11
Temporal perspective moderates self-reported responses to social exclusion Amy M. Garczynski a, , Christina M. Brown b a Department of Psychology, Saint Louis University, USA b Department of Psychology, Arcadia University, USA HIGHLIGHTS Framing (past or present tense) of self-reports shapes responses to exclusion. Exclusion hurts more when reported in the past tense. Researchers should use care when assessing temporal effects of exclusion. abstract article info Article history: Received 1 May 2013 Revised 11 October 2013 Available online 15 February 2014 Keywords: Social exclusion Ostracism Social rejection Tense Temporal perspective Framing Self-reported feelings of personal distress (i.e., thwarted needs for belonging, lowered self-esteem) following social exclusion are commonly used as the sole determinant of whether an event was experienced as rejection as well as whether a person has recovered from the experience (e.g., Zadro, Williams, & Richardson, 2004). However, the present research reveals that the temporal framing (past or present tense) of self-report measures shapes responses. In two studies, we manipulated social exclusion and the tense of self-report personal distress measures (i.e., basic needs satisfaction and self-esteem). The results suggest that differences based on tense are the result of biased self-reports (due to social desirability concerns or implicit theories of change over time), rather than representing actual recovery from exclusion. The present research highlights the importance of attending to question tense when assessing reactions to social exclusion. © 2014 Elsevier Inc. All rights reserved. Introduction Although people have many essential physical and psychological needs, the need for social belonging is high in motivational priority (e.g., Baumeister & Leary, 1995; Kenrick, Griskevicius, Neuberg, & Schaller, 2010). When this need is thwarted through social exclusion, rejection, or ostracism, people self-report lowered feelings of belonging, self-esteem, control, and meaningful existence, as well as exhibit marked changes in behavior ranging from increased aggression to in- creased pursuit of social afliation (e.g., Williams, 2007a). Researchers have begun to identify some of the contextual factors that moderate how people respond to social rejection (see Smart Richman & Leary, 2009, for a review), but one contextual factor that has not been explored is the temporal perspective taken when reecting on the experience of social exclusion. Self-reported feelings of social exclusion are the most widely used measures assessing the impact of social rejection and ostracism (e.g., Bernstein, Sacco, Young, Cook, & Hugenberg, 2010; Gonsalkorale & Williams, 2006; Wirth, Sacco, Hugenberg, & Williams, 2010). In this paper we reveal that the temporal framing, or phrasing, of these self-report measures shapes people's responses. Temporal theory of rejection We propose that whether socially excluded people are asked to con- sider either their current feelings or their feelings during the exclusion event will inuence the level of distress reported, with greater distress re- ported in the past than in the present. One potential explanation for this phenomenon is articulated in Williams' (2001, 2007a, 2007b) temporal theory of emotional responses to rejection. This theory stipulates that re- actions to social exclusion can be separated into an immediate reexivestage and a subsequent reectivestage. The reexive stage is character- ized by an immediate feeling of social pain, an experience akin to physical pain (e.g., Eisenberger, Lieberman, & Williams, 2003). People in the reex- ive stage also have a conscious experience of personal distress, which takes the form of feeling that important needs (including belonging, con- trol, self-esteem, and meaningful existence) have been thwarted. In addi- tion to thwarted needs, there has been mixed evidence that this experience of personal distress can also include feeling negative emotions Journal of Experimental Social Psychology 53 (2014) 4050 Corresponding author at: Institutional Research, Ashford University, USA. E-mail address: [email protected] (A.M. Garczynski). http://dx.doi.org/10.1016/j.jesp.2014.02.003 0022-1031/© 2014 Elsevier Inc. All rights reserved. Contents lists available at ScienceDirect Journal of Experimental Social Psychology journal homepage: www.elsevier.com/locate/jesp

Temporal perspective moderates self-reported responses to social exclusion

Embed Size (px)

Citation preview

Journal of Experimental Social Psychology 53 (2014) 40–50

Contents lists available at ScienceDirect

Journal of Experimental Social Psychology

j ourna l homepage: www.e lsev ie r .com/ locate / j esp

Temporal perspective moderates self-reported responses tosocial exclusion

Amy M. Garczynski a,⁎, Christina M. Brown b

a Department of Psychology, Saint Louis University, USAb Department of Psychology, Arcadia University, USA

H I G H L I G H T S

• Framing (past or present tense) of self-reports shapes responses to exclusion.• Exclusion hurts more when reported in the past tense.• Researchers should use care when assessing temporal effects of exclusion.

