anxiety-sensitive relevant concerns or alter state fear.
Expectancy measure. The Expectancy Task (Cabeleira et al.,
2010) used in Study 1 was also used in Study 2 to evaluate the
tendency to expect relatively positive versus negative future events
to occur in the described scenarios. However, this study presented
only the 32 scenarios relating to physical concerns (and excluded
the 32 scenarios relevant to social concerns). Additionally, given
the Expectancy Task was originally developed in Australia, minor
modifications were made so that the wording was more prototypical of American English (e.g., “queuing” was changed to “waiting
in line”).
Physical sensation stressor. To evaluate how expectancy bias
is related to anxiety following an anxiety sensitivity-relevant stressor, participants were asked to complete the Straw Breathing task.
Note this task was completed after the Expectancy Task by all
participants, unlike the Candle-Blowing task, which was a
between-subjects manipulation that preceded the Expectancy Task.
Similar to the Candle-Blowing task, the Straw Breathing task was
derived from the widely used Panic Control Treatment manual
(Barlow & Craske, 1994) and has been used in previous studies
examining anxiety reactions (e.g., Gordon & Teachman, 2008;
Steinman & Teachman, 2010). In the Straw Breathing task, participants were asked to breathe through a thin straw for up to two
minutes, while holding their nostrils shut. Similar to the CandleBlowing task, the straw breathing task is harmless, but elicits
temporary sensations, such as dizziness, suffocation, and lightheadedness. Anxiety was measured by the PANAS-FS by asking
participants to indicate how they felt when their anxiety was at its
peak during the task.
Procedure. Participants were informed that the purpose of the
study was to investigate how people decide what happens next
after reading short stories. Participants were unaware that they
were recruited for the study based on their level of anxiety sensi3
To be invited to the LAS group, individuals also had to score a 9.71 or
below on the ASI Physical Concerns Subscale (ASI-PC; matching the one
standard deviation cutoff above the mean of a healthy nonanxious sample;
Teachman, Smith-Janik, & Saporito, 2007). To be invited to the HAS
group, individuals also had to score a 12.66 or greater on the ASI-PC
(matching the one standard deviation cutoff below the mean of a sample
with panic disorder; Teachman, Smith-Janik, & Saporito, 2007).
4
The materials reported are part of a larger study assessing cognitive
biases in anxiety sensitivity. For a complete listing of measures, please
contact Shari A. Steinman.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
EXPECTANCY BIAS AND ANXIETY
tivity. After informed consent, participants completed a brief demographics form, followed by the PANAS-FS to get a baseline
measure of state fear. Next, participants were assigned to the Prime
(n ⫽ 70) or No Prime (n ⫽ 68) condition. The prime conditions
were balanced for AS group and for gender. Participants in the
Prime condition completed the Candle-Blowing task, while those
in the No Prime condition completed the Word Search task. The
PANAS-FS was administered after both tasks. Next, all participants completed four practice Expectancy Task scenarios that were
unrelated to physical concerns, followed by the full Expectancy
Task (and the PANAS-FS). All participants then completed the
Straw Breathing physical sensation stressor followed by the
PANAS-FS. Next, participants completed a final administration of
the PANAS-FS to ensure they did not have residual anxiety at the
end of the study. Finally, all participants were fully debriefed.
Results
Descriptive statistics. As predicted, chi-square tests revealed
that the LAS and HAS groups did not differ by gender (2 ⫽ .09,
p ⫽ .761) or race (2 ⫽ 5.77, p ⫽ .217). An independent samples
t test demonstrated there was not a significant difference between
the LAS and HAS groups in terms of age, t(136) ⫽ 1.88, p ⫽ .063,
d ⫽ .32. As expected, the LAS group had significantly lower
baseline levels of state anxiety (M ⫽ 6.62, SD ⫽ 1.43) relative to
the HAS group (M ⫽ 7.81, SD ⫽ 2.52; t(107.87) ⫽ 3.42, p ⫽ .001,
d ⫽ .58). Additionally, as expected, chi-square tests revealed that
gender ratio (2 ⫽ .09, p ⫽ .761) and race (2 ⫽ 3.05, p ⫽ .550)
did not differ between participants assigned to the Prime and No
Prime conditions. Further, independent samples t tests revealed
that there was no significant difference between the prime conditions in baseline fear as measured by the PANAS-FS, t(135) ⫽ .78,
p ⫽ .439, d ⫽ .13 or age, t(136) ⫽ 1.63, p ⫽ .106, d ⫽ .28. See
Table 3 for descriptive statistics separated by Anxiety Group and
Prime Condition.
