The Journal of Social Psychology, 2009, 149(6), 609–626

Copyright © 2009 Heldref Publications

Personal, Interpersonal, and Situational

0022-4545
VSOC
The Journal of Social Psychology
Psychology, Vol. 149, No. 6, Oct 2009: pp. 0–0

Influences on Behavioral Self-Handicapping

CHRISTINA M. BROWN
The Journal
Brown & Kimble
of Social Psychology

Saint Louis University

CHARLES E. KIMBLE
University of Dayton

ABSTRACT. This study explored the combined effects of personal factors (participant
sex), interpersonal factors (experimenter sex), and situational factors (performance feed-
back) on two forms of behavioral self-handicapping. Participants received non-contingent
success or failure feedback concerning their performance on a novel ability and were
given the opportunity to self-handicap before performing again. Behavioral self-handicap-
ping took the form of (a) exerting less practice effort (practice) or (b) choosing a perfor-
mance-debilitating tape (choice). Men practiced least after failure feedback and chose a
debilitating tape if they were interacting with a female experimenter. Generally, across all
participants in both choice and practice conditions, high performance concern and the
presence of a male experimenter led to the most self-handicapping. Results are interpreted
in terms of self-presentational concerns that emphasize a desire to impress or an awareness
of the female or male experimenter’s acceptance of self-handicappers.
Keywords: choice, performance concern, practice, self-handicapping, sex

BEHAVIORAL SELF-HANDICAPPING is a self-protective strategy used when

people care about their performance but doubt the likelihood of their success,
especially in new endeavors (Berglas & Jones, 1978). Self-handicapping provides
an opportunity to diminish the threat of failure by obscuring low ability as the
reason for failure. In the case of behavioral self-handicapping, this is achieved by
actively creating an obstacle that impedes performance. However, while this
reduces internal attributions in the event of failure, it also makes failure more
likely to occur. The negative impact of behavioral self-handicapping on present

We are grateful for the substantial efforts of Christopher G. Murrer, Tarika Daftary,
Wei-Yi Chen, Christine M. Abraham, Lisa Boeckman, and Omar D. Carlo in helping us to
conduct this experiment.
Address correspondence to Dr. Christina M. Brown, Department of Psychology, Saint
Louis University, 221 N. Grand Blvd., St. Louis, MO 63109, USA; cbrown81@slu.edu (e-mail).

609
610 The Journal of Social Psychology

and future performance makes understanding the phenomenon important (cf.

Zuckerman, Kieffer, & Knee, 1998). The extensive literature in self-handicapping
has revealed that multiple factors—such as sex (Hirt, McCrea, & Kimble, 2000),
self-esteem (Harris & Snyder, 1986), self-focus (Kimble & Hirt, 2005), and self-
presentation concerns (Kolditz & Arkin, 1982)—determine whether self-
handicapping occurs and the form it takes.
This study examines both personal (who self-handicaps?) and situational
(when is handicapping likely to occur?) factors associated with behavioral self-
handicapping. It is important to note that behavioral self-handicapping is distinct
from self-reported handicapping. Self-reported self-handicapping involves
claiming a disadvantageous condition exists and thus provides an external
attribution for failure. In contrast, behavioral self-handicapping occurs when the
person acts to create a disadvantage by his or her own behavior, which not only
affects attributions but also directly impacts performance (Leary & Shepperd,
1986). Behavioral self-handicapping is particularly risky, because creating a
disadvantage for oneself in an evaluative situation is likely to make the individ-
ual fail or do poorly. Yet people have been shown to self-handicap behaviorally
in order to protect their positive self-evaluation (Berglas & Jones, 1978).
For example, the first experiment on behavioral self-handicapping by
Berglas and Jones (1978) showed that men who were given unfounded or non-
contingent success feedback on a first test chose a performance-debilitating drug
rather than a performance-facilitating one before a second test. Similarly, Kimble
and Bryant (2002) found that men made similar performance-hindering choices
after non-contingent success feedback on a first test. Kolditz and Arkin (1982)
and others (e.g., Shepperd & Arkin, 1989, 1991) have also used a choice between
performance-facilitating and performance-debilitating conditions to study behav-
ioral self-handicapping.
Alternatively, other investigators studied behavioral self-handicapping by
assessing whether participants practice or prepare less before an important evalu-
ation (Hirt, Deppe, & Gordon, 1991; Hirt et al., 2000; Kimble & Hirt, 2005;
Rhodewalt, Saltzman & Wittmer, 1984). These two approaches to examining
behavioral self-handicapping—choosing a handicapping condition for one’s
future performance (e.g., Berglas & Jones, 1978) and acting in a less effortful
manner that hinders future performance—are distinctive self-handicapping activ-
ities that may operate under different principles. Therefore, we examined these
two indices of behavioral self-handicapping in the same study to discover possi-
ble similarities and differences.
In this experiment, we manipulated feedback in three ways. We told some of
the participants that they had done very well on Test 1 when this was impossible
(non-contingent success feedback, as Berglas & Jones, 1978, had done); others
were told they had done very well on Test 1 when they most likely had
performed successfully (contingent success, as in Berglas & Jones); a third group
of participants were told that they had done poorly when it was very likely that
 Brown & Kimble 611