⁎ Corresponding author at: Institutional Research, AshfE-mail address: [email protected] (A.M. Garc

http://dx.doi.org/10.1016/j.jesp.2014.02.0030022-1031/© 2014 Elsevier Inc. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 1 May 2013Revised 11 October 2013Available online 15 February 2014

Keywords:Social exclusionOstracismSocial rejectionTenseTemporal perspectiveFraming

Self-reported feelings of personal distress (i.e., thwarted needs for belonging, lowered self-esteem) followingsocial exclusion are commonly used as the sole determinant of whether an event was experienced as rejectionas well as whether a person has recovered from the experience (e.g., Zadro, Williams, & Richardson, 2004).However, the present research reveals that the temporal framing (past or present tense) of self-report measuresshapes responses. In two studies, we manipulated social exclusion and the tense of self-report personal distressmeasures (i.e., basic needs satisfaction and self-esteem). The results suggest that differences based on tense arethe result of biased self-reports (due to social desirability concerns or implicit theories of change over time),rather than representing actual recovery from exclusion. The present research highlights the importance ofattending to question tense when assessing reactions to social exclusion.

© 2014 Elsevier Inc. All rights reserved.

Introduction

Although people have many essential physical and psychologicalneeds, the need for social belonging is high in motivational priority(e.g., Baumeister & Leary, 1995; Kenrick, Griskevicius, Neuberg, &Schaller, 2010). When this need is thwarted through social exclusion,rejection, or ostracism, people self-report lowered feelings of belonging,self-esteem, control, and meaningful existence, as well as exhibitmarked changes in behavior ranging from increased aggression to in-creased pursuit of social affiliation (e.g., Williams, 2007a). Researchershave begun to identify some of the contextual factors that moderatehow people respond to social rejection (see Smart Richman & Leary,2009, for a review), but one contextual factor that has not been exploredis the temporal perspective taken when reflecting on the experience ofsocial exclusion. Self-reported feelings of social exclusion are the mostwidely used measures assessing the impact of social rejection andostracism (e.g., Bernstein, Sacco, Young, Cook, & Hugenberg, 2010;Gonsalkorale & Williams, 2006; Wirth, Sacco, Hugenberg, & Williams,

ord University, USA.zynski).

2010). In this paper we reveal that the temporal framing, or phrasing,of these self-report measures shapes people's responses.

Temporal theory of rejection

We propose that whether socially excluded people are asked to con-sider either their current feelings or their feelings during the exclusioneventwill influence the level of distress reported,with greater distress re-ported in the past than in the present. One potential explanation for thisphenomenon is articulated in Williams' (2001, 2007a, 2007b) temporaltheory of emotional responses to rejection. This theory stipulates that re-actions to social exclusion can be separated into an immediate “reflexive”stage and a subsequent “reflective” stage. The reflexive stage is character-ized by an immediate feeling of social pain, an experience akin to physicalpain (e.g., Eisenberger, Lieberman, &Williams, 2003). People in the reflex-ive stage also have a conscious experience of personal distress, whichtakes the form of feeling that important needs (including belonging, con-trol, self-esteem, andmeaningful existence) have been thwarted. In addi-tion to thwarted needs, there has been mixed evidence that thisexperience of personal distress can also include feeling negative emotions

41A.M. Garczynski, C.M. Brown / Journal of Experimental Social Psychology 53 (2014) 40–50

like anger and sadness (Blackhart, Nelson, Knowles, & Baumeister, 2009;Gerber & Wheeler, 2009; Williams, 2007a, 2007b). Williams (2007a,2007b) argues that all people respond in the sameway during the reflex-ive stage, as evidenced by studies showing that personality does notmod-erate the degree of distress participants remember feeling during theexclusion event (e.g., Zadro, Boland, & Richardson, 2006). However, it'simportant to note that this research has asked people to report howthey felt during the exclusion event after the event has occurred.(In fact, personality factors have been found to moderate responses tosocial exclusion when affect is measured during ostracism instead ofretrospectively; Wesselmann, Wirth, Mroczeka, & Williams, 2012.)

