The last five are the articles that have to be summarized and the first one is the full instructions on the paper with exact details on how it is to be done.

Journal of Experimental Social Psychology 43 (2007) 233–240 www.elsevier.com/locate/jesp

0022-1031/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jesp.2006.01.010

The perseverance eVect in the debrieWng paradigm: Replication and extension !

Cathy McFarland a,¤, Adeline Cheam a, Roger Buehler b

a Department of Psychology, Simon Fraser University, Burnaby, BC, Canada V5A 1S6 b Wilfrid Laurier University, Waterloo, Ont., Canada N2L 3C5

Received 9 November 2005; revised 11 November 2005 Available online 29 March 2006

Abstract

A classic study conducted by Ross, Lepper, and Hubbard (1975) revealed a perseverance eVect wherein people who received positive performance feedback on an alleged social perceptiveness test reported more favorable self-perceptions in this domain than those who received negative feedback despite the fact that they had received standard outcome debrieWng (i.e., been informed about the false, prede- termined, and random nature of the feedback) prior to reporting self-assessments. The present studies extend this past research by reveal- ing that (a) there is a form of outcome debrieWng (i.e., informing participants about the bogus nature of the test as well as the bogus nature of the feedback) that eVectively eliminates the perseverance eVect, (b) the perseverance eVect that occurs after standard outcome debrieWng is limited to perceptions of speciWc task-relevant skills rather than more global abilities, and (c) aVective reactions do not underlie the perseverance eVect that occurs in the false feedback paradigm. © 2006 Elsevier Inc. All rights reserved.

Keywords: The perseverance eVect; DebrieWng; Self-perception; False feedback

In a study investigating how threats to self-esteem aVect social perceptions, Charlotte received feedback indicating that she performed at the 30th percentile on a test that pur- portedly measured intellectual ability. Shortly thereafter, she evaluated a target person, and was then thoroughly debriefed. The experimenter explained why it was necessary to present a cover story, stressed that her score was randomly determined in advance, and highlighted that her score reXected nothing about her true intellectual ability. Nonethe- less, as she headed home, Charlotte found herself wondering about her ability to succeed in university, and seriously con- templated whether she really has what it takes to be a lawyer. This perseverance eVect, wherein people cling to newly

formed beliefs even when the evidential basis for those beliefs is completed refuted, was demonstrated convincingly in a classic study conducted by Ross et al. (1975). Their research revealed that participants who were provided with false feed- back indicating that they performed well on a test of social perceptiveness ability provided more favorable self-evalua- tions after debrieWng than did participants who were told that they performed poorly. In other words, the self-perceptions of task-relevant skills that were elicited by the feedback “per- severed” despite extensive debrieWng. Presumably, such beliefs persist because subjects attempt to generate explana- tions for the initial outcome (e.g., I am a pretty outgoing per- son and that is why I did so well), and the causal factors they identify continue to predict the outcome even when the out- come that prompted the explanation is invalidated (Ander- son, Lepper, & Ross, 1980; Ross, Lepper, Strack, & Steinmetz, 1977).

Although conducted over a quarter century ago, Ross et al.’s research continues to have important ethical implications for researchers in social psychology. The false

! This research was supported by SSHRC standard research grants awarded to the Wrst and third authors. We thank Celeste Alvaro, Paul Conway, Vivian Hsing, and Diane Lines for their assistance with the research.

* Corresponding author. E-mail address: [email protected] (C. McFarland).

234 C. McFarland et al. / Journal of Experimental Social Psychology 43 (2007) 233–240

feedback paradigm has been used to study many social psy- chological phenomena, including reactions to social compar- isons (e.g., Crocker, Thompson, McGraw, & Ingerman, 1987; McFarland & Buehler, 1995; Tesser, Millar, & Moore, 1988); the impact of threats to self-esteem on self-evaluation (e.g., Brown & Smart, 1991; Dunning, Leuenberger, & Sherman, 1995), social comparison judgments (e.g., Brown & Galla- gher, 1992), aggression (Stucke & Sporer, 2002), and preju- dice (e.g., Fein & Spencer, 1997); the role of individual diVerences in emotional, cognitive, and behavioral reactions to performance feedback (e.g., Baumeister, Heatherton, & Tice, 1993; Blaine & Crocker, 1993; Di Paula & Campbell, 2002; McFarlin, 1985; Wood, Giordano-Beech, Taylor, Michela, & Gaus, 1994, 1999); the relation between mood and cognition (e.g., Forgas, 2000; Schwarz & Clore, 1996); and the nature of emotions and emotional regulation in the aftermath of distressing events (Brown & Dutton, 1995; Brown & Marshall, 2001; Dodgson & Wood, 1998; Forgas & Ciarrochi, 2002; McFarland & Buehler, 1998; Nummenmaa & Niemi, 2004). Given the continued and widespread use of the false feedback paradigm, it is essential that researchers have at their disposal debrieWng techniques that can eVec- tively eliminate the perseverance phenomenon. The goal of our research was to critically examine the debrieWng process and further clarify the precise features that make for an eVec- tive debrieWng in this paradigm.

The current research

We conducted two studies that were designed to extend Ross et al.’s (1975) Wndings regarding the nature of an eVec- tive debrieWng. Before presenting our precise goals, it is important to consider their research in greater detail. In their main study, participants were Wrst informed that the research was examining “physiological responses during decision making.” Next, while attached to a recording device, they completed a decision-making task (i.e., distinguishing real from fake suicide notes) that purportedly assessed social per- ceptiveness ability.1 After receiving either success or failure feedback, they were assigned to one of three groups. In the outcome debrieWng condition, participants were informed of the “true” purpose of the study (“to examine physiological reactions to feedback”), and that their feedback was false, randomly assigned, predetermined, and non-reXective of their actual ability. In the process debrieWng condition, partic- ipants received the same information as that provided to out- come debrieWng participants, with one important addition: they were informed about the perseverance phenomenon and encouraged to avoid engaging in this cognitive process. In the no debrieWng condition, participants did not receive a debrieWng at this point in the session. Participants then com- pleted three assessments of belief perseverance: (1) estimates

of current performance and predictions regarding future per- formance on an equally diYcult set of notes, (2) evaluations of ability on the speciWc task (i.e., identifying real suicide notes), and (3) ratings of abilities presumably related to social perceptiveness (i.e., recognizing falsehood, sensitivity to others’ feelings, and test-taking skills). A perseverance eVect would be revealed if post-debrieWng self-perceptions in the success group were more favorable than those in the fail- ure group. The results indicated that although process debrieWng was eVective in eliminating perseverance on all measures, outcome debrieWng yielded a perseverance eVect on both performance estimates and ratings of ability on the speciWc task.2

Our research had four objectives. First, we assessed the possibility that there is a form of outcome debrieWng that can be eVective in eliminating perseverance. One notewor- thy feature of the original Ross et al. (1975) research is that although debriefed participants were told that their score on the test was bogus, they were led to believe that the test itself was a valid test of an important underlying ability, and that therefore, a “real” score on the test actually existed. It seems possible that this feature of the debrieWng might make the perseverance eVect more likely. Presum- ably, participants who have received outcome debrieWng might Wnd themselves wondering what their real score on the test is, and might even use the fake score as an anchor with which to estimate their real score (Cervone & Palmer, 1990; Mussweiler, Strack, & PfeiVer, 2000; Wegner, Coul- ton, & WenzlaV, 1985).3 In essence, curiosity about one’s real score could engender a train of thought that leads par- ticipants to construct a scenario or image of their actual performance that is consistent with their randomly assigned performance. This, image, in turn, might engender further self-relevant thoughts (e.g., attributions to stable qualities, self-praise or criticism) that ultimately create per- severance in self-perceptions. In our research, we developed a revised form of outcome debrieWng that included infor- mation indicating that the test itself was not a real validated test of social perceptiveness ability. We expected that the addition of this feature would eliminate the perseverance eVect that is normally obtained with outcome debrieWng.

Second, we explored the generality of the perseverance eVect that occurs after standard outcome debrieWng. In Ross et al. (1975), perseverance was strongest on self-perceptions that were speciWc to the test-domain (i.e., performance esti- mates and ratings of the ability to identify real suicide notes). However, although they assessed perceptions of a few “related

1 Dr. Lepper (personal communication, August, 1997) has conWrmed that participants were told that the decision-making task assessed general social perceptiveness ability. The original (1975) report did not mention this point explicitly.

2 We focused our discussion on Study 2 of the Ross et al. (1975) report because it included a “no debrieWng” control group and a process de- brieWng group. The results of their Study 1, which included only the out- come debrieWng condition, were comparable to those of Study 2, except that perseverance was obtained on the “related abilities” measure.

3 Some anecdotal support for this assumption can be found in the spon- taneous comments made by participants from other studies in our lab. We regularly use false feedback, and have found that it is not uncommon for participants to ask for information about their “real” score after being de- briefed. Indeed, it is these comments that sparked the current research.

C. McFarland et al. / Journal of Experimental Social Psychology 43 (2007) 233–240 235

skills,” they did not ask participants to rate themselves on the precise global dimension allegedly assessed by the suicide note task—social perceptiveness ability. In our research, we included several items to assess participants’ perceptions of their global social perceptiveness skills. Based on Ross et al.’s Wndings, we expected that perseverance would likely be lim- ited to speciWc ratings. However, it seemed worthwhile to assess the generality of the perseverance eVect because their preliminary study (see Footnote 2) did reveal an eVect on the “related skills” measure. Moreover, given that the test is por- trayed as assessing a general ability, it seems plausible that perseverance could occur on more global perceptions. The generality of the perseverance eVect is an important issue. If the phenomenon extends to perceptions of global abilities then the prescription that researchers conduct debrieWngs that completely eliminate perseverance is rendered ever more pressing.

Finally, we examined the possibility that the persever- ance eVect occurring after standard outcome debrieWng might be due to “aVective perseverance” (e.g., Sherman & Kim, 2002). Ross et al. discussed, but were not able to test, the possibility that perseverance in self-perceptions could derive from mood reactions that are not completely elimi- nated through debrieWng. Presumably, participants’ post- debrieWng moods could inXuence self-perceptions via mood-congruent processing (e.g., Brown & Mankowski, 1993; Sedikides & Green, 2001). We included a post- debrieWng mood assessment to evaluate this notion.

Study 1

Using a procedure closely modeled after that used by Ross et al. (1975), participants were exposed to either suc- cess or failure feedback on an alleged test of social percep- tiveness that involved distinguishing real from fake suicide notes. They then received one of four debrieWng inductions: (1) standard outcome debrieWng (i.e., participants were told about the “true” purposes, and that their score was false, predetermined, and randomly assigned), (2) revised out- come debrieWng (i.e., participants received standard out- come debrieWng and learned that the test was bogus), (3) process debrieWng (i.e., participants received standard out- come debrieWng and information regarding perseverance), or (4) no debrieWng. After the debrieWng variation, partici- pants provided current and future performance estimates, ratings of their speciWc and global abilities, and mood rat- ings. We predicted that perseverance would occur in only the standard outcome debrieWng condition, and that it would be limited to ratings of performance and speciWc task-relevant skills. We also evaluated whether aVective reactions mediated the perseverance eVect.

Method

Participants and design The participants were 67 female and 61 male SFU

undergraduates who took part individually and were pro-

vided with course credit for participating. They were ran- domly assigned to the conditions of a 2 (feedback: success vs. failure) £ 4 (debrieWng condition: no debrieWng vs. stan- dard outcome debrieWng vs. revised outcome debrieWng vs. process debrieWng) between-subjects factorial design. Males and females were distributed approximately equally across the conditions. Preliminary analyses indicated that there were no interactions involving gender; thus, the primary analyses do not include gender as a factor.

Procedure Participants were Wrst provided with a cover story indi-

cating that the researchers were exploring “personality traits and physiological responses during decision-making.” Accordingly, they Wrst completed a personality survey, after which they were attached to physiological recording equip- ment. During a “baseline assessment of arousal,” they read an information sheet indicating that the decision-making task involved reading 15 pairs of suicide notes, and select- ing the one “real” note from each pair. The task was described as a widely used measure of social perceptiveness- “the ability to make accurate judgments about other peo- ple’s behaviors and motives.” This ability was depicted as an important attribute that is linked to a wide variety of positive outcomes. Participants then completed the test, after which they were told to sit still while a “post-task reading” was obtained. At this time, the experimenter left the room brieXy to “score the test.”

Feedback manipulation The experimenter returned and presented the participant

with a feedback sheet tucked inside the test booklet. The feedback sheets were prepared in advance, ensuring that the experimenter remained blind to condition. Participants were told that their score was either 14/15 (success) or 4/15 (failure), and that the average was 9/15.

DebrieWng manipulation After a few minutes, the experimenter removed the elec-

trodes, and delivered one of the debrieWng variable induc- tions. Participants in the no debrieWng condition were told that while their credit form was being prepared they were to complete a Wnal “thoughts and reactions” questionnaire (see below). In the debrieWng conditions, the experimenter explained that the study was actually examining “the eVects of feedback on physiological responses,” and that it had therefore been necessary to provide false test feedback. Par- ticipants in the standard outcome debrieWng condition were informed that their score was a fake score that had been randomly assigned to them prior to their arrival. Addition- ally, they were shown a “random assignment schedule,” and the experimenter emphasized that the score contained absolutely no information about the participant’s actual performance or underlying abilities.

Participants in the revised outcome debrieWng condition were provided with the same information as that provided to those in the standard outcome debrieWng group.

236 C. McFarland et al. / Journal of Experimental Social Psychology 43 (2007) 233–240

Importantly, however, they were also told that the suicide- note test was a fake test that had been made to look like a real test. The experimenter explained that all of the suicide notes in the test were fake notes, and that the test did not measure any underlying abilities.

Participants in the process debrieWng condition were also provided with the same information as that provided to stan- dard outcome debrieWng participants. Additionally, however, they were informed about the nature of the perseverance eVect and how it might have personal relevance for them in this context. The experimenter highlighted that people’s beliefs sometimes persist even after debrieWng because they generate independent evidence that explains the feedback, and urged participants to avoid thinking in this way.