they had (failure). Berglas and Jones found that participants self-handicapped
after they had received non-contingent success feedback, which was presumed to
have undermined their certainty of future success, but not after contingent
success. Recently, Smith and Kimble (2003) found that failure feedback produced
more self-handicapping among men, while neither type of success feedback pro-
duced self-handicapping among men or women. We sought to discover whether
failure or non-contingent success was more likely to promote self-handicapping
in the present contexts. Non-contingent success encourages self-handicapping
because the individual does not know how his or her initial success was achieved
and is uncertain about whether it can be attained again. Thus, this creates an attri-
bution for future failure so his or her previous success cannot be called into ques-
tion (Berglas & Jones, 1978). Therefore, impression management is likely to be a
large factor in self-handicapping following non-contingent success (e.g., Kolditz
& Arkin, 1982). Failure, on the other hand, is unambiguous. Individuals who
self-handicap following failure may also be motivated by impression manage-
ment concerns (i.e., individuals who fail at a task and then practice considerably
are likely to feel even worse if they fail a second time; thus, choosing not to prac-
tice could prevent a further decrease in mood). In contrast, failure may also be
more likely to motivate increased effort (the opposite of self-handicapping) for
participants who see an opportunity for improvement. In other words, while non-
contingent success is likely to lead to self-handicapping, failure may lead to
either self-handicapping or increased effort.
A second situational factor that may influence self-handicapping is the pres-
ence of specific others and their evaluative implications. Kolditz and Arkin (1982)
showed that behavioral self-handicapping was more likely to occur under public
conditions than under private conditions, thereby calling attention to audience char-
acteristics as an important factor. Recent research has shown that audiences or
observers of self-handicapping behavior evaluate self-handicappers differently
(Hirt, McCrea, & Boris, 2003; Luginbuhl & Palmer, 1991). Luginbuhl and Palmer
showed that when participants saw a tape in which a man went to a movie instead
of studying for an exam, women predicted that this self-handicapper’s future test
score would be lower than men did. When he scored a C on the actual test, women
also rated his knowledge of the subject lower than men did. In a recent study, Hirt
et al. (2003) found that women view behavioral self-handicappers as irresponsible
and self-destructive more than men do, and women are more critical and skeptical
of self-handicappers’ motives. If women perceive self-handicapping acts in a nega-
tive way, then potential self-handicappers may behaviorally self-handicap less
when they are observed (and perhaps evaluated) by women. Therefore, we predict
that when a woman is the experimenter who observes, instructs, and informs the
participant, all participants will self-handicap less.
Throughout the literature, sex differences in self-handicapping inclinations
are evident. Specifically, men are more likely to self-handicap behaviorally than
women (Berglas & Jones, 1978; Harris & Snyder, 1986; Hirt, Deppe, & Gordon,
612 The Journal of Social Psychology

1991; Kimble & Hirt, 2005; Rhodewalt & Davison, 1986; Shepperd & Arkin,
1989, 1991). Hirt et al. (1991) showed that men who were habitual self-handicappers
practiced less for an upcoming test (behaviorally self-handicapped) than other
men and all women. Kimble and Hirt (2005) found that even when the self-
evaluative implications were made salient to women and men by manipulating
self-focus, women (unlike men) did not respond by self-handicapping behavior-
ally. Other studies in which behavioral self-handicapping has been found to
occur used only male participants (Greenberg, 1985; Higgins & Harris, 1988;
Kolditz & Arkin, 1982; Rhodewalt, Morf, Hazlett, & Fairfield, 1991; Shepperd
& Arkin, 1991, Experiment 2). Given the importance of sex in past research on
self-handicapping, this study highlights the role of sex in examining conditions
that produce self-handicapping.
In contrast, Oleson, Poehlmann, Yost, Lynch, and Arkin (2000) have taken an
individual differences approach to understanding overachievement and self-
handicapping. In their subjective overachievement scale, they have measured self-
doubt and performance concern as chronic traits. They have found that high self-
doubt and high performance concern predict subjective overachievement, which is
the tendency to perform better than expected because of extra effort, and that high
self-doubt and low performance concern predict trait self-handicapping (Jones &
Rhodewalt, 1982). To assess the impact of momentary levels of self-doubt and per-
formance concern, in the current study we created state measures of self-doubt and
performance concern that were administered before the self-handicapping opportu-
nity to see if they affected self-handicapping in the experimental situation.
Although these were not our primary factors of interest, we expected that high self-
doubt and low performance concern would lead to more self-handicapping.
This study was an attempt to answer many exploratory questions about behav-
ioral self-handicapping, and as such, no specific hypotheses were advanced. The
issues addressed were: (1) whether types of self-handicapping available (choosing
a hindering condition or exerting less practice effort) affect self-handicapping; (2)
whether different types of feedback affect self-handicapping; (3) whether men and
women are consistently different in behavioral self-handicapping; and (4) whether
being observed by men or women who differ in their approval of self-handicappers
affects self-handicapping. Our goal was to advance a dynamic understanding of
this psychological domain where achievement and self-evaluation desires meet.

METHOD

Measures

Current Feelings

After receiving feedback about their test performance, participants were

asked to complete a survey designed to assess their current feelings. The first part
 Brown & Kimble 613

of this survey contained ten questions to assess how participants currently felt
about themselves and how well they expected to do on the second test, such as “I
am confident that I will perform well,” “I will be disappointed with myself if I
perform poorly,” “It is important to me that I do well on this test,” and “I am
uncertain about how well I will do” (cf. Hirt et al., 2000).