Effect of prime condition on state fear. To evaluate the
effect of the physical sensation prime condition on state fear
(measured by the PANAS-FS), a repeated measures ANOVA with
two between-subjects factors: Anxiety Group (LAS, HAS) and
Prime Condition (Prime, No Prime), and one within-subjects factor: Time (Baseline, Post Prime Condition Task), was conducted.
Results indicated a main effect of Time, such that on average, all
participants reported higher levels of state fear following the prime
condition task relative to baseline, F(1, 132) ⫽ 5.30, p ⫽ .023,
p2 ⫽ .04. Not surprisingly, there was also a main effect of Anxiety
group, indicating that participants in the HAS group reported
higher levels of state fear than participants in the LAS group, F(1,
132) ⫽ 15.85, p ⬍ .001, p2 ⫽ .11. Finally, there was the expected
significant Time by Anxiety group by Prime Condition interaction,
F(1, 132) ⫽ 5.93, p ⫽ .016, p2 ⫽ .04.
Follow-up tests to understand the interaction showed that, for
the LAS group, there was a main effect of Prime Condition, such
that participants in the Prime condition reported higher state fear
than those in the No Prime condition, F(1, 65) ⫽ 5.14, p ⫽ .027,
p2 ⫽ .07. However, no main or interactive effects with Time
emerged (all p ⬎ .10). For the HAS group, there was not a
significant main effect of Prime Condition, F(1, 67) ⫽ .38, p ⫽
.543, p2 ⫽ .01, or Time, F(1, 67) ⫽ 3.81, p ⫽ .055, p2 ⫽ .05, but
there was the expected significant Time by Prime Condition in-
9
teraction, F(1, 67) ⫽ 6.50, p ⫽ .013, p2 ⫽ .09. As expected, for
HAS participants in the No Prime condition, reported state fear
was not significantly different at the two time points, t(33) ⫽ .43,
p ⫽ .673, d ⫽ .07. However, for HAS participants in the Prime
condition, reported state fear was significantly higher following
the Candle-Blowing task, relative to baseline, t(34) ⫽ 3.16, p ⫽
.003, d ⫽ .55. Overall, these results suggest that the Candle
Blowing prime increased state fear for the HAS group, but did not
change state fear over time for the LAS group, supporting interpretation of the prime as an anxiety sensitivity-relevant stressor.5
Of note, on average, PANAS-FS scores following the prime condition tasks were low (M ⫽ 7.49, SD ⫽ 2.24, range ⫽ 6 –17),
suggesting that although the prime increased fear for the HAS
group, the prime was somewhat weak.
Evidence for expectancy effects.6 As in Study 1, an expectancy bias index was calculated by subtracting the likelihood of
future negative events from the likelihood of future positive
events. A summary of the mean probability ratings for negative
and positive future event statements is presented in Table 4, with
mean ratings organized by Passage Valence condition (Negative
Valence, Positive Valence, Conflicting Valence), Anxiety Group
(LAS, HAS), and Prime condition (No Prime, Prime). A repeated
measures ANOVA with two between-subjects factors: Anxiety
Group (LAS, HAS) and Prime Condition (Prime, No Prime), and
one within-subjects factor: Passage Valence Condition (Conflicting Valence, Positive Valence, Negative Valence) was conducted.
As in Study 1, the ANOVA revealed a main effect of Anxiety
Group (F(1, 134) ⫽ 4.02, p ⫽ .047, p2 ⫽ .03), such that the HAS
group had a more negative/less positive expectancy bias index
score (M ⫽ 0.53, SD ⫽ 0.43) than the LAS group (M ⫽ 0.68,
SD ⫽ 0.44). Thus, this second study confirmed that anxious
(compared to nonanxious) participants demonstrated a lowered
relative expectation for positive (vs. negative) future events.7 As
can be seen in Figure 4, the average expectancy bias score (averaged across the difference passage valence conditions) was significantly different from zero, suggesting a general tendency to expect positive relative to negative future events, t(137) ⫽ 16.18,
p ⬍ .001, d ⫽ 1.38. However, the tendency to display a positive
expectancy bias was lower for HAS participants, relative to LAS
participants. Also as in Study 1, there was a main effect of Passage
Valence Condition, F(2, 133) ⫽ 52.45, p ⬍ .001, p2 ⫽ .44.
Follow-up analyses showed that expectancy bias index scores
across the three passage valence types all significantly differed
from each other in the anticipated direction (all p ⬍ .001), with
5
Note that PANAS data were positively skewed, so analyses were rerun
with log transformed data. This did not change the pattern of results.
Additionally, each group (i.e., LAS No Prime, LAS Prime, HAS No Prime,
HAS Prime) had at least one outlier at both time points (baseline PANASFS, and after the Prime condition administration of PANAS-FS) before
transformation, so it is unlikely that outliers specific to one group are
driving effects. Outliers were defined as data points at least 1.5 box lengths
outside of interquartile range when looking at boxplots.