Dependent measures Next, the experimenter asked participants to complete a

“thoughts and reactions” questionnaire. They were assured of anonymity and asked to place the completed question- naire into a sealed unmarked envelope and place it amongst other unmarked envelopes. The Wrst section consisted of Wller questions (e.g., clarity of instructions) that validated the cover story. Participants then completed two items assessing their perceptions of performance on the suicide note task. One item requested that they estimate their score on the suicide note test out of 15. The wording of this ques- tion was carefully tailored to each debrieWng condition to ensure that the question made sense in light of the details of each condition. Participants in the no-debrieWng condition were asked to recall the score they had just been assigned. Participants in the standard outcome debrieWng and process debrieWng conditions (who were told that their score was fake) were asked to estimate their “actual score on the sui- cide note test (/15).” Participants in the revised outcome debrieWng condition, who were told that the test was not real, were asked the following: “Even though you now know that the task was not actually a real test of your abil- ity to distinguish genuine from Wctitious suicide notes, imagine for a moment right now that it had been a real test (i.e., one including real and fake notes). If you took this test now, what would you estimate to be your score on the test?.” Participants also estimated their future performance on a diVerent, but equally diYcult, set of genuine and fake notes. As well, they evaluated the speciWc ability to distin- guish real from fake suicide notes (1 D much lower ability than average; 11 D much higher ability than average).

Next, participants responded to 8 scales (range 1–11) assessing their perceptions of their global abilities: good at detecting another’s distress (extremely poor-extremely good); good at understanding why people behave in certain ways (extremely poor-extremely good); likelihood of pursuing a job requiring social perceptiveness skill (extremely unlikely-extremely likely); good at being a psychologist (extremely poor-extremely good); socially percep- tive (extremely low-extremely high); sensitive to others’ feel- ings (extremely insensitive-extremely sensitive); good at taking tests under pressure (extremely poor-extremely

good); ability to recognize falsehood (extremely poor- extremely good). The latter three items constitute Ross et al.’s (1975) “related skills” measure. Finally, participants rated their current moods (happy, satisWed, pleased, disappointed, sad, proud, and competent; 1Dnot at all; 9Dextremely).

Final debrieWng Participants in all conditions received a Wnal “process”

debrieWng. The experimenter explained the exact hypothe- ses, the necessity for the elaborate deception, and the fake nature of the test.

Results and discussion

Creation of indexes We Wrst constructed several indexes: (1) performance

estimate index (i.e., the average of current and future score estimates, ! D .81), (2) global abilities index (i.e., the average of the 8 global ratings, ! D .80), and (3) mood (i.e., the aver- age of the 7 mood ratings with the 2 negative items reverse scored, ! D .85).

Performance estimates We predicted that a perseverance eVect on performance

estimates would occur in only the standard outcome debrieWng condition. Revised outcome and process debrieWng were expected to eliminate perseverance. A 2 (feedback)£ 4 (debrieWng) ANOVA performed on this index revealed two main eVects (debrieWng: F (3, 120)D8.90, p < .0001; feedback: F (1, 120)D48.90, p < .0001) that were qualiWed by a signiWcant interaction eVect (F (3, 120)D57.76, p < .0001) that supported the prediction (see Table 1). Not surprisingly, among those who were not debriefed, participants who received success feedback reported higher estimates (M D13.18) than those who received failure feedback (M D4.46), t (120) D15.03, p < .001. Importantly, though, even among those who received standard outcome debrieWng, recipients of success feedback reported higher estimates (MD9.33) than recipients of failure feedback (MD7.96), t (120) D2.49, p < .05 (i.e., a perseverance eVect). The success–failure diVerence was not signiWcant (ts < 1) in the revised and process debrieWng groups, indicating that perseverance was eliminated in these conditions.4

4 On the performance estimate measure, the variability in the control groups was lower than that found in the debrieWng groups, probably be- cause control participants found it relatively easy to recall the score they received a short while before. Consequently, the overall error term may be somewhat lower than it would be if these control groups were not included in the analysis. To assess whether the perseverance eVect obtained in the standard outcome debrieWng group would remain signiWcant if an alterna- tive error term was used, we conducted an independent-groups t test that used the two pertinent cell variances to represent error variability. The per- severance eVect was maintained, t (30) D 2.40, p < .05. As well, we assessed whether the signiWcant interaction eVect would be preserved if a more stringent critical F value were used. We conducted the conservative Box test for heterogeneous variances outlined in Howell (2002), and the inter- action eVect remained signiWcant, p < .05. In sum, the Wndings on the per- formance estimate measure are not attributable to the lower variances in the control conditions.

C. McFarland et al. / Journal of Experimental Social Psychology 43 (2007) 233–240 237

SpeciWc ability rating Again, a perseverance eVect was anticipated in only the

standard outcome debrieWng condition. A 2 £ 4 ANOVA performed on the speciWc ability rating revealed two main eVects (debrieWng: F (3, 120) D 4.72, p < .01; feedback: F (1, 120) D 39.62, p < .0001) that were qualiWed by a signiW- cant interaction eVect (F (3, 120) D 50.59, p < .0001) that supported the prediction (see Table 1). As expected, non- debriefed participants who received success feedback reported a greater ability to distinguish real from fake sui- cide notes (M D 9.36) than those who received failure feed- back (M D 3.07), t (120) D 13.97, p < .001. Consistent with predictions, recipients of standard outcome debrieWng revealed a marginally signiWcant perseverance eVect (suc- cess: M D 7.20; failure M D 6.47), t (120) D 1.78, p < .08. Per- severance did not occur in the revised or process debrieWng groups (success vs. failure ts < 1, ns).

Global ability ratings One goal of our research was to examine whether perse-

verance occurs on the more global dimension purportedly assessed by the suicide note task. A 2 £ 4 ANOVA per- formed on the global ability index revealed that the perse-

verance eVect was not obtained on evaluations of general social perceptiveness skills. Although a signiWcant interac- tion was obtained (F (3, 120) D 2.63, p < .05), it occurred solely because non-debriefed “success” participants reported greater global ability (M D 7.84) than non- debriefed “failure” participants (M D 6.82), t (120) D 2.62, p < .01. No evidence for perseverance was obtained within any debrieWng condition (i.e., all ps > .20). Separate analyses of the individual items that were included in the overall global index revealed a similar pattern of eVects—no evi- dence of perseverance in any debrieWng group (again, all relevant ps > .20). Thus, it appears that when the persever- ance eVect occurs, it is limited to perceptions of speciWc task-relevant features (i.e., score estimates, and ratings of task speciWc ability). It is worth noting as well that although failure feedback decreased global ability ratings (relative to the combined failure debrieWng condition means) it does not appear that positive feedback increased global ratings. The debrieWng condition averages can be interpreted as representing a “baseline” global ability rating, and the mean in the no debrieWng/success group did not diVer from the mean of the combined success (or failure) debrieWng groups. It seems, then, that participants had rather

Table 1 Dependent variables as a function of type of performance feedback and debrieWng technique

Note. Higher scores on the performance estimate index reXect higher performance (/15). Higher scores on the ability measures indicate greater ability (1– 11). Within rows and columns for each measure, means not sharing a common subscript letter diVer at the .05 level (two-tailed). There is one exception: on the speciWc ability rating, the success vs. failure diVerence is marginally signiWcant within the standard debrieWng condition (p < .08).

Measure and type of feedback Type of debrieWng

No debrieWng Standard outcome debrieWng Revised outcome debrieWng Process debrieWng

Performance estimates Failure

M 4.46a 7.96b 10.38c 8.85b N 14 17 17 17 SD .49 1.40 1.81 2.11

Success M 13.18d 9.33c 10.08 c 8.03b N 14 15 17 17 SD .99 1.79 1.66 1.50

SpeciWc ability rating Failure

M 3.07a 6.47c 7.53d 7.12c, d N 14 17 17 17 SD .83 1.17 1.55 1.49

Success M 9.36 b 7.20d 7.24 d 6.65d N 14 15 17 17 SD 1.08 1.08 1.30 .93

Global ability ratings Failure

M 6.82a 7.69c 7.48c, a 7.47c, a N 14 17 17 17 SD 1.45 1.07 1.23 1.15

Success M 7.84c 7.45c 7.86c 7.01c N 14 15 17 17 SD 1.10 1.09 1.08 .85

238 C. McFarland et al. / Journal of Experimental Social Psychology 43 (2007) 233–240

favorable preexisting views of their global social perceptive- ness abilities, and that positive feedback did not further ele- vate these self-views.

Mood ratings To test whether post-debrieWng mood reactions engender

perseverance in self-perceptions through mood-congruent processing, we Wrst conducted a 2 £ 4 ANOVA on the mood index. This analysis revealed a marginally signiWcant interac- tion eVect (F (3, 120) D 2.41, p < .07) wherein the diVerence between the success and failure conditions was signiWcant only among non-debriefed participants (success M D 6.77; failure M D 5.47; t (120)D 3.25, p < .001). If aVective reactions underlie the perseverance eVect obtained in the standard out- come debrieWng condition, participants in this group who received success feedback should have reported more posi- tive moods (M D 6.33) than those who received failure feed- back (M D 6.42), and they did not, t < 1, ns. As an additional test, we reran the ANOVAs conducted on the performance estimate and speciWc ability measures (i.e., the measures that revealed perseverance) using mood as a covariate. Both inter- action eVects were maintained when mood was controlled (estimates (F (3, 119) D 53.23, p < .0001); speciWc ability rating (F (3, 119)D 46.83, p < .0001)). Thus, the perseverance eVect occurring on these measures does not appear to be due to aVective perseverance.

Study 2

Study 1 revealed that a slightly revised form of out- come debrieWng can be as eVective as process debrieWng in eliminating perseverance. When participants were informed not only that their score was false, but also that the test was not a valid measurement tool, their post- debrieWng self-perceptions were uninXuenced by false feedback. In Study 2, we explored the possibility that revised outcome debrieWng may be more eVective than standard outcome debrieWng because it preempts a pat- tern of thought that occurs in the latter form of debrieWng. When participants receive standard outcome debrieWng, they probably Wnd themselves contemplating or ruminat- ing about their “real score” on what they have been led to believe is a valid test. This curiosity about their real score may lead them to use their assigned score as a subjective anchor point with which to estimate their real score (Weg- ner et al., 1985), and the constructed score representation could lead to further thoughts that reXect explanations of the hypothetical score (Fleming & Arrowood, 1979; Ross et al., 1975). These types of thoughts would be expected to yield perseverance in self-perceptions. In contrast, in revised outcome debrieWng, participants learn that the test is bogus, and thus they should be much less likely to rumi- nate about what their real score might be. Study 2 explored this reasoning. Recipients of failure feedback received one of three forms of debrieWng: standard out- come, revised outcome, or process debrieWng. Immedi- ately after debrieWng, we obtained several measures

reXecting the degree of contemplation regarding actual performance levels. We expected that recipients of revised outcome debrieWng would ruminate less about their real performance than participants in the other conditions.

Method

Participants Participants were 6 male and 24 female SFU undergrad-

uates who received course credit.

Procedure The procedure closely paralleled that used in the fail-

ureEdebrieWng conditions of Study 1 up to the point at which the initial debrieWngs were completed. At this point, the experimenter left the room for 4 min, and returned with the “thoughts and reactions questionnaire.” After a couple of Wller questions, participants were informed that we were interested in the thoughts that people have after learning about our true purposes, and that they should indicate the degree to which they had certain thoughts during the past few minutes. On a scale ranging from 1 (not at all) to 11 (a great deal), they indicated the degree to which they found themselves (a) wondering about or contemplating what their real score on a test of social perceptiveness might be, (b) wondering about what their actual ability level is in the domain of social perceptiveness, and (c) thinking that the score they were given on the test might be a useful starting point for estimating their actual performance/ability on a test of social perceptiveness. They were then fully debriefed.5

Results

We predicted that participants in the revised outcome debrieWng condition would think less about their real score or ability level than participants in the other debrieWng con- ditions. A one-way ANOVA performed on an index reXect- ing the average of the three contemplation items (! D .76) revealed support for this prediction, F (2, 27) D 5.56, p < .01. Immediately after debrieWng, recipients of revised outcome debrieWng reported thinking about their score and ability level signiWcantly less (M D 5.16) than did recipients of stan- dard outcome debrieWng (M D 7.56; t (27) D 2.75, p < .01) or process debrieWng (M D 7.76; t (27) D 2.98, p < .01).

5 We did not include the perseverance measures because we reasoned that they would be tainted by the previous act of completing the contem- plation measures, and thus would not yield the typical perseverance eVect (i.e., lower ratings in the standard outcome debrieWng condition than the other conditions). We reasoned that even participants who reported think- ing little about their real score during the previous few minutes (i.e., those in the revised outcome debrieWng group) would be prompted to think about it after completing items that inquired explicitly about this score. Thus, the contemplation measures are reactive in the sense that completing them would likely create perseverance in most participants.

C. McFarland et al. / Journal of Experimental Social Psychology 43 (2007) 233–240 239

General discussion

Prior to the publication of Ross et al.’s research in 1975, it seems likely that most social psychologists simply assumed that their debrieWngs were eVective. After all, why would research participants who are told that they lack an important intellectual or social skill not embrace informa- tion conveying that the feedback is invalid? By demonstrat- ing the perseverance eVect, and revealing a strategy to eliminate this phenomenon, Ross and colleagues provided the Weld with both a “wake-up call” and a solution. In the past quarter century, process debrieWng has become the most widely accepted protocol for the proper conduct of debrieWngs (e.g., Aronson, Ellsworth, Carlsmith, & Gonza- les, 1990) and authors often mention explicitly that this is the form of debrieWng that they delivered.

The current studies build upon this classic early work in several ways. First, our Wndings revealed that there is a form of outcome debrieWng that is as eVective as process debrieWng in eliminating perseverance. It appears that the addition of one piece of information (i.e., information that the test is invalid or bogus), is suYcient to prevent perse- verance in performance-related self-perceptions. Based on our Wndings, one might be tempted to suggest that researchers use revised outcome debrieWng in lieu of pro- cess debrieWng. This position is not without merit. Revised outcome debrieWng appears to work because it preempts the ruminative processing that normally serves to solidify feedback-based self-perceptions. Successful process debrieWng, in contrast, probably requires that participants engage in a more eVortful or controlled corrective process (e.g., Wegener & Petty, 1997) wherein they adjust rumina- tion-inspired self-perceptions in a direction that opposes the feedback. Study 2 oVered some indirect support for this possibility: Participants in the process debrieWng con- dition were as likely as those in the standard outcome debrieWng condition to report ruminating about their per- formance; nonetheless, they managed to avoid persever- ance (see Study 1). If this general reasoning is correct, it could be argued that revised outcome debrieWng will be more eVective than process debrieWng among participants who are less motivated or able to engage in an eVortful correction process. Despite this advantage, however, we believe that a “combined” debrieWng would be most eYcacious [i.e., one that incorporates the best features of both revised outcome (i.e., information about test invalid- ity) and process debrieWng (i.e., information about perse- verance)]. This form of debrieWng would not only preempt perseverance among participants who are less inclined to engage in correction, but also provide a long-term strat- egy for all participants to use in the event that they experi- ence feedback-consistent self-thoughts after leaving the laboratory. Combined debrieWng does require that researchers use test items that do not actually represent valid measures of underlying ability (e.g., that they use word completions rather than GRE test items as measures of “intellectual aptitude”); however, this should not gen-

erally pose a problem because studies incorporating the false feedback paradigm rarely call for the use of vali- dated tests of underlying abilities. Participants need only believe that a test measures important abilities for this paradigm to be eVective.

Our research also extends Ross et al.’s research by conWrming that the perseverance eVect is restricted to people’s perceptions of speciWc performances and abili- ties. This lack of perseverance on evaluations of global ability cannot be easily attributed to methodological fac- tors. The global ability index included multiple items, was internally consistent, and was sensitive to negative feedback. Moreover, the cover story delivered to partici- pants stressed that the test measured important general abilities. We can only speculate, but the absence of perse- verance on this index may reXect the operation of self- enhancement or self-veriWcation motives. Much past research has revealed that the average person processes information in a manner that conWrms a generally posi- tive stable self-view (e.g., Swann, Pelham, & Krull, 1989; Taylor & Brown, 1988). The results on the global ability index are consistent with this general Wnding. “Failing” participants appear to have been quite willing to relin- quish negative global perceptions that were formed tem- porarily in reaction to the feedback. Additionally, “succeeding” participants appear to have been unwilling to modify (in response to either feedback or debrieWng) the highly positive self-views that they “brought with them” to the experiment. Overall, the Wndings on the global ability index can be taken as “good news” for researchers who use false feedback. In an attempt to cap- ture the strong motivational forces and threatening emo- tions that aVect people in everyday life, researchers often portray their tests as measuring highly important and global abilities. Our results imply that when researchers use this strategy, their participants are unlikely to suVer long-term negative consequences.

Finally, our results revealed that the perseverance eVect obtained in the debrieWng paradigm is not due to “aVective perseverance.” Participants’ aVective reactions revealed no evidence of perseverance, and perseverance in self-perceptions occurred even when mood reactions were controlled. Thus, debrieWng of any kind appears to be eVective in eliminating the mood changes produced by false feedback. Although some researchers have obtained evidence for aVective perseverance in other paradigms (Sherman & Kim, 2002), perseverance in the debrieWng paradigm appears to be due to more cognitive (e.g., rumi- native) processes.

In conclusion, our research highlights that there are sev- eral critical features of an eVective debrieWng in the false feedback paradigm. In particular, the distinction between “debrieWng about the false nature of the feedback” and “debrieWng about the false or unvalidated nature of the test itself” appears to be an important one, and we hope that future researchers will construct debrieWngs that reXect both of these components.

240 C. McFarland et al. / Journal of Experimental Social Psychology 43 (2007) 233–240

References

Anderson, C. A., Lepper, M. R., & Ross, L. (1980). Perseverance of social theories: The role of explanation in the persistence of discredited infor- mation. Journal of Personality and Social Psychology, 39, 1037–1049.

Aronson, E., Ellsworth, P. C., Carlsmith, J., & Gonzales, M. H. (1990). Methods of research in social psychology. New York: McGraw-Hill.

Baumeister, R. F., Heatherton, T. F., & Tice, D. M. (1993). When ego threats lead to self-regulation failure: Negative consequences of high self-esteem. Journal of Personality and Social Psychology, 64, 141–156.

Blaine, B., & Crocker, J. (1993). Self-esteem and self-serving biases in reac- tions to positive and negative events. In R. Baumeister (Ed.), Self- esteem: The puzzle of low self-regard (pp. 55–85). New York: Plenum.

Brown, J. D., & Gallagher, F. M. (1992). Coming to terms with failure: Pri- vate self-enhancement and public self-eVacement. Journal of Experi- mental Social Psychology, 28, 3–22.

Brown, J. D., & Dutton, K. A. (1995). The thrill of victory, the complexity of defeat: Self-esteem and people’s emotional reactions to success and failure. Journal of Personality and Social Psychology, 68, 712–722.

Brown, J. D., & Marshall, M. A. (2001). Self-esteem and emotion: Some thoughts on feelings. Personality and Social Psychology Bulletin, 27, 575–584.

Brown, J. D., & Smart, S. A. (1991). The self and social conduct: Linking self-representations to prosocial behaviour. Journal of Personality and Social Psychology, 60, 368–375.

Brown, J. D., & Mankowski, T. A. (1993). Self-esteem, mood, and self-eval- uation: Changes in mood and the way you see you. Journal of Personal- ity and Social Psychology, 64, 421–430.

Cervone, D., & Palmer, P. W. (1990). Anchoring biases and the persever- ance of self-eYcacy beliefs. Cognitive Therapy and Research, 14, 401–416.

Crocker, J., Thompson, L., McGraw, K. M., & Ingerman, C. (1987). Down- ward comparison, prejudice, and evaluations of others: EVects of self- esteem and threat. Journal of Personality and Social Psychology, 52, 907–916.

Di Paula, A., & Campbell, J. D. (2002). Self-esteem and persistence in the face of failure. Journal of Personality and Social Psychology, 83, 711–724.

Dodgson, P. G., & Wood, J. V. (1998). Self-esteem and the cognitive acces- sibility of strengths and weaknesses after failure. Journal of Personality and Social Psychology, 75, 178–197.

Dunning, D., Leuenberger, A., & Sherman, D. A. (1995). A new look at motivated inference: Are self-serving theories of success a product of motivational forces? Journal of Personality and Social Psychology, 69, 58–68.

Fein, S., & Spencer, S. J. (1997). Prejudice as self-image maintenance: AYrming the self through derogating others. Journal of Personality and Social Psychology, 73, 31–44.

Fleming, J., & Arrowood, A. J. (1979). Information processing and the per- severance of discredited self-perceptions. Personality and Social Psy- chology Bulletin, 7, 257–263.

Forgas, J. P. (2000). Feeling and thinking: The role of aVect in social cogni- tion. New York: Cambridge University Press.

Forgas, J. P., & Ciarrochi, J. V. (2002). On managing moods: Evidence for the role of homeostatic cognitive strategies in aVect regulation. Person- ality and Social Psychology Bulletin, 28, 336–345.

Howell, D. C. (2002). Statistical methods for psychology. PaciWc Grove, CA: Duxbury.

McFarland, C., & Buehler, R. (1995). Collective self-esteem as a moderator of the frog-pond eVect in reactions to performance feedback. Journal of Personality and Social Psychology, 68, 1055–1070.

McFarland, C., & Buehler, R. (1998). The impact of negative aVect on autobiographical memory: The role of self-focused attention to moods. Journal of Personality and Social Psychology, 75, 1424–1440.

McFarlin, D. B. (1985). Persistence in the face of failure: The impact of self-esteem and contingency information. Personality and Social Psy- chology Bulletin, 11, 152–163.

Mussweiler, T., Strack, F., & PfeiVer, T. (2000). Overcoming the inevitable anchoring eVect: Considering the opposite compensates for selective accessibility. Personality and Social Psychology Bulletin, 26, 1142–1150.

Nummenmaa, L., & Niemi, P. (2004). Inducing aVective states with suc- cess-failure manipulations: A meta-analysis. Emotion, 4, 207–214.

Ross, L., Lepper, M. R., & Hubbard, M. (1975). Perseverance in self-per- ception and social perception: Biased attributional processes in the debrieWng paradigm. Journal of Personality and Social Psychology, 32, 880–892.

Ross, L., Lepper, M. R., Strack, F., & Steinmetz, J. (1977). Social explana- tion and social expectation: EVects of real and hypothetical explana- tions on subjective likelihood. Journal of Personality and Social Psychology, 35, 817–829.

Schwarz, N., & Clore, G. L. (1996). Feelings and phenomenal experiences. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Hand- book of basic principles (pp. 433–465). New York: Guilford Press.

Sedikides, C., & Green, J. D. (2001). AVective inXuences on the self-concept: Qualifying the mood-congruency principle. In J. P. Forgas (Ed.), Hand- book of aVect and social cognition (pp. 145–160). London: Erlbaum.

Sherman, D. K., & Kim, H. S. (2002). AVective perseverance: The resis- tance of aVect to cognitive invalidation. Personality and Social Psychol- ogy Bulletin, 28, 224–237.

Stucke, T. S., & Sporer, S. L. (2002). When a grandiose self-image is threat- ened: Narcissism and self-concept clarity as predictors of negative emotions and aggression following ego-threat. Journal of Personality, 70, 509–532.

Swann, W. B., Pelham, B. W., & Krull, D. S. (1989). Agreeable fancy or dis- agreeable truth: Reconciling self-enhancement and self-veriWcation. Journal of Personality and Social Psychology, 57, 782–791.

Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social-psy- chological perspective on mental health. Psychological Bulletin, 103, 193–210.

Tesser, A., Millar, M., & Moore, J. (1988). Some aVective consequences of social comparison and reXection processes: The pain and pleasure of being close. Journal of Personality and Social Psychology, 54, 49–61.

Wegener, D. T., & Petty, R. E. (1997). The Xexible correction model: The role of naïve theories of bias in bias correction. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 29, p. 141). San Diego, CA: Academic Press.

Wegner, D. M., Coulton, G. F., & WenzlaV, R. (1985). The transparency of denial: BrieWng in the debrieWng paradigm. Journal of Personality and Social Psychology, 49, 338–346.

Wood, J. V., Giordano-Beech, M., & Ducharme, M. J. (1999). Compensat- ing for failure through social comparison. Personality and Social Psy- chology Bulletin, 25, 1370–1386.

Wood, J. V., Giordano-Beech, M., Taylor, K. L., Michela, J. L., & Gaus, V. (1994). Strategies of social comparison among people with low self- esteem: Self-protection and self-enhancement. Journal of Personality and Social Psychology, 67, 713–731.