State Self-Doubt and Performance Concern

Presented below, after the 10 current feelings items, were the self-doubt and
performance concern subscales of Oleson et al.’s (2000) achievement orientation
scale. These subscales were used to measure participants’ current or “state” feel-
ings of self-doubt and concern with performance. Examples of items on the per-
formance concern subscale are, “It is important that I succeed in all that I do,”
“Failure is unacceptable to me,” and “Sometimes I am more comfortable when I
lose or do poorly” (reverse-coded). Examples of items on the self-doubt subscale
are, “For me, avoiding failure has a greater emotional impact (e.g., sense of
relief) than the emotional impact of achieving success (e.g., joy, pride),” “More
often than not I feel unsure of my abilities,” and “I sometimes find myself
wondering if I have the ability to succeed at important activities.”

Participants and Procedure

The 126 students (64 women, 62 men) who participated in this study were
randomly assigned to either a tape condition or a practice condition. Furthermore,
participants within each condition were randomly assigned to one of three feed-
back conditions: contingent success, non-contingent success, and failure. Partici-
pants interacted with one of seven experimenters (4 women, 3 men) so we could
explore how audience sex may affect the propensity to self-handicapping.1
In both the practice and the tape conditions, participants were informed that
they would be taking two tests of integrative orientation (TIO), and they were
told the tests have been shown to be a consistent predictor of future success in
academics and in the workplace. Integrative orientation was further described as
an independent aspect of intelligence that is assessed by few other tests, and
they were led to believe that an organization conducting a national project
would be contacting people who performed exceptionally well. Practice-condition
participants were told that they would have the opportunity to practice before
the second test, although they could not practice before the first test. Tape-
condition participants were told that they would be listening to sound recordings
during the second test so we could compare the effect of the recordings on test
performance.
Participants were then given the first test, which consisted of 10 vocabulary
analogy questions. The students in the contingent success condition received a
normal solvable test, while both non-contingent success and failure participants
614 The Journal of Social Psychology

received an unsolvable version of the same test (the correct answer had been
changed so that none of the options were correct). This was done to make the test
seem more difficult, which should induce feelings of uncertainty about future
performance in the participants, who were told they had performed extremely
well on such a difficult test (non-contingent success). Likewise, participants
assigned to the failure condition were given the unsolvable test so they would not
be surprised when they received a low score. Next, participants were told they
would have 10 minutes to complete the test, and they were encouraged to answer
all questions, even if that meant guessing.
Following the test, participants were given false feedback about their test
results. Those in the failure condition were told that they had scored 3/10 correct
and that approximately 80% of the people tested before them had scored higher.
The feedback given in the contingent and non-contingent success conditions was
a score of 9 out of 10, which was said to be in the 95th percentile and the best
score seen at the University of Dayton since the beginning of testing. All partici-
pants were shown a laminated score card illustrating a bar graph of percentile
ranking by the number of problems correct, with a blue line indicating their score
and rank.
At this point, participants in the practice condition and tape condition were
given different instructions about the second test. The practice condition instruc-
tions emphasized the importance of practice, which subtly presented the opportu-
nity to self-handicap by implying that a low score does not necessarily mean the
test-taker has low ability if that person did not practice much:

On this test, you’ll be able to practice beforehand. This second subtest of the Test of
Integrative Orientation is negatively affected by lack of practice. In other words, peo-
ple who have not had much practice with these problems when they take the test tend
to get a score that is significantly below their true level of ability. These people who
are “under-practiced” have better integrative orientation abilities than their score on
the test would suggest. In other words, their lack of practice makes people appear to
have less integrative orientation ability on the test than they really have. Since we are
interested in the norms for the true level of integrative orientation and not simply test
scores, we need to control for the amount of practice in determining your level of inte-
grative orientation. Therefore, we are going to allow you to practice as much or as lit-
tle as you like before taking the test. Again, the amount of practice you put in will
improve the test’s ability to accurately measure your level of integrative orientation.

The option to practice for any length of time gave participants the opportu-
nity to self-handicap. Before practicing, participants were asked to complete the
Current Feelings survey, the state self-doubt, and the state performance concern
described earlier. Participants were then told that the second TIO would require
them to determine which figure should come next in a patterned sequence, and
they were given three pre-practice problems to insure that they understood the
directions. The experimenter graded these pre-practice problems, but the first two
 Brown & Kimble 615

were always marked as correct and the last one as incorrect to induce some
uncertainty in all participants. After the experimenter reiterated that they could
practice as much or as little as they liked, participants were left to practice. Time
was started when the experimenter left the cubicle and ceased when participants
indicated via the intercom that they had practiced enough. The number of prac-
tice problems completed out of 40 possible problems and the length of time spent
practicing were the primary dependent measures of self-handicapping in the
practice condition.
Participants in the tape condition were told that they would be doing the
second test while listening to tape recordings so we could study how the recorded
sound affects their performance. They were also told that they would be listening
to either a tape that would help performance or a tape that would hinder perfor-
mance. However, the experimenter then became flustered and embarrassed,
admitting that he or she had forgotten to assign that participant to a sound condi-
tion, so he or she was just going to let the participant pick any of the sound tapes.
Before participants made their choice, they were given the Current Feelings
survey, the state self-doubt, and the state performance concern scales to com-
plete, after which the experimenter presented participants a box containing four
tapes. Each tape case was labeled with one or two dots that were colored either
red or green. Participants were told that the green label tapes contained facilitat-
ing sound, with the two-dot green tape being very helpful and the one-dot green
tape being somewhat helpful. The red label tapes were said to have debilitating
sound, with the two-dot red tape having the worst effect on performance and the one-
dot red tape having a negative effect that was not as severe as the two-dot tape. After
the participants chose a tape (the dependent measure of self-handicapping), the exper-
imenter asked why they chose that tape and quickly wrote down participants’
responses.
After choosing a tape or signaling that they had finished practicing, partici-
pants were asked to complete a “Personal Reactions” form before taking the
actual test. The form asked practice participants to select a number from 1 (very
poorly) to 9 (very well) to complete the statement “I did __ on the first subtest.”
The instructions were to answer according to how they felt they did—not the
score they were given—and this was intended to be a manipulation check of
feedback and the feelings induced by the type of test (i.e., solvable vs. unsolv-
able). If they rated their performance between 1 and 5, they were instructed to
indicate their agreement using a scale of 1 (strongly disagree) to 5 (strongly
agree) for a series of statements such as, “I thought that I had done poorly on the
first subtest,” “I was surprised I had done so poorly on the first subtest,” and “I
did worse than I thought on the first subtest.” If they had rated their performance
between 6 and 9 on the first manipulation check question, they were asked to rate
similar statements that used “well” to describe their performance instead of
“poorly,” and “better” instead of “worse.” In addition, participants in the practice
condition were asked to rate how much effort they had put into practicing for the
616 The Journal of Social Psychology

second test, from 1 (no effort) to 7 (exceptional effort). This was intended as a
measure of self-reported handicapping. Participants were also asked whether—
according to the instructions—practice improves, decreases, or has no effect on
test performance. This served as a manipulation check that they remembered the
instructions given by the experimenter. After completing the survey, participants
were told they would not be taking the second test. They were asked to write a
sentence or two about their reaction to the study and what they thought the pur-
pose of the study was. Participants were thoroughly debriefed and asked if they
had any additional questions before being thanked and excused.

RESULTS

Manipulation Checks

Suspicion

We analyzed participants’ written reactions to the study for suspicions about

their score on the first subtest. Approximately 7% of participants (three out of 43)
who received failure feedback had suspicions that the feedback was false or that
we were actually interested in their tape choice or practice effort, as did 14% of
contingent success participants and 32% of non-contingent success participants.
These differences are consistent with how believable the success or failure feed-
back should have seemed to participants, considering the version of the test they
received. Failure participants received a difficult test because it was, in fact,
unsolvable, so few doubted the score they received. Conversely, non-contingent
success participants had the most suspicions about their score, which is under-
standable because they were told they had succeeded even though the test was
impossible. However, the analyses excluding suspicious participants were in the
same direction as the analyses with all participants, so only the analyses using all
participants are reported below because they provide more stable cell sizes.

Instructions About Practice

All but one participant in the practice condition chose the response, “Practice
improves test performance.”2 This confirms that participants remembered the
instructions about the importance of practice in obtaining an accurate estimate of
participants’ ability.

Reactions to Feedback

A one-way ANOVA revealed that feedback significantly affected partici-

pants’ ratings of how they felt they had done on the first subtest, F(2, 123) = 73.49,
p < .001. On a scale of 1 (very poorly) to 9 (very well), participants in the failure
 Brown & Kimble 617

feedback condition reported performing worse (M = 2.16, SD = 1.07) than partic-

ipants in the non-contingent success (M = 6.14, SD = 2.20; p < .001)3 and contin-
gent success (M = 6.71, SD = 2.17, p < .001) conditions, which were not
significantly different from each other. This suggests that the unsolvable nature
of the first test made failure participants believe they had done poorly, whereas
non-contingent success participants were influenced more by the verbal feedback
of success than by their memory of how difficult the test was.
What is important is that the responses of contingent success and non-
contingent success participants did not differ on any of the questions related to
how well they thought they did on the first test. For example, the success condi-
tions did not significantly differ on the items, “I was surprised I had done so well
on the first subtest” (contingent success, M = 4.23, SD = .88; non-contingent
success, M = 4.11, SD = .92; t(57) = –.51, ns), and “I did better than I thought on
the first subtest” (contingent success, M = 4.39, SD = .80; non-contingent
success, M = 4.18, SD = 1.09; t(57) = –.84, ns). This finding suggests that the
contingency manipulation was inadequate because only participants in the non-
contingent success condition who took the unsolvable test should have been
surprised by their success, yet the contingent success participants were equally
surprised. In fact, we scored the actual tests that were given to contingent suc-
cess participants and found that participants had performed quite poorly, averag-
ing only 4.78 (SD = 1.75) of 10 problems correctly. In other words, contingent
success participants—who were supposed to believe their success was
deserved—found the test difficult and were surprised by their results, as
reported above. This indicates that our manipulation did not make these partici-
pants truly experience contingent success. Consequently, participants in the
contingent success condition whose score was 50% or less (N = 28) were reas-
signed to the non-contingent success condition, because their success would
have felt surprising and undeserved, leaving only 13 participants who should
have believed their success was deserved (as intended by the manipulation).
Considering the small sample size and the number of variables we were inter-
ested in analyzing, we decided to exclude the contingent success condition from
all primary analyses.

Measures of Current Feelings After Receiving Feedback

We conducted one-way ANOVAs to examine if the type of feedback had

any effect on participants’ responses to the Current Feelings measures. On a scale
of 1 (not at all) to 7 (very much), failure participants’ response to the statement,
“I am confident I will perform well,” indicated they felt significantly less confi-
dent (M = 4.12, SD = 1.18) than non-contingent success participants (M = 5.00,
SD = 1.09), F(1, 111) = 16.44, p < .001. Failure participants were also signifi-
cantly more uncertain about how they would do, F(1, 111) = 4.91, p < .05, and
expected to do more poorly on the next test, F(1, 111) = 34.01, p < .001.
618 The Journal of Social Psychology

Furthermore, non-contingent success participants indicated feeling better about

themselves at the moment, F(1, 111) = 11.28, p < .01. Thus, it was clear that fail-
ure feedback led to more uncertainty about future performance and more nega-
tive feelings. However, feedback did not affect state performance concern or
state self-doubt (Fs < 1, ns).

Practice Condition: Practice Effort

To analyze the effect of feedback, participant sex, and experimenter sex on

practice effort, the amount of time participants spent practicing and the number
of problems they practiced were standardized and combined to form an index of
practice effort. A 2 (feedback: failure, non-contingent success) × 2 (participant
sex: male, female) × 2 (experimenter sex: male, female) ANOVA revealed a sig-
nificant three-way interaction, F(1, 46) = 5.32, p < .05. To interpret this interac-
tion, we conducted 2 (participant sex) × 2 (experimenter sex) ANOVAs
separately for each feedback condition (see Figure 1).
In the non-contingent success condition, the 2 × 2 ANOVA revealed a
significant main effect of experimenter sex, F(1, 29) = 7.19, p < .05. As seen in
Figure 1, participants practiced less when the experimenter was male (M = −.62,
SD = 2.23) than when the experimenter was female (M = 1.04, SD = 1.45).
Participant sex did not significantly affect practice, nor did it interact with exper-
imenter sex (Fs < 3, ns).
However, in the failure condition, the ANOVA revealed there was a signifi-
cant main effect of participant sex, (F(1, 17) = 5.07, p <. 05), such that men
practiced less (M = −.1.36, SD = 1.36) than women (M = −.23, SD = 1.8).

Non-contingent Success Feedback Failure Feedback

2 2

1 1
Practice Effort

Male experimenter
0 0 Female experimenter

–1 –1

–2 –2
Male Female Male Female
Participant Sex Participant Sex

FIGURE 1. Practice effort by participant sex, experimenter sex, and feedback.

 Brown & Kimble 619

However, this was qualified by a marginal interaction between participant sex

and experimenter sex, (F(1,17) = 3.32, p = .086). Simple effects analyses revealed
that women’s practice effort was not affected by experimenter sex, but men prac-
ticed significantly less when the experimenter was male (M = −1.84, SD = 1.49)
compared to when the experimenter was female (M = −1.04, SD = 1.31),
F(1,17) = 6.23, p < .05.

Practice Condition: Claimed Effort

To assess the effect of our independent variables on claimed self-handicapping,

we conducted a 2 (feedback) × 2 (participant sex) × 2 (experimenter sex)
ANOVA on participants’ responses to the question, “How much effort did you
put into practicing for the second test?” There was a significant effect of exper-
imenter sex, F (1, 46) = 4.90, p < .05, with participants claiming to have put
more effort into practicing when the experimenter was female (M = 5.08, SD = .84)
than male (M = 4.22, SD = 1.56). This was qualified by a marginal experi-
menter sex × feedback interaction, F (1, 46) = 3.85, p = .056. When non-
contingent success feedback was given by female experimenters, participants
claimed to have exerted greater effort (M = 5.32, SD = .78); less effort (more
self-handicapping) was reported when the same feedback was given by male
experimenters (M = 4.00, SD = 1.90), F(1,46) = 11.24, p < .01. There was no
difference in claimed effort between the two failure conditions, F < 1, ns. Thus,
participants claimed more effort when female experimenters were involved in
delivering non-contingent success feedback than when a male experimenter
delivered the same feedback.

Tape Condition: Tape Chosen

Tape choice was denoted with a lower number indicating greater self-
handicapping behavior, with two green dots coded as “4” and two red dots coded
as “1.” An average score below 2.5 (a score of 2 being the one red dot tape) indi-
cates a tendency towards selecting a debilitating tape and, therefore, the decision
to self-handicap.
Participants’ tape choice was significantly affected by their sex (F(1, 51) = 9.24,
p < .01) and the interaction between experimenter sex and feedback (F(1, 51) = 5.90,
p < .05). However, these effects were qualified by a marginal three-way interac-
tion between participant sex, experimenter sex, and feedback, F(1, 51) = 3.19,
p = .08. To interpret this interaction, we conducted separate 2 (participant sex) × 2
(experimenter sex) ANOVAs for the failure and non-contingent success feedback
conditions. The means for tape choice are presented in Figure 2.
In the non-contingent success condition, there were significant main effects
of participant sex (F(1, 33) = 5.67, p < .05) and experimenter sex (F(1, 33) = 10.01,
p < .01), but no interaction. Specifically, male participants (M = 2.05, SD = 1.19)
620 The Journal of Social Psychology

Non-contingent Success Feedback Failure Feedback

4 4
Tape Chosen

3 3
Male experimenter
Female experimenter

2 2

1 1
Male Female Male Female
Participant Sex Participant Sex

FIGURE 2. Tape chosen by participant sex, experimenter sex, and feedback.

chose a more debilitating tape than female participants (M = 2.82, SD = 1.19). In

addition, participants chose a more debilitating tape when the experimenter was
male (M = 3.17, SD = 1.03; female experimenter M = 2.04, SD = 1.17).
Similarly, there was a marginal effect of participant sex in the failure feed-
back condition, F(1, 18) = 3.81, p = .067, such that men (M = 2.45, SD = 1.29)
chose a more debilitating tape than women (M = 3.36, SD = 1.03). However, this
marginally interacted with experimenter sex, F(1, 18) = 3.54, p = .076. Men
chose a more debilitating tape when the experimenter was female (M = 2.17,
SD = 1.47; male experimenter M = 2.8, SD = 1.10) and women chose a more
debilitating tape when the experimenter was male (M = 2.83, SD = 1.17; female
experimenter M = 4.0, SD = 0.0). In other words, there was a marginal trend for
participants to self-handicap more (choose a more debilitating tape) in the pres-
ence of an experimenter of the opposite sex.