Williams (2007a) proposed that the reflexive stage is followed by areflective stage duringwhich the person appraises the exclusion experi-ence, such as the reason for the exclusion.Williams posits that reactionsto exclusion in the reflective stage aremoderated by personality and thecircumstances surrounding the exclusion. Because he believes peoplequickly move from the reflexive stage to the reflective stage, Williamsargues that self-report measures of personal distress (e.g., loweredbasic needs) should be phrased in the past tense to capture how partic-ipants felt when they were still in the painful reflexive stage. He writes:

This distinction becomes important because the available evidencesuggests that the reflexive pain/distress signal is quickly followedby appraisals and coping mechanisms that direct the individual to-ward thoughts and feelings that alleviate the pain. To be includedin this section, the subsequent dependent variable measures must,therefore, have been taken during or immediately following the os-tracism experience and must pertain to their responses during theostracism experience.

[Williams, 2007a, p. 432]

In other words, if participants report less distress when asked in thepresent tense than in the past, it is because the coping and appraisalprocesses of the reflective stage have reduced their distress. But whatif framing self-report measures in the present tense does not merely re-duce excluded participant's distress, but eliminates the effect of rejec-tion altogether?

If there is a null effect of exclusion on self-reported present distresswhen feelings are reported immediately after the experience, to comportwith Williams' temporal theory this would require that recovery fromexclusion is near-instantaneous, a conclusion that contradicts the resultsof other research. A number of studies reveal that excluded participantsare still reacting intensely to the exclusion at this time, as displayed byphysiological and behavioral measures taken immediately after the ex-clusion manipulation or even after self-report surveys (e.g., Baumeister,Twenge, & Nuss, 2002; Stroud, Tanofsky-Kraff, Wilfley, & Salovey,2000; Zhong & Leonardelli, 2008; Zwolinski, 2012). In the sectionsbelow, we compare responses to exclusion recorded immediately afterthe event and after a delay.

Responses immediately following exclusion.Several experiments support the claim that individuals still exhibit

strong responses to social exclusion shortly after the ostracism experi-ence. Physiological measures of emotional arousal taken immediately fol-lowing social exclusion indicate elevated blood pressure and cortisol (astress hormone; Stroud et al., 2000). Excluded people also embody thefeeling of social isolation in the form of physical coldness, as measuredby reporting colder room temperatures and desiring hot foods immedi-ately following social exclusion (Zhong & Leonardelli, 2008). Additionally,social exclusion impairs performance on cognitive tests (Baumeister et al.,2002), which suggests that attention remains focused on or distracted bythe exclusion experience.

Delayed responses following exclusion

These physiological effects can persist for an extended period of timeas well. For instance, 20 min after a social inclusion or exclusion task,

women who were the targets of social exclusion still exhibited higherlevels of cortisol than women who had been socially included 20 minearlier (Zwolinski, 2012). The same study found that 20 min followingan exclusion manipulation, excluded men exhibited greater hostilitythan included men. It is important to note that while this finding hasbeen corroborated by researchdemonstrating heightened cortisol levelsat 30 and 45min following social rejection (e.g., Blackhart, Eckel, & Tice,2007), other research teams have found no cortisol increases after ex-clusion when examining women only (Zöller, Maroof, Weik, &Deinzer, 2010) and both women and men (Weik, Maroof, Zöller, &Deinzer, 2010). However, both studies that found a null-effect of exclu-sion on cortisol did obtain a significant exclusion effect on self-reportednegative feelings (in Zöller et al., 2010, the affect measure was retro-spective; the framing used in Weik et al., 2010, is unclear).

Interestingly, Weik et al. (2010) found evidence that women experi-ence a delayed physiological effect of exclusion: Althoughwomen's cor-tisol levels did not increase immediately after exclusion, when theywere subsequently placed in a situation that does reliably increasecortisol (public speaking), their cortisol response was blunted relativeto included participants. The authors speculate that women's well-established “tend and befriend” response to social stress may includeincreased oxytocin and vasopressin, which are known to inhibit stress-induced cortisol (Weik et al., 2010). Critically, this cortisol-bluntingresponse in excluded women was found over 1 h after the exclusionexperience, suggesting that the effect of exclusion can be prolonged.