6
Similar to Study 1, an ANOVA revealed that the order of events (e.g.,
negative-neutral vs. neutral-negative) presented in the Scenario Presentation Component of the task did not influence the ratings made in the
Expectancy Rating Component of the task (all p ⬎ .05). Thus, order of
events was not included as a factor in the analyses presented.
7
Note that when the outlier attributable to age is included, this main
effect becomes F(1, 135) ⫽ 3.40, p ⫽ .067, 2p ⫽ .03.
CABELEIRA ET AL.
10
Table 3
Study 2: Descriptive Statistics for No Prime and Prime Conditions in Low and High Anxiety
Sensitive Groups
Low Anxiety Sensitive (LAS) n ⫽ 68
High Anxiety Sensitive (HAS) n ⫽ 70
Measures
No prime condition
n ⫽ 33
M ⫾ SD
Prime condition
n ⫽ 35
M ⫾ SD
No prime condition
n ⫽ 35
M ⫾ SD
Prime condition
n ⫽ 35
M ⫾ SD
Age
ASI
PANAS-FS
18.82 ⫾ 0.98
7.44 ⫾ 3.64
6.24 ⫾ 0.66
18.57 ⫾ 0.78
7.71 ⫾ 3.32
6.97 ⫾ 1.82
19.11 ⫾ 1.02
34.89 ⫾ 8.67
7.88 ⫾ 2.63
18.86 ⫾ 0.88
35.93 ⫾ 8.49
7.74 ⫾ 2.44
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Note. ASI ⫽ Anxiety Sensitivity Index; PANAS-FS ⫽ Positive and Negative Affect Schedule – Fear Subscale
(Baseline).
Negative Valence passages (M ⫽ 0.09, SD ⫽ 0.77) yielding a
relatively less positive expectancy index compared to Conflicting
Valence passages (M ⫽ 0.63, SD ⫽ 0.53), which in turn were less
positive than Positive Valence passages (M ⫽ 1.10, SD ⫽ 0.71).
These findings replicate Study 1. Notably, there was not a significant Anxiety Group by Passage Valence Condition interaction,
F(2, 133) ⫽ .91, p ⫽ .406, p2 ⫽ .01. Hence, both anxious and
nonanxious individuals’ expectations concerning future events
were equally affected by the valence of initial events that occurred
in each scenario, indicating that extrapolation from the valence of
initial events did not vary by anxiety level, though the HAS group
expected relatively less positive future events in general. This
replicates the pattern of effects shown by HTA participants on
these same physical passages in Study 1.
Finally, there were no main or interactive effects of Prime
Condition (all p ⬎ .10), suggesting that prime condition did not
affect expectancy. This is consistent with some prior null findings
for the effect of anxiety-relevant primes on cognitive processing
biases (e.g., Watts & Dalgeish, 1991), but counter to the expectation that the prime would make the physical concerns salient,
enhancing the expression of an expectancy bias.
Relationship between expectancy bias and response to the
stressor. To test whether the expectancy bias is related to in vivo
fear responding, a correlation was computed between the expectancy index score and fear in response to the physical sensation
stressor (Straw-Breathing). Scores on the PANAS-FS following
the physical sensation stressor ranged from 6 to 24, with a mean of
11.69 (SD ⫽ 4.36). As expected, results suggested that a less
positive expectancy bias is related to greater state fear during the
physical sensation stressor, as measured by the PANAS-FS
(r(135) ⫽ ⫺.19, p ⫽ .023).
Discussion
As was found for HTA individuals in Study 1, participants with
HAS also displayed a less positive expectancy bias relative to
participants with LAS. Furthermore, replicating the pattern of
effects shown on physical scenarios by HTA participants in Study
1, the anxiety-linked expectancy bias observed on these physical
scenarios in Study 2 was unaffected by the valence of the preceding events that occurred within these physical scenarios. Although
the current study included a larger sample and thus greater power
to detect significant effects than Study 1, the replicated absence of
an interaction with valence still needs to be interpreted with
caution given it relies on a null result. However, Pervasive-EBH
remains the most parsimonious account for the obtained results.
Given that Study 2 only included physical scenarios, we do not
know what type of expectancy bias HAS individuals would have
when presented with scenarios relevant to other domains (e.g.,
social). In retrospect, it would have been ideal to include social
scenarios along with physical scenarios in Study 2; however, for
practical reasons (e.g., added time it took to do candle blowing and
Table 4
Study 2: Summary of Mean Probability Ratings for Negative and Positive Future Event
Statements, With Mean Ratings Organized by Passage Valence Condition (Negative Valence,
Positive Valence, Conflicting Valence), Anxiety Group (LAS, HAS), and Prime Condition (No
Prime, Prime)
Low Anxiety Sensitive
(LAS)
Passage valence
Negative valence
Positive valence
Conflicting valence
High Anxiety Sensitive
(HAS)
Future valence
No prime
n ⫽ 33
M ⫾ SD
Prime
n ⫽ 35
M ⫾ SD
No prime
n ⫽ 35
M ⫾ SD
Prime
n ⫽ 35
M ⫾ SD
Negative
Positive
Negative
Positive
Negative
Positive
2.3 ⫾ 0.6
2.6 ⫾ 0.5
1.8 ⫾ 0.4
3.1 ⫾ 0.5
2.1 ⫾ 0.5
2.9 ⫾ 0.3
2.5 ⫾ 0.5
2.4 ⫾ 0.4
1.9 ⫾ 0.4
3.1 ⫾ 0.5
2.2 ⫾ 0.5
2.9 ⫾ 0.4
2.4 ⫾ 0.5
2.5 ⫾ 0.5
1.9 ⫾ 0.5
3.0 ⫾ 0.4
2.3 ⫾ 0.4
2.9 ⫾ 0.3
2.4 ⫾ 0.4
2.5 ⫾ 0.5
2.0 ⫾ 0.5
3.0 ⫾ 0.5
2.3 ⫾ 0.4
2.8 ⫾ 0.4
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
EXPECTANCY BIAS AND ANXIETY
Figure 4. Study 2: Expectancy bias by valence condition and level of
anxiety sensitivity. Expectancy bias was scored by subtracting average
ratings of negative future events from average ratings of positive future
events, so higher scores indicate a more positive bias.
straw breathing tasks) we were unable to include the social scenarios in this study.
Further, Study 2 demonstrated a relationship between expectancy bias index scores and fear experienced during a physical
sensation stressor. This suggests that expectancy bias is related to
actual anxiety experienced in response to a stressor (vs. only a
questionnaire measure of trait levels of anxiety sensitivity), and
highlights the predictive validity of expectancy bias. Finally, results suggested that priming a physical stressor did not independently or interactively affect expectancy bias. Although this is
contrary to the hypothesis that the prime would make physical
concerns salient, and in turn augment the anxiety-linked expectancy bias, it is consistent with past null findings for the effect of
anxiety-relevant primes on processing biases (e.g., Watts & Dalgeish, 1991). Future studies might consider using a stronger prime
(that increases state anxiety more than the prime used in the
current study) to evaluate whether a stronger physical prime would
have a significant impact on the anxiety-linked expectancy bias.
General Discussion
In this article, we report two studies that use the Expectancy
Task, a recently developed paradigm that provides individuals with
different forms of valenced information, and then evaluates their
tendency to expect positive or negative events to occur in the
future. The studies shared two goals: a) to examine expectancy
bias in anxious (HTA and HAS) samples, and b) to determine the
conditions that give rise to expectancy bias. Study 2 had the
additional goal of exploring the effect of priming physical sensations on expectancy.
In both Studies 1 and 2, high anxious (both trait anxious and
anxiety sensitive) groups showed a more negative/less positive
expectancy bias relative to low anxious individuals. In this respect,
the results confirm previous evidence of threat expectancy bias in
trait anxious participants (e.g., MacLeod et al., 1997; Miranda &
11
Mennin, 2007) and extend this evidence to anxiety sensitive participants. Additionally, the results suggested a relationship between expectancy bias and various indicators of anxiety, such that
a greater negative expectancy bias was associated with higher
levels of subjective fear experienced during a stressor. This further
supports an association between anxiety and expectancy bias.
An important feature of the Expectancy Task is that it is designed not only to determine whether an expectancy bias exists,
but also to illuminate possible cognitive underpinnings of this bias
by evaluating the experimental conditions that give rise to the bias.
In Study 1, for passages related to social concerns, the HTA group
only displayed elevated expectancy for negative future events
when initial scenarios included negative events, or a mix of negative and positive events. Although speculative, one reason that the
negative expectancy bias may diminish in the context of positive
social scenarios may be that even HTA individuals likely have
more experience with social interactions going well, so this may
seem more plausible than a potential physical concern turning out
well. Along these lines, it will be interesting in future research to
evaluate the expectancy bias in highly socially anxious individuals,
given this group may more pervasively expect social interactions
to go badly.