Relation Between Current Feelings and Self-Handicapping

Pearson bivariate correlations between the current feelings items and the
measures of self-handicapping (practice effort, claimed practice effort, and tape
chosen) were conducted to see which situational feelings measures were related
to self-handicapping measures. These correlations showed that greater confi-
dence about future performance was associated with choosing a debilitating tape,
r = −.264, p < .05. In addition, feeling it was important to do well on the test
(r = .349, p < .01) and wanting to do as well as possible (r = .358, p < .01) were
strongly related to claiming to have put more effort into practicing. However,
these items were not significantly related to practice effort.
 Brown & Kimble 621

As for the state measures of self-doubt and performance concern recorded

during the experiment, performance concern predicted both tape choice and prac-
tice effort, but self-doubt affected neither type of self-handicapping. Specifically,
performance concern was significantly related to practice effort (the combined
index; r = −.279, p < .05) and marginally related to tape chosen (r = −.228, p = .082),
such that participants self-handicapped more when they were experiencing
greater performance concern.
We also examined sex differences in current feelings. There were differ-
ences between men and women on the items relating to “confidence that I will
perform well,” F(1, 111) = 14.35, p < .001; “concern about how others might
regard my performance,” F(1, 111) = 6.28, p < .05; and “feeling good about
myself now,” F(1, 110) = 4.98, p < .05. Men expressed more confidence about
performance (men, M = 5.07, SD = 1.09; women, M = 4.26, SD = 1.17), while
women expressed more concern about others’ opinions (men, M = 2.55, SD = 1.56;
women, M = 3.32, SD = 1.67), and men felt better about themselves in the
experimental situation (men, M = 5.53, SD = 1.09; women, M = 5.02, SD = 1.32).
Similarly, there was a sex difference in current self-doubt, F(1, 111) = 11.43,
p < .01, with women reporting more current self-doubt (M = 29.07, SD = 6.01)
than men (M = 25.54, SD = 5.06).

DISCUSSION

We studied the opportunity to self-handicap by explicitly choosing a

hindering condition for performance or by exerting low effort. We had two
behavioral measures of self-handicapping, one involving effort-based self-
handicapping (practice effort) and another involving decisional self-handicapping
(tape choice). Claimed self-handicapping was measured through self-rated
effort.

Practice Effort

There was a consistent effect of sex when self-handicapping took the form of
reduced practice effort, such that men generally self-handicapped more than
women by practicing less. Interestingly, men self-handicapped somewhat more
after failure than after non-contingent success. Thus, consistent with Smith &
Kimble (2003), failure may produce more self-handicapping among men than
non-contingent success.
In terms of audience effects, failure participants self-handicapped less in the
presence of the opposite sex, whereas non-contingent success participants consis-
tently self-handicapped less in the presence of women. There may be two sepa-
rate factors driving these effects: First, non-contingent success is classically
believed to produce self-handicapping to maintain an internal attribution for past
success and to prevent an internal attribution for future failure. Exerting less
622 The Journal of Social Psychology

effort following non-contingent success is clearly self-handicapping, whereas

exerting less effort following failure feedback can be motivated by both desire to
prevent attributions about poor ability as well as hopelessness (“I did terribly on
the first task, and no amount of effort is going to make me do better on the
second”; it is possible that these two motives in conjunction led to even greater
self-handicapping among men in general). Given that women are less accepting
of self-handicapping than men (Hirt et al. 2003), it is likely that non-contingent
success participants (whose reduced effort clearly indicated self-handicapping)
reduced self-handicapping in the presence of women who would look down upon
such behavior. In contrast, increased effort following failure feedback indicates
an honest attempt to improve. Participants who received failure feedback prac-
ticed more in the presence of the opposite sex, perhaps suggesting that they
wanted to look more appealing to the opposite sex by showing how determinedly
they would try to improve. In other words, it might be more acceptable to “give
up” in front of the same sex, because such a reaction would not make a positive
impression on the opposite sex. Thus, audience sex affected effort following non-
contingent success such that participants who were clearly self-handicapping
tried to temper this behavior in front of someone who would be disapproving
(female experimenter), whereas audience sex influenced effort following failure
by leading participants to try harder in front of the opposite sex.

Tape Choice

Similar behavior was evident when self-handicapping was decisional in nature.

Specifically, participants who received non-contingent success chose a more debili-
tating tape when the experimenter was male, mirroring the results of the practice
condition. Reduced effort following non-contingent success is unambiguously self-
handicapping, and so participants were less eager to obviously self-handicap in front
of a female experimenter who was likely to disapprove of such behavior.
In contrast, the behavior of participants in the failure condition contradicts the
behavior of failure participants for whom practice effort was the dependent mea-
sure. When self-handicapping took the form of reduced practice effort,
participants who had failed “tried harder” in front of the opposite sex. In contrast,
when self-handicapping took the form of choosing a performance-debilitating
tape, participants chose a more debilitating tape in the presence of the opposite
sex. One possible explanation concerns participant expectations about the source
of improvement. When practice effort was the available behavior, failure partici-
pants expected that improvement would be through their own effort and the oppo-
site-sex experimenter would know this. However, when a tape was available,
failure participants might have expected success following a performance-enhancing
tape to be undeserved. In other words, this wouldn’t lead to a positive impression
of oneself, because it was the tape that improved performance, not the partici-
pant’s own ability. Given that a positive impression of oneself was unlikely, for
 Brown & Kimble 623

these participants, preventing additional negative attributions about their ability

was the best alternative. Thus, failure feedback led to being concerned about the
impression one was making in front of the opposite sex. When improvement could
be attributed to oneself, participants attempted to project a positive impression by
exerting more effort. In contrast, when improvement could be attributed to exter-
nal help (the performance-enhancing tape), which might actually make the indi-
vidual look worse (i.e., can only improve with the help of something else),
participants attempted to project a less negative impression by self-handicapping,
thus preventing further ability-based attributions for failure.
Failure feedback, it seems, increases impression management concerns,
especially in regards to the opposite sex. These concerns may revolve around
projecting a generally positive impression of oneself; failure “looks bad,” and so
participants are motivated to find ways to “look good” (especially in front of the
opposite sex, for whom a positive impression is particularly important). Non-
contingent success also increases impression management concerns, although
these concerns are specifically focused on ability. The unexpected nature of the
success captures participants’ attention and concern, so the desired impression is
to “continue deserving past success” and “avoid appearing to have low ability,”
rather than a general focus on positive or negative impressions. Audience effects
reflect the specific nature of impression concerns: Participants self-handicap to
protect ability attributions, and they only temper this behavior in front of females
who—if realizing their behavior indicates self-handicapping—would disapprove
of such behavior, thus making it a less effective way to attain a positive impres-
sion regarding their ability. This is supported by the results of the claimed self-
handicapping measure: Non-contingent success participants claimed more effort
when the experimenter was female, again suggesting they realized she would not
approve of self-handicapping. Admittedly, these interpretations are speculative,
and additional research is necessary to isolate participants’ specific concerns,
expectations, and motives.
It’s important to note that performance concern was related to more self-
handicapping for both effort and choice behaviors. This outcome was unex-
pected, considering Oleson et al.’s (2000) finding that trait self-handicapping was
associated with low performance concern. It may be that momentary (state)
variations in performance concern affect behavior differently than chronic (trait)
levels of performance concern. According to Oleson et al. (2000), low trait
performance concern should have led to frivolous self-handicapping; but instead,
high state performance concern did so in the current study. However, on closer
inspection of performance concern items, such as “I strive to be successful at all
times,” “It is important that I succeed in all that I do,” and “Failure is unaccept-
able to me,” it seems apparent that high performance concern is a mental state
that would make a person act protectively to avoid a negative evaluation by self-
handicapping. This performance concern seems to be the state of mind that
Berglas and Jones (1978) tried to create in the first experiment by manipulating
624 The Journal of Social Psychology

non-contingent success, and it may be that these momentary feelings have a

unique effect on self-handicapping that is independent of chronic concerns about
performance.
In general, failure feedback leads males to self-handicap the most. Whether
females self-handicap may depend on their psychological reactions to the situa-
tion, such as worrying about their actual performance and how others will evalu-
ate them. We decided to further examine females’ ratings of current feelings,
self-doubt, and performance concern, because their self-handicapping reactions
did not show any consistent patterns in response to feedback. While there were
no significant correlations between emotional reactions and behavior for males,
there were many relationships between emotional reactions and behavior for
females. Among females, reporting more current self-doubt was associated with
selecting a self-handicapping tape (r = −.447, p < .05). Moreover, setting higher
standards (r = .368, p = .05), greater importance of doing well (r = .474, p < .01),
and expecting to do poorly (r = .381, p < .041) were also related to claiming
more effort. These findings suggest that females’ inconsistent self-handicapping
reactions are influenced more so by particular emotional states than by feedback.
In other words, feedback does not affect females’ self-handicapping directly, but
their individual reactions to the feedback may have a more immediate effect on
self-handicapping.

Conclusions

One pivotal feature of behavioral self-handicapping discovered in this

study was how there are multiple ways to create a disadvantage before an
important evaluation. For example, was it done by verbally choosing a
performance-detracting condition for an upcoming evaluation or by exerting less
effort in preparation for an evaluation? Decisional or choice self-handicapping was
a single behavior and was stated to an experimenter; practice effort self-handicapping
involved several implicit decisions made while completing practice prob-
lems. A previous decisional self-handicapping study similar to our choice
condition showed that more men chose a performance-debilitating tape than a
performance-facilitating tape (Kimble & Bryant, 2002), as in the present study.
Women observers had the effect of decreasing self-handicapping in practice
situations but increasing self-handicapping in tape choice situations. While mak-
ing a risky verbal choice to an experimenter may show an adventurous nature,
actually exerting less effort may not seem so positive when the observer is likely
to disapprove of the behavior (Hirt et al., 2003). Women observers’ likely disap-
proval of self-handicapping appears to have inhibited low practice effort.
According to Oleson et al. (2000), low performance concern should have led
to frivolous self-handicapping; instead, high performance concern did so in the
current study. However, on inspection of performance concern items, such as “I
strive to be successful at all times,” “It is important that I succeed in all that I do,”
 Brown & Kimble 625

and “Failure is unacceptable to me,” it seems apparent that high performance

concern is a mental state that would make a person act protectively to avoid a
negative evaluation by self-handicapping. This performance concern seems to be
the state of mind that Berglas and Jones (1978) tried to create in the first experi-
ment by manipulating non-contingent success.
Finally, while men self-handicapped when they received failure feedback,
women seemed to be affected more by their emotional reactions to the evalua-
tion situation than they were affected by failure or non-contingent success.
Self-handicapping by women was more associated with self-doubt, worrying
about how others reacted to their performance, expecting to do poorly on the
test, and not feeling confident in the evaluation context. In conclusion, self-
handicapping is a dynamic process that is affected by a confluence of personal
factors (such as sex and performance concern) and situational factors (like
feedback and audience sex).