Collectively, these and other studies indicate that recovery from re-jection is far from complete by the time self-reported feelings areassessed. Contrary to these findings, Williams' (2007a) earlier quote in-dicates that self-report measures administered after exclusion can onlycapture the reflexive stage if they ask participants how they felt mo-ments before (when the ostracism occurred), whereas asking partici-pants how they feel right now accurately captures current feelings,which may reflect differences in coping in the reflective stage. Assupport for this model, Williams (2007b) cites a study by Zadro et al.(2006) inwhich thepersonality factor of social anxiety did notmoderateself-reported personal distress immediately after ostracism experiencedvia an online ball-toss game (Cyberball)whereas it didmoderate contin-ued personal distress 45min later (i.e., everyone is upset at the reflexivestage, but individual differences in social anxiety shape coping duringthe reflective stage). However, a critical flaw with this study is that thetime of measurement, which is intended to assess reactions at the re-flexive and reflective stage, is confounded with the temporal phrasingof the questions. Their first self-report measure of personal distressasked participants “to answer the questions according to how they felt‘while playing the game’” (p. 693), whereas participants answered the“second test according to how they felt ‘right now’” (p. 694). It is impor-tant to note that other research has also confounded item tense withtime (using past tense immediately after ostracism and present tenseafter a delay) to capture the persistence of social exclusion (Goodwin,Williams, & Carter-Sowell, 2010; Wirth & Williams, 2009). We do notdoubt that personality moderates recovery from exclusion (indeed,Wesselmann et al., 2012, found evidence for personality moderationduring exclusion when feelings are measured currently and continuous-ly), but we suggest that confounding tense with time can lead themag-nitude of the effect of personality moderators to be overestimated.

What if both immediate and delayed measures were in the sametense, either past or present? Earlierwe explainedWilliams' (2007a) ar-gument for assessing personal distress in the past tense to capture thereflexive stage. If people don't report current (present tense) personaldistress immediately after ostracism,wemight conclude fromWilliams'theory that people are now in the reflective stage and have alreadycoped with the exclusion or appraised it as meaningless. Yet our earlierreview of past research clearly shows that most people haven't recov-ered from social exclusion by the time researchers administer their de-pendent measures. So if people do not report personal distress in thepresent tense immediately after social exclusion, another explanation

42 A.M. Garczynski, C.M. Brown / Journal of Experimental Social Psychology 53 (2014) 40–50

is necessary. In the next section we introduce alternate reasons to be-lieve that question framing might moderate self-reported feelings ofrejection.

Framing effects through social desirability and implicit theories

We believe that question framing and phrasing, rather than time it-self, will shape responses to rejection for two possible reasons: Peoplemay not want to admit current weakness or hurt, or they might retro-spectively overestimate the intensity of their affect.

First, it is possible that people are not comfortable reporting currentfeelings of personal distress. For instance, research on social desirabilitydemonstrates that people exhibit a tendency to over-report good behav-ior or outcomes and under-report bad behavior or outcomes (Crowne &Marlowe, 1960). Hence, social desirability concerns can drive people toprovide answers on questionnaires that create a positive social imageof themselves. As a simple example, masculine self-presentational con-cerns cause men to report different emotional reactions to stimuli thatare stereotypically feminine when alone versus in the presence ofother men (Fisher & Dubé, 2005). In the case of social exclusion, it mayseem more socially desirable to report past personal distress than toadmit a current emotional weakness. This could cause rejection effectsto appear more pronounced when reported in the past tense than inthe present tense. If so, a social desirability explanation posits that pasttense self-reports are more honest, whereas self-reports in the presenttense are biased as a result of self-presentation concerns.

A second explanation, unrelated to social desirability, is that retro-spective judgments are exaggerated to be more intense. To illustrate,Smith, Leffingwell, and Ptacek (1999) asked students to complete dailymeasures of how they were coping while preparing for a major exam.When students recalled their coping a week later, they overestimatedhowmuch they had actually engaged in coping behaviors. Additionally,researchers who manipulated wording tense found that people recallmore intense emotional responses when reporting events in the pasttense, relative to present or past-perfect tense (Hart, 2012; Schwarz,1999; Thomas & Diener, 1990). This research suggests that rejection ef-fects aremore negativewhen reported in the past tense than thepresenttense simply because memories become exaggerated.

If the second interpretation—that people overreport their past emo-tions—is correct, why does this phenomenon occur? One explanationcomes from research on implicit theories about change over time.Specifically, people tend to retrospectively change memories of them-selves to fit their beliefs about stability over time. For example, peoplewho believe attitudes remain stable “remember” their past self asbeing more similar to their current self and people who believe ingrowth over time “remember” their past self as weaker than theircurrent self (e.g., Kanten & Teigen, 2008; Karniol & Ross, 1996). In thedomain of social exclusion, perhaps people who are still feeling thesting of rejection believe their feelings must have been more intenseduring the actual exclusion event because this fits their implicit theoryabout emotional dissipation over time.

To summarize, there are several reasons that people might reportless distress when asked to consider their past feelings. Social desirabil-ity concerns could motivate people to underreport how distressed theycurrently are because they don't want to appear “weak,” whereas im-plicit theories could cause individuals to overestimate the intensity oftheir past emotions because they believe feelings during the actual ex-clusion event should bemore intense thanwhat they feel now. And last-ly, Williams' (2007a, 2007b) temporal theory argues that when presentfeelings are less negative, it is because the person is accurately reportinglessened distress as a result of reflection.

The present research

We hypothesized that the tense used when asking excluded partic-ipants to report their feelings (of belonging, self-esteem, etc.) influences

responses.We conducted two studies inwhichwe induced social exclu-sion (using the virtual ostracism paradigm of Cyberball in Experiment 1and an autobiographical memory task in Experiment 2) and asked par-ticipants to report feelings of social exclusion and self-esteem, manipu-lating whether these questions were asked in the past or present tense.For comparison, we includedmanipulations of social inclusion and con-trol experiences.

We expected excluded participants to report more rejected feelingsand lowered self-esteemwhen asked in the past tense than the presenttense. The goal of the present research was to establish that questionframing shapes self-reported distress after social exclusion. It is impor-tant to establish the effect of word tense on self-reported distress be-cause these measures are commonly used as the primary dependentmeasure to establish if an event was experienced as social exclusion(e.g., Bernstein et al., 2010; Gonsalkorale & Williams, 2006; Wirthet al., 2010; Zadro, Williams, & Richardson, 2004; Zadro et al., 2006). Itseems intuitive to assess whether people are currently feeling rejectionand lowered state self-esteem—after all, the goal is usually to establishthat the participant's phenomenological state is one of being excluded—yet Williams (2007a) explicitly argues for measuring past feelings.Indeed, commonly used measures of state self-esteem and currentmood are explicitly in the present tense (e.g., Baumeister & Leary, 1995;Mayer & Gaschke, 1988), and many researchers advise recording currentfeelings to minimize retrospective recall errors (Schwarz, 2007; Smithet al., 1999). Rejection and social pain are affective experiences, so whyshould they be different?

In fact, if temporal perspective does moderate self-reported distressin response to social exclusion, it may explain some inconsistencies inthe literature. There are two recent but conflictingmeta-analyses on so-cial exclusion's effect on emotions; one found that it consistently lowersself-esteem and mood (Gerber & Wheeler, 2009) while another con-cluded that there are no effects on the same measures (Blackhartet al., 2009). Interestingly, many of the studies finding no effect ofexclusion on mood and self-esteem assess these in the present tense(e.g., Baumeister et al., 2002; Twenge, Baumeister, DeWall, Ciarocco, &Bartels, 2007), while those finding large effects measure them in thepast tense (e.g., Gonsalkorale & Williams, 2006; Wirth & Williams,2009). Although there may certainly be other critical moderators, suchas how exclusion is manipulated (e.g., the former citations used a falsepersonality feedbackmanipulation, the latter used Cyberball), temporalperspective may contribute to identifying when social exclusion has alarge effect on reported distress (past tense) versus a small or non-significant effect (present tense).

If excluded participants report more intense distress when asked inthe past than present tense, why does this occur?We believe that com-paring excluded participants to included and control participantsin each tense condition can help answer this question. Although thepresent research does not conclusively identify the mechanism, wecan speculate on how participants would behave if their responseswere driven by either implicit theories, coping (Williams' temporal the-ory), or social desirability. First, an implicit theory of emotional dissipa-tion would state that participants' are thinking, “I feel bad now, butfeelings are always more intense when the event is happening so Imust have felt worse in the past.” If this was driving responses to exclu-sion, then present-tense excluded participants would report less dis-tress than past-tense excluded participants but still report significantlymore distress than present-tense control and included participants.Such a pattern is also amenable to a social desirability explanation,with people less willing to report intense personal distress in the pres-ent than the past. Williams' theory would explain this pattern as partic-ipants having entered the reflective stage and thus starting to recover.

But if excluded participants report feeling the same as control partic-ipants when asked in the present tense and only report feeling moredistress when asked in the past tense, both Williams' temporal theoryand an implicit theory explanation would require that participantshave coped with the experience and are honestly reporting a lack of

43A.M. Garczynski, C.M. Brown / Journal of Experimental Social Psychology 53 (2014) 40–50

rejected feelings at the time of measurement (e.g., with an implicit the-ory, participants' experience would be, “I don't feel rejected now, butrejection hurts so I must have felt bad in the past”). However, as men-tioned before, behavioral evidencemakes it unlikely that excluded indi-viduals have truly recovered from their hurt when reporting theirfeelings less than a minute after the exclusion event. Therefore, if sucha pattern is obtained, a social desirability explanation is more likely.

We report the results of two social exclusion studies in which wemanipulated the tense (past or present) in which participants wereasked to report their personal distress (i.e., satisfaction of basic needsand state self-esteem). In the past tense condition, participants wereasked to report how they felt during the exclusion event, whereas thepresent tense condition asked participants to report how they felt atthat moment. The personal distress measures were always adminis-tered immediately (mere seconds) after the exclusion manipulation.

Experiment 1

Method

ParticipantsParticipants were 87 undergraduate students from Saint Louis Uni-

versity. Four participants whowere assigned to play Cyberball acciden-tally exited the program early and were therefore excluded from theanalyses. The final sample consisted of 83 participants. The averageage was 19.02 years, with a standard deviation of 1.24. The samplewas primarily female (72.3%). Amajority of participants ethnically iden-tified as White/Caucasian (62.7%) followed by Asian/Pacific Islander(20.5%), and then Black/African American (7.2%). The remaining partic-ipants (9.6%) were of other ethnicities or did not report race.

ProcedureThe present study used a 3 (Cyberball: exclusion, inclusion, control)

× 2 (tense: past, present) between-participants design. Participantsbegan by completing a baseline measure of Williams and colleagues'basic needs scale (Zadro et al., 2004), with these instructions: “In thenext task, you will be asked to indicate how you currently feel. Foreach question, please select the response that best represents how youfeel AT THIS MOMENT.” Participants then began the manipulation,which was disguised as a mental visualization task. All participantsfirst completed a questionnaire about their ability to mentally visualizesituations; this questionnaire was used to strengthen the cover storyand it was not scored (e.g., Bernstein et al., 2010; Brown, Young,Sacco, Bernstein, & Claypool, 2009).1 Next, participants in the exclusionand inclusion conditions completed a virtual ball toss game with twoother players online, which involved 30 ball tosses and lasted approxi-mately 2 min. However, in reality, the other players were part of thecomputer program (Williams, Cheung, & Choi, 2000). In the exclusioncondition, the participant received the ball twice from the other playersat the start of the game, and then did not receive it again for the remain-der of the game. In the inclusion condition, the participant received one-third of the tosses. In the control condition, participants completed thequestionnaire about their ability to mentally visualize situations, butdid not participate in the virtual ball toss game.

Lastly, all participants completed three measures of self-esteem andwell-being: a post-test administration of the basic needs scale, Leary,Tambor, Terdal, and Downs's (1995) self-esteem scale, and Heathertonand Polivy's (1991) state self-esteem scale. The order in which

1 The instructions read, “We are interested in how mental visualization affects humanbehavior. One example of howmental visualization is used is in sports. A Coach might tella player to mentally visualize the ball going through the hoopwhen shooting a free throwin basketball in order to help the player make the shot. People vary in terms of how theymentally visualize. We want to ask you some questions about how likely and easy it is foryou to mentally visualize.” Participants then completed 9 items asking about their mentalvisualization habits, answered on a 1 (strongly disagree) to 5 (strongly agree) scale (e.g.,“When I read a book, I can see the main characters clearly in my mind”).

participants completed these three scales was randomized. Thesethree scales are all continuous measures of current feelings (i.e., thereare no cut-off scores to classify people into high and low self-esteem).Participants were randomly assigned to complete these dependent var-iable measures in the past tense or the present tense, a between-subjectsmanipulation. Those in the present tense condition read the fol-lowing instructions prior to beginning the measures: “In the next task,you will be asked to indicate how you currently feel. For each question,please select the response that best represents how you feel AT THISMOMENT.” Those in the past tense condition read the following instruc-tions prior to completing these measures: “In the next several question-naires you will be asked to indicate how you felt when you completedthe mental visualization task. Please respond to the following questionsbased on how you felt DURING THE MENTAL VISUALIZATION TASK.”After completing these measures, participants reported their demo-graphic characteristics and were debriefed.

Materials

Basic needs change. Williams and colleagues (Wirth & Williams, 2009;Williams et al., 2000; Zadro et al., 2004) created a 20 item scale tomeasure satisfaction of four basic needs: belongingness, meaningfulexistence, self-esteem, and control, with five items measuring each ofthe basic needs. The scale has been used by Williams and colleaguesto determine whether an event was experienced as ostracizing,both as a manipulation check and as a primary dependent measure(e.g., Gonsalkorale & Williams, 2006; Zadro et al., 2004). In the presentexperiment, basic needs were assessed twice: at the start of the studyas a baseline and following the manipulation. Therefore, basic needschange was computed by subtracting Time 1 (baseline) scores fromTime 2 (post-manipulation) scores, with negative scores indicating adrop in basic needs satisfaction and positive scores indicating an in-crease in basic needs satisfaction. A sample item, with the manipulatedtense in brackets, is “I [feel]/[felt] rejected.” Participants responded toeach item on a scale of 1 (strongly disagree) to 5 (strongly agree), andthe scale was coded such that high scores indicated higher satisfactionof each basic need. This scale demonstrated adequate internal consis-tency in the present study (overallα= .92; past tenseα= .94; presenttense α = .71).

Self-esteem (Leary). Leary's self-esteem scale is a 16 item measure an-swered on a 7-point scale with opposing attributes at each anchor(e.g., confident vs. insecure) and is intended to capture self-evaluativefeelings from a sociometer perspective. Twelve of the items were fromLeary and colleagues' self-esteem measure reported in Leary et al.(1995), and the other four items were added by Kavanagh, Robins,and Ellis (2010) to measure belonging specifically. Participants wereinstructed to respond to the questions regarding how they currentlyfeel (present tense), or how they felt when completing themental visu-alization task (past tense). Typically, the Leary self-esteem scale asks re-spondents to answer in the present tense. Following Kavanagh et al.(2010), we computed the mean of all 16 items to measure self-esteem, with higher scores indicating higher self-esteem (Kavanaghet al., 2010; Leary et al., 1995). This scale demonstrated adequate inter-nal consistency in the present study (α = .97; past tense α = .98;present tense α = .94).

Self-esteem (Heatherton & Polivy). Heatherton and Polivy's (1991) stateself-esteem scale is a 20-item measure which assesses state well-beingon the dimensions of performance self-esteem, social self-esteem, andappearance self-esteem. Participants responded to items on a scale of 1(strongly disagree) to 7 (strongly agree). Items were phrased either inthe past tense or present tense, depending on the study condition,although the original version worded the items in the present tense.A sample item is, “I [feel]/[felt] self-conscious.” An overall meanscore was computed for state self-esteem. This scale demonstrated

1

1.5

2

2.5

3

3.5

4

4.5

5

Exclusion Inclusion Control

Past Present

Fig. 1. Mean basic needs scale (BNS) scores as a function of Cyberball condition andquestion tense.

44 A.M. Garczynski, C.M. Brown / Journal of Experimental Social Psychology 53 (2014) 40–50

adequate internal consistency in the present study (α= .90; past tenseα = