Social Comparison and Performance 1
This is a preprint of an article accepted for publication in the
European Journal of Social Psychology copyright ? 2006
RUNNING HEAD: Domain Matching, Social Comparison and Performance
When Different is Better:
Performance Following Upward Comparison
Camille S. Johnson
Diederik A. Stapel
University of Groningen
This publication was supported in part by grant number T32 MH19728 from the National Institute of Mental Health, and National Research Service Award F31 MH64238-01 awarded to the first author. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Institute of Mental Health.
This publication was also supported in part by a grant from the Dutch Science Foundation (Nederlandse Organisatie voor Wetenschappelijk Onderzoek) and a research grant of the Heymans Institute of the University of Groningen awarded to the second author.
Social Comparison and Performance 2
The attainability of upward social comparisons is known to affect self-evaluative responses. The consequences for performance, however, are less well understood. We suggest
that demoralizing upward comparisons with unattainable targets may lead to improved performance when the target and performance domains are mismatched. For example, comparison with a target that has been successful in an analytic domain should lead to better performance in a verbal domain. This improvement in performance occurs because increased performance in alternative domains provides an opportunity for self-evaluation maintenance. In three studies, we demonstrate that upward comparisons to targets whose successes are perceived as threatening lead to improved performance when the task and performance domain do not match, but no improvements when the domains match.
Social Comparison and Performance 3
When Different is Better:
Performance Following Upward Comparison
In judging ourselves, we look to the accomplishments of others. Their achievements serve as comparison points by which we measure our own progress (Festinger, 1954). In addition to providing information, comparison with another individual may affect our feelings, thoughts, and behaviors. Social comparison researchers have effectively documented how others affect our self-evaluations. Researchers have focused on the direction of comparison (Wood, 1989), the type of comparison (Stapel & Suls, 2004), the role of individual differences (Marx, Stapel, & Muller, 2005), the characteristics of the comparison target (Lockwood & Kunda, 1997) and the conditions under which comparison occurs (Mussweiler, Ruter, & Epstude, 2004). Additionally, researchers interested in performance have shed light on the relation between social comparisons and performance (Blanton, Buunk, Gibbons, & Kuyper, 1999; Marx & Roman, 2002; Seta, 1982; Seta, Seta, & Donaldson, 1991). However, the underlying relation among social comparison, self-evaluations and performance is less well explored. In this paper, we investigate how self-evaluative responses to upward comparison targets lead to strategic performance improvements.
We begin with a distinction first offered by Lockwood and Kunda (1997): the perceived attainability of comparison targets. Lockwood and Kunda exposed participants to targets who were either younger or older than the participants. Older comparison targets represented attainable comparison, because participants could foresee similar accomplishments, whereas younger comparison targets were considered unattainable. They found that attainable comparison targets inspired participants and raised self-evaluations, whereas unattainable targets demoralized participants and lowered self-evaluations. We extend these findings to performance outcomes and make what, at first, may appear to be an ironic prediction. We suggest that under some conditions, the same comparison targets that
Social Comparison and Performance 4
lead to lowered self-evaluations may lead to improved performance and that the targets that boost self-evaluations may fail to impact performance at all.
This prediction is based on the reasoning that although individuals may not have an opportunity to surpass the accomplishments of an unattainable target in one domain, the possibility of surpassing the target in other domains exists. Moreover, the unattainable
comparison target threatens self-evaluation and that threat must be resolved (Tesser, 1988). One means of responding to this threat is to increase performance in alternative or mismatching domains. Thus, we predict unattainable comparison targets will lead to improved performance, but only in domains that mismatch the target‟s domain. However,
attainable targets are not threatening to self-evaluations. In fact, they may even be inspiring and ego-boosting (Lockwood & Kunda, 1997). Thus, following comparison with attainable targets, there is no threat to be resolved, and the domain of performance should not be relevant: no performance improvement is expected, regardless of the domain.