NOTES
1. All experimenters were trained to behave similarly and to follow a common script.
In addition, experimenters were average in attractiveness (e.g., neither supermodels nor
disfigured).
2. Removing this participant does not change the analyses.
3. All post hoc analyses use the Bonferroni procedure.

AUTHOR NOTES
Christina M. Brown is an assistant professor in the Department of Psychology
at Saint Louis University. Charles E. Kimble passed away in March 2009 after a
short illness. Dr. Kimble was a professor in the Department of Psychology at the
University of Dayton.

REFERENCES
Berglas, S., & Jones, E. E. (1978). Drug choice as a self-handicapping strategy in
response to noncontingent success. Journal of Personality and Social Psychology, 36,
405–417.
Greenberg, J. (1985). Unattainable goal choice as a self-handicapping strategy. Journal of
Applied Social Psychology, 15, 140–152.
Harris, R. N., & Snyder, C. R. (1986). The role of uncertain self-esteem in self-handicapping.
Journal of Personality and Social Psychology, 51, 451–458.
Higgins, R.L. & Harris, R. N. (1988). Strategic “alcohol” use: Drinking to self-handicap.
Journal of Social and Clinical Psychology, 6, 191–202.
Hirt, E. R., Deppe, R. K., & Gordon, L. J. (1991). Self-reported versus behavioral self-
handicapping: Empirical evidence for a theoretical distinction. Journal of Personality
and Social Psychology, 61, 981–991.
Hirt, E. R., McCrea, S. M., & Boris, H. I. (2003). “I know you self-handicapped last
exam”: Sex differences in reactions to self-handicapping. Journal of Personality and
Social Psychology, 84(1), 177–193.
626 The Journal of Social Psychology

Hirt, E. R., McCrea, S. M., & Kimble, C. E. (2000). Public self-focus and sex differences
in behavioral self-handicapping: Does increasing self-threat still make it “just a man’s
game?” Personality and Social Psychology Bulletin, 26, 1131–1141.
Jones, E. E., & Rhodewalt, F. (1982). The self-handicapping scale. (Available from
F. Rhodewalt, Department of Psychology, University of Utah, Salt Lake City, UT 84112.)
Kimble, C. E. & Bryant, J. (2002). Sex and behavioral self-handicapping: Competi-
tive men make self-handicapping choices. Unpublished manuscript, University of
Dayton.
Kimble, C. E., & Hirt, E. R. (2005). Self-focus, sex, and habitual self-handicapping: Do
they make a difference in behavioral self-handicapping. Social Behavior and Personal-
ity, 33, 43–56.
Kolditz, T. A., & Arkin, R. M. (1982). An impression management interpretation of the
self-handicapping strategy. Journal of Personality and Social Psychology, 43, 492–502.
Leary, M. R., & Shepperd, J. A. (1986). Behavioral self-handicaps versus self-reported
handicaps: A conceptual note. Journal of Personality and Social Psychology, 51,
1265–1268.
Luginbuhl, J., & Palmer, R. (1991). Impression management aspects of self-handicapping:
Positive and negative. Personality and Social Psychology Bulletin, 17, 655–662.
Oleson, K. C., Poehlmann, K. M., Yost, J. H., Lynch, M. E., & Arkin, R. M. (2000). Sub-
jective overachievement: Individual differences in self-doubt and concern with perfor-
mance. Journal of Personality, 68, 491–523.
Rhodewalt, F. & Davison, J. Jr. (1986). Self-handicapping and subsequent performance:
Role of outcome valence and attributional certainty. Basic and Applied Social Psychol-
ogy, 7, 307–322.
Rhodewalt, F., Morf, C., Hazlett, S., & Fairfield, M. (1991). Self-handicapping: The role
of discounting and augmentation in the preservation of self-esteem. Journal of Person-
ality and Social Psychology, 61, 122–131.
Rhodewalt, F., Saltzman, A. T., & Wittmer, J. (1984). Self-handicapping among competi-
tive athletes: The role of practice in self-esteem protection. Basic and Applied Social
Psychology, 5, 197–210.
Shepperd, J. A., & Arkin, R. M. (1989). Self-handicapping: The moderating roles of pub-
lic self-consciousness and task importance. Personality and Social Psychology Bulletin,
15, 252–265.
Shepperd, J. A. & Arkin, R. M. (1991). Behavioral other-enhancement: Strategically
obscuring the link between performance and evaluation. Journal of Personality and
Social Psychology, 60, 79–88.
Smith, M. R. & Kimble, C. E. (2003). Behavioral self-handicapping as a function of
success and failure feedback, sex, and personality variables. Unpublished manuscript.
University of Dayton.
Zuckerman, M., Kieffer, S. C., & Knee, C. R. (1998). Consequences of self-handicapping:
Effects on coping, academic performance, and adjustment. Journal of Personality and
Social Psychology, 74, 1619–1628.

Received May 16, 2007

Accepted April 5, 2008
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.