In Study 2, which used only the scenarios related to physical
concerns, HAS participants demonstrated heightened expectancy
for less positive/more negative future events. This was unaffected
by the valence of the information provided concerning the initial
events that previously took place in each scenario, and is in line
with the Pervasive-EBH. This replicates results from the physical
scenarios from Study 1, and is in line with findings reported by
Steinman et al. (2013) from an online version of the Expectancy
Task given to an unselected sample varying in trait anxiety. Past
research has suggested that anxious individuals tend to disqualify,
or not learn from, past positive experiences (e.g., Beck, 1976;
Heimberg & Becker, 2002), which may explain why the HAS
group in Study 2 expected negative future events to occur, even
following Positive Valence or Conflicting Valence scenarios. Future attempts to manipulate expectancy bias, as is done in cognitive
bias modification research (e.g., Mathews & Mackintosh, 2000),
can help determine whether the bias is causally related to the onset
or maintenance of anxiety problems.
It is intriguing that valence moderated the anxiety-linked expectancy bias when scenarios were related to social concerns (in Study
1), but not when events were related to physical concerns (in both
Studies 1 and 2). It would be interesting to determine whether HAS
individuals also show greater extrapolation from negative events
when processing social scenarios. If they do, this would suggest
that the tendency for anxious individuals to extrapolate more
strongly from prior negative events is more readily observed when
these negative events are of a social nature. In contrast, if HAS
individuals do not display this pattern of results even when processing social scenarios, this would instead suggest that the mechanisms that underpin negative expectancy bias may differ across
anxious samples.
Although we have demonstrated that expectancy bias, observed
for the physical and social scenarios, differs in groups of participants selected on the basis of their differing levels of anxiety, this
does not permit the conclusion that anxiety alone is implicated in
the observed effect. Anxiety is correlated with other dimensions of
temperament, such as depression and neuroticism (Jylhä & Isom-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
12
CABELEIRA ET AL.
etsä, 2006). Therefore it would be valuable to include measures of
depression, and perhaps neuroticism, in addition to anxiety, in
future studies of this nature, to potentially differentiate their respective contribution to observed patterns of expectancy bias. We
speculate that the observed expectancy biases are not unique to
anxiety alone, given the prominence of a negative future focus
(e.g., hopelessness) in other disorders, like depression. However,
the specific nature of expectancy biases may differ in anxiety and
in depression. For example, it may be that attenuation of positive
expectancies may characterize depression, whereas amplification
of negative expectancies may be more characteristic of anxiety.
Delineating the patterns of expectancy associated with each dimension of emotion will be an important objective for future
research.
It is important to consider the potential role of response bias in
the observed effects. A response bias effect, reflecting an inflated
tendency to endorse more extreme negative responses, regardless
of actual expectancy, could mimic an expectancy bias in this kind
of assessment task. The fact that we obtained different findings for
the physical and social scenarios weighs against the plausibility of
a response bias account. Moreover, the finding that anxiety-linked
expectancy effects were moderated by the valence of previous
events precludes a response bias explanation of the present group
difference. Nevertheless, it would be prudent for future research to
more directly assess for response bias effects. For instance, if
negative and positive foil items (reflecting future events that do not
plausibly follow from the previously presented scenarios) were
included in the Expectancy Task, then negative response bias
would be directly revealed by preferential endorsement of the
negative foils. Further, to mitigate the impact of such response
bias, future investigators may usefully seek to develop methods of
assessing expectancy that bypass the need for self-report. For
example, it may be possible to infer a higher level of expectancy
for certain continuation events on the basis of speeded comprehension latencies to encode such events in a self-paced reading
task. More implicit assessment of expectancy bias would also
reduce the degree to which observed effects could plausibly be
attributed to experimenter demand.
Of note, neither of the present studies found evidence to support
the Emotional Weighting-EBH. In other words, in both studies, the
anxiety-linked expectancy bias was not found to be the strongest in
the Conflicting Valence condition. This finding suggests that anxious individuals’ heightened expectancy for future negative events
(relative to nonanxious individuals) cannot be solely attributed to
the conflicting nature of previously presented information and the
construed emotional valence of such information. Given the current finding, it seems unlikely that anxious individuals’ less positive expectations of the future would be remediated by only
reducing their tendency to misconstrue emotionally contradictory
information as predominantly negative in tone. Rather, it may be
beneficial for clinicians to directly target anxious clients’ biased
patterns of future expectations across both emotionally conflicting
and nonconflicting situations, to encourage the development of an
increased expectation for positive events and a reduced expectation for negative events.
In Study 1, when observing expectancies within each anxiety
group, LTA participants were more positive in their expectancies following positive information alone, compared with negative information alone or information conflicting in emotional
valence. However, they tended to have a positive absolute bias
(e.g., above zero) in their expectancies across all of these
differently valenced conditions, and this was true for both
social and physical scenarios. In contrast, the expectancies of
HTA participants were more consistent with the valence of
previous information, with a positive absolute bias when previous information was positive, a negative bias when previous
information was negative, and no expectancy bias (neither
positive or negative) when preceding information was of a
conflicting valence. In Study 2, both high and low anxiety
sensitive participants tended to have a positive absolute expectancy bias across passage valence conditions, highlighting the
importance of looking at both relative and absolute bias.
Finally, no effects of priming physical sensations on expectancy bias were found in Study 2. This is consistent with past
null findings for effects of anxiety-relevant primes on processing biases (e.g., Watts & Dalgeish, 1991), but inconsistent with
other findings suggesting that state anxiety can enhance (MacLeod & Mathews, 1988) or attenuate (Mathews & Sebastian,
1993) cognitive biases, and contrary to our hypothesis that the
prime would highlight physical concerns and magnify the
anxiety-linked expectancy bias. Given that the prime findings
were null, it is difficult to tease apart whether the lack of effects
is meaningful or attributable to a methodological issue. One
possibility is that the effect of the prime may have simply
dissipated quickly, obscuring the opportunity to see its effects.
Another possibility is that the prime’s effect was not strong
enough to affect expectancy bias. Given the mixed findings in
the literature about the effects of state anxiety on cognitive
biases more broadly, it is clear that more work is needed to
determine the relationship between state anxiety and expectancy bias, and the moderators of this relationship.
The current research has some limitations. First, the samples
comprised only young adult, college students, which are not representative of the general population. Second, analogue, rather
than clinical, samples were used. However, the anxious participants had very high levels of anxiety symptoms, similar to those
found in diagnosed samples. Third, as mentioned earlier, depressive symptoms were not assessed in these studies. More focused
investigation of depressive symptoms would be helpful in evaluating the unique contribution of depression and anxiety to different
forms of expectancy bias.
Fourth, our choice to use a relative strategy to calculate the
bias index (i.e., subtracting average ratings of negative future
events from average ratings of positive future events) has the
advantage of simultaneously accounting for ratings of both
positive and negative events, but has the disadvantage that we
cannot determine to what extent the observed expectancy biases
are driven by an evaluation that positive effects are especially
unlikely or that negative effects are especially likely; an interesting question for future research. Similarly, interpretation of
within-groups effects (e.g., evaluating whether bias is positive,
negative, or does not exist) are somewhat clouded by our use of
a relative strategy to calculate bias, given that we do not have
an objective measure of whether the positive and negative
events occur equally often (i.e., whether they have, or are perceived to have, comparable base rates). However, several measures
were taken to validate the Scenario Event Sets (e.g., ensuring that
positive and negative events had equivalent valence intensity; see
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
EXPECTANCY BIAS AND ANXIETY
Cabeleira, 2010), which allows for more confidence in interpreting
the results.
Fifth, although it seems unlikely that presenting scenarios in
blocks of four (before requiring ratings to be made) would place
significant demands on memory for younger adults, there was no
explicit measure of the role of memory in this task. This leaves
open the possibility that memory may influence the results found
when using this task. Future research could explore this possibility
by varying the number of scenarios presented before the Rating
Component of the task. Lastly, it is possible that participants’
perceptions of the frequency with which the candidate future
events tend to occur in general (i.e., perceived base rates) might
have influenced their responses, separately from their perceptions
concerning the likelihood that the candidate events would occur
for themselves in particular. It could be interesting, in future
research, to ask participants to provide two expectancy ratings:
how likely each event is to occur to others, and how likely each
event is to occur to oneself.
Despite these limitations, these studies provide valuable new
information about anxiety-linked expectancy bias, revealed
through the recently developed Expectancy Task. The results
confirm that anxious individuals do display a less positive/more
negative expectancy bias. Furthermore, they demonstrate that
the conditions that give rise to this bias may differ depending
upon the type of information being processed, with the inflated
expectation of future negative social events reflecting a heightened tendency to extrapolate from present negative social
events, while the inflated expectation of future negative physical events occurred regardless of the valence of prior information. A better understanding of the cognitive conditions that
give rise to the negative expectancy bias shown by anxious
individuals may not only shed light on why they “expect the
worst,” but could also identify the processes that can be targeted
therapeutically to attenuate such expectations.
References
Barlow, D., & Craske, M. (1994). Mastery of your anxiety and panic-II.
San Antonio, TX: The Psychological Corporation.
Barnes, L. L. B., Harp, D., & Jung, W. S. (2002). Reliability generalization
of scores on the Spielberger State-Trait Anxiety Inventory. Educational
and Psychological Measurement, 62, 603– 618. doi:10.1177/
0013164402062004005
Beck, A. T. (1976). Cognitive therapy and the emotional disorders. Oxford, England: International Universities Press.
Beck, A. T., & Clark, D. A. (1988). Anxiety and depression: An information processing perspective. Anxiety Research, 1, 23–36. doi:10.1080/
10615808808248218
Borkovec, T. D., Alcaine, O. M., & Behar, E. (2004). Avoidance theory of
worry and generalized anxiety disorder. In R. G. Heimberg, C. L. Turk,
& D. S. Mennin (Eds.), Generalized anxiety disorder: Advances in
research and practice (pp. 77–108). London, England: Guilford Press.
Cabeleira, C. (2010). Age differences in emotional appraisal (Unpublished
honor’s thesis). University of Western Australia, Australia.
Cabeleira, C., Bucks, R., Teachman, B. A., & MacLeod, C. (June, 2010).
Emotional appraisal of ambiguous scenarios in older and younger
adults. Poster presented at the annual meeting of the World Congress of
Behavioral and Cognitive Therapies, Boston, MA.
Clark, D. M. (1986). A cognitive approach to panic. Behaviour Research
and Therapy, 24, 461– 470. doi:10.1016/0005-7967(86)90011-2
13
Cox, B. J., Parker, J. D. A., & Swinson, R. P. (1996). Anxiety sensitivity:
Confirmatory evidence for a multidimensional construct. Behaviour
Research and Therapy, 34, 591–598. doi:10.1016/00057967(96)00006-X
Dugas, M. J., Freeston, M. H., Ladouceur, R., Rhéaume, J., Provencher,
M., & Boisvert, J. (1998). Worry themes in primary GAD, secondary
GAD, and other anxiety disorders. Journal of Anxiety Disorders, 12,
253–261. doi:10.1016/S0887-6185(98)00013-9
Gordon, T. L., & Teachman, B. A. (2008). Ethnic group differences in
affective, behavioral, and cognitive markers of anxiety. Journal of
Cross-Cultural Psychology, 39, 424 – 446. doi:10.1177/
0022022108318224
Heimberg, R. G., & Becker, R. E. (2002). Cognitive-behavioral group
therapy for social phobia: Basic mechanisms and clinical strategies.
New York, NY: Guilford Press.
Jylhä, P., & Isometsä, E. (2006). The relationship of neuroticism and
extraversion to symptoms of anxiety and depression in the general
population. Depression and Anxiety, 23, 281–289. doi:10.1002/da.20167
Kendall, P. C., & Ingram, R. E. (1989). Cognitive-behavioral perspectives:
Theory and research on depression and anxiety. In P. C. Kendall, K. M.
Simon, L. E. Beutrler, H. Arkowitz (Eds.) Comprehensive handbook of
cognitive therapy (pp. 27–53). New York, NY: Plenum Press.
MacLeod, A. K., & Byrne, A. (1996). Anxiety, depression, and the anticipation of future positive and negative experiences. Journal of Abnormal
Psychology, 105, 286 –289. doi:10.1037/0021-843X.105.2.286
MacLeod, A. K., Tata, P., Kentish, J., & Jacobsen, H. (1997). Retrospective and prospective cognitions in anxiety and depression. Cognition and
Emotion, 11, 467– 479. doi:10.1080/026999397379881
MacLeod, C., & Mathews, A. (1988). Anxiety and the allocation of
attention to threat. The Quarterly Journal of Experimental Psychology
A: Human Experimental Psychology, 40, 653– 670. doi:10.1080/
14640748808402292
Mathews, A., & Mackintosh, B. (2000). Induced emotional interpretation
bias and anxiety. Journal of Abnormal Psychology, 109, 602– 615.
doi:10.1037/0021-843X.109.4.602
Mathews, A. M., & Sebastian, S. (1993). Suppression of emotional Stroop
effects by fear-arousal. Cognition and Emotion, 7, 517–530. doi:
10.1080/02699939308409203
Miranda, R., & Mennin, D. S. (2007). Depression, generalized anxiety
disorder, and certainty in pessimistic predictions about the future. Cognitive Therapy and Research, 31, 71– 82. doi:10.1007/s10608-0069063-4
Peterson, R. A., & Reiss, S. (1992). Anxiety sensitivity index manual (2nd
ed.). Worthington, OH: International Diagnostics Systems.
Peterson, R. A., & Reiss, S. (1993). Anxiety Sensitivity Index Revised test
manual. Worthington, OH: IDS Publishing Corporation.
Pratto, F., & John, O. P. (1991). Automatic vigilance: The attentiongrabbing power of negative social information. Journal of Personality
and Social Psychology, 61, 380 –391. doi:10.1037/0022-3514.61.3.380
Reiss, S. (1991). Expectancy model of fear, anxiety, and panic. Clinical
Psychology Review, 11, 141–153. doi:10.1016/0272-7358(91)90092-9
Reiss, S., & McNally, R. J. (1985). The expectancy model of fear. In S.
Reiss & R. R. Bootzin (Eds.), Theoretical issues in behavior therapy (pp.
107–121). New York, NY: Academic Press.
Reiss, S., Peterson, R. A., Gursky, D. M., & McNally, R. J. (1986).
Anxiety sensitivity, anxiety frequency and the predictions of fearfulness. Behaviour Research and Therapy, 24, 1– 8. doi:10.1016/00057967(86)90143-9
Smith-Janik, S. B., & Teachman, B. A. (2008). Impact of priming on
explicit memory in spider fear. Cognitive Therapy and Research, 32,
291–302. doi:10.1007/s10608-007-9122-5
Spielberger, C. D. (1983). Manual for the State-Trait Anxiety Inventory
(Form Y). Palo Alto, CA: Consulting Psychologists Press.
CABELEIRA ET AL.
14
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Steinman, S. A., & Teachman, B. A. (2010). Modifying interpretations
among individuals high in anxiety sensitivity. Journal of Anxiety Disorders, 24, 71–78. doi:10.1016/j.janxdis.2009.08.008
Steinman, S. A. Smyth, F. L., Bucks, R. S., MacLeod, C., & Teachman,
B. A. (2013). Anxiety-linked expectancy bias across the adult lifespan. ⬎Cognition and Emotion, 27, 345–355. doi:10.1080/02699931.2012.711743
Stöber, J. (2000). Prospective cognitions in anxiety and depression: Replication and methodological extension. Cognition and Emotion, 14,
725–729. doi:10.1080/02699930050117693
Teachman, B. A., Smith-Janik, S. B., & Saporito, J. (2007). Information
processing biases and panic disorder: Relationships among cognitive
and symptom measures. Behaviour Research and Therapy, 45, 1791–
1811. doi:10.1016/j.brat.2007.01.009
Watson, D., & Clark, L. A. (1994). Manual for the positive and negative affect
schedule (expanded form). Iowa City, IA: University of Iowa.
Watts, F. N., & Dalgeish, T. (1991). Memory for phobia-related words in
spider phobics. Cognition and Emotion, 5, 313–329. doi:10.1080/
02699939108411043
Wheaton, M. G., Berman, N. C., & Abramowitz, J. S. (2010). The contribution of experiential avoidance and anxiety sensitivity in the prediction
of health anxiety. Journal of Cognitive Psychotherapy, 24, 229 –239.
doi:10.1891/0889-8391.24.3.229
Appendix
Examples of Scenarios and Components of the Expectancy Task
Example of a Scenario Event Set
Title: Going to the Doctor
Orienting Sentence: You go to the doctor’s rooms.
Negative: You find out you need a biopsy done.
Negative: The doctor prescribes you medication that can have
bad side effects.
Negative: The doctor warns you all your family is at risk of
diabetes.
Positive: The doctor says your heart sounds very healthy.
Positive: The doctor informs you that you are at a healthy
weight.
Positive: The doctor says she is happy with your exercise
regime.
Neutral: A bird flies past the window.
Neutral: The telephone rings.
Neutral: You notice a car drive by outside.
Negative: You find out you need a biopsy done.
Negative: The doctor prescribes you medication that can have
bad side effects.
Neutral: A bird flies past the window.
Neutral: You notice a car drive by outside.
Conflicting Valence
Title: Going to the Doctor
Orienting Sentence: You go to the doctor’s rooms.
Negative: You find out you need a biopsy done.
Negative: The doctor prescribes you medication that can have
bad side effects.
Positive: The doctor informs you that you are at a healthy
weight.
Positive: The doctor says she is happy with your exercise
regime.
Sample Scenarios of Different Valence Types Based on
Above Scenario Event Set
Example of an Expectancy Rating Trial
Positive Valence
Title: Going to the Doctor
Orienting Sentence: You go to the doctor’s rooms.
Positive: The doctor informs you that you are at a healthy
weight.
Positive: The doctor says she is happy with your exercise
regime.
Neutral: A bird flies past the window.
Neutral: You notice a car drive by outside.
Negative Valence
Title: Going to the Doctor
Orienting Sentence: You go to the doctor’s rooms.
View publication stats
Title: Going to the Doctor
Orienting Sentence: You go to the doctor’s rooms.
How likely is it that . . .
Negative: The doctor warns you all your family is at risk of
diabetes.
Positive: The doctor says your heart sounds very healthy.
Neutral: The telephone rings.
Received March 7, 2013
Revision received October 8, 2013
Accepted December 12, 2013 䡲