Several social psychological theories support our hypothesis. According to the self-evaluation maintenance (SEM) model, being outperformed by another may threaten self-evaluations (Tesser, 1988). For example, if a first-year chemistry student compares herself with an outstanding fellow first-year chemistry student, her self-evaluations may be threatened. In the face of this threat, the SEM suggests that she may use one of several strategies to maintain her self-evaluations. One means of dealing with this threat to self-evaluations would be to improve her chemistry performance. However, having been outperformed once in the chemistry domain, rewarding future comparisons in that domain may not be likely and further comparisons in the same domain may only confirm incompetence. On the other hand, comparison with the target in an alternative domain, such as artistic expression, may present an opportunity for superior performance and restoration of self-evaluations. Withdrawing from domains in which further threats to self-evaluations may
Social Comparison and Performance 5
be anticipated and seeking alternative domains in which comparisons may be positive allows one to maintain positive self-regard (Crocker & Major, 1989; Steele, 1997). If our chemistry student leaves the chemistry domain, she is less likely to face more unflattering upward comparisons and more likely to encounter self-enhancing downward comparisons. Applying this logic to the experimental situation, we expect that our participants, when presented with a task from a domain other than the target‟s domain of success, will show improved
performance. However, when the performance domain matches the target‟s domain of success, no change in performance is expected.
As we move through daily life we encounter successful others from many different domains, which may match, mismatch, or encompass our own areas of interest. For example, our chemistry student is as likely to encounter a successful chemistry student as she is to encounter someone who is globally or generally successful. The globally successful student may constitute an ambiguous match – she neither matches nor mismatches our chemistry
student‟s domain. In fact, the domain of the globally successful student may be described as encompassing the chemistry domain. What are the consequences of comparison with such a global target? We suggest that in this situation, our chemistry student can adopt “socially
creative” strategies in which she could carve out a specific domain in which she could be successful. This suggest that when a target‟s success is not restricted to a particular domain
but is described as global or ambiguously matching, individuals may view any specific domain as an opportunity to improve, and performance in that specific domain will improve following comparison. That is, after seeing a globally successful student, individuals are expected to perform better on verbal, analytical, or other specifically described task.
These current studies seek to demonstrate that individuals respond to unattainable comparison targets as they would to other self-evaluation threats. In particular, we predict
Social Comparison and Performance 6
that following exposure to an unattainable comparison target, individuals will increase performance in domains that mismatch the domain of the comparison target. By contrast, attainable comparison targets are expected to boost immediate self-evaluations, but not affect performance.
Three studies are presented here. Building on a foundation provided by previous research (Lockwood & Kunda, 1997), the current studies use manipulations of attainability to create threats to self-evaluations. However, in addition to examining how attainability impact self-evaluations, the current studies also examine how attainable and unattainable comparison targets impact performance. Thus, rather than extrapolating from self-evaluative responses to predict behavior, the current studies provide evidence of the actual performance consequences. In all studies, participants read about comparison targets who were either the same age (unattainable) or older (attainable), then completed measures of performance or self-evaluation. Each study provides support for a different aspect of our model.
The first study examined the effects of strict matching (i.e., target excels in literature and verbal ability is tested) and mismatching (i.e., target excels in mathematics and verbal ability is tested) of domains on performance. The second study directly examines the relationship among social comparisons, test domains, performance expectations and performance. In addition to measures of performance, Study 2 includes measures of performance expectations and tests the hypothesis that alternative domains lead to better performances when individuals have been exposed to unattainable comparison targets, because alternative domains allow individuals to have higher performance expectancies. The third study examines the generalizability of these effects by examining the effects of matching (i.e., target excels in literature, verbal ability is tested) versus overlapping of domains (i.e., target is globally successful, verbal ability is tested) on performance. Study 3 tests the hypothesis that individuals who experience threat will respond to an encompassing
Social Comparison and Performance 7
mismatching condition as mismatching, and among those exposed to an unattainable comparison target, will experience an improvement in performance.
Study 1 varies the degree to which the target and test domain match. We suggest that better performance is the result of self-evaluation maintenance behaviors and will only occur when both threats to self-evaluations and opportunities to recover self-evaluations are present. Therefore, better performance is predicted when the participants are exposed to a threatening, unattainable comparison target and provided with an opportunity to perform in an alternative domain.
In addition to performance, participant perceptions of the test are also measured. When participants are doubtful or concerned about their ability level in the test domain, they are expected to view the performance task as more difficult. Thus, we predict that participants who have been exposed to the unattainable comparison target and are faced with a task in a matching domain will anticipate that test as being difficult. After taking the test, those who used the testing as an opportunity to repair self-regard (unattainable/mismatch condition) should also feel greater satisfaction with their performance.
Although previous published research has found that older comparison targets are regarded as attainable and inspiring and that younger comparison targets are regarded as unattainable and discouraging (Lockwood & Kunda, 1997), given the importance of that construct in our theory, it was necessary to verify that the comparison targets used in our studies similarly affected feelings of inspiration and threat. Moreover, given our hypotheses regarding domain-specific influences of older and younger comparison targets, it was necessary to demonstrate that disparate effects of older and younger comparison targets on
Social Comparison and Performance 8
performance expectations. Therefore, a pilot study of our experimental materials was conducted.
Forty-two Dutch university students read about a successful fellow student. All materials were presented in Dutch, and translations are presented here. The paragraph outlined the process by which Hans de Groot won a (fictional) prestigious prize. It described Hans as “one of five finalists chosen ….to participate in three days of „intellectual
challenges”. The paragraph also explained that Hans won the prize because he showed
“remarkable intellectual ability during the completion of a variety of tasks” and “especially
because of his verbal ability”. The paragraph also contained quotes from judges of the
competition praising Hans‟ exceptional talent for seeing “linguistic, verbally well thought out
solutions in everyday problems." Finally, the paragraph gave information about the size of the prize (4000 euros). In the older comparison target conditions, the paragraph ended by reiterating the fact that Hans competed against other advanced students, and in the younger target conditions, the paragraph reminded participants that Hans competed against beginning university students.
After reading about the comparison target, participants were asked to rate the target along a series of traits, including likeability, intelligence, success, creativeness, outgoing, and laziness. Age of the comparison targets did not affect any of these ratings, t < 1.0. Successful
younger targets were not not rated as more successful or more liked than older targets.
Participants also rated how similar the comparison targets were to themselves, how attainable they were, how threatening, and inspiring. Again, the expected effects were found. Participants found the older targets to be more attainable (M= 4.38, t(41) = 2.43, p =.02), 1
inspiring (M= 3.33, t(41) = 2.55, p =.01), and similar (M= 4.90, t(41) = 2.12, p =.01) than 2 3
the younger targets (M = 3.59, M= 2.45, M= 4.05). Additionally, older targets were 12 3
Social Comparison and Performance 9
viewed as less threatening (M =5.77) than the younger comparison targets, (M = 5.00), t(41)
= -2.97, p =.005.
Participants were also asked two questions regarding their expectations of future success: to what extent they expected to be successful in their own way and to what extent they expected that they could be successful in the same way as the target. As predicted, participant ratings of the likelihood of success in their own way did not differ as a result of exposure to an younger role model (M = 4.68) or to an older target (M = 4.62), t < 1.0.
However, when participant ratings of likelihood of success in the target‟s domain was affected by the age of the target. When reading about a younger comparison target, ratings of success in the target‟s domain were statistically significantlyly lower (M = 4.05) than when
reading about an older comparison target (M = 4.86), t(41) = 2.42, p = 02.
The results of the pilot study suggested that our manipulation of threat – comparison
with older or younger comparison targets, is appropriate. Moreover, they provide evidence that people feel that it is less likely that they will succeed in the same domain as a successful other, perhaps leading to a shift in domains.
Participants were 98 Dutch university students and all materials were presented in Dutch. All students received partial credit towards a course requirement. Although gender information was not collected, participants were representative of the overall participant pool, which consisted of 70% females. All participants were undergraduate students in their first two years of school.
A 2 (type of comparison target: attainable versus unattainable) x 2(type of domain: domain match versus domain mismatch) between-participants design was employed.
Social Comparison and Performance 10
Participants viewed the comparison targets then completed the performance task.
Comparison Target. The comparison targets were the same as used in the pilot study except that the descriptions varied in the abilities to which the target‟s success was attributed.
In the matching conditions, the target‟s success was ascribed to his verbal abilities (literature domain) and in the mismatching conditions to his analytical abilities (mathematics domain). In mismatching conditions, where participants read about the target who was successful because of his mathematics ability, participants were told that Hans had been successful “especially because of his analytical ability”. The paragraph also contained quotes from judges of the competition praising Hans‟ exceptional talent for seeing “logical, analytically well thought out solutions in everyday problems."
Performance measure. The test was described to participants as a measure of verbal ability. The test consisted of 20 remote associate task items (RAT, McFarlin & Blascovich, 1984). Typical RAT items present three related words (e.g., television, window, and computer)
and ask participants to fill in a fourth related word (e.g., screen).
Participants were brought to the lab to participate in several unrelated studies. They were asked to read about the comparison target as part of a study of media influence. After reading about the comparison target, they were asked to take a moment to reflect on the target.
After reading about the comparison target, participants received test instructions. In all conditions the performance task was described as a measure of verbal ability. Participants were told that they should work on the test until they either answered all of the questions or felt that they could not answer any more questions correctly.
After seeing the test instructions, participants rated how difficult the test appeared, then completed the RAT. After the test, participants were asked how satisfied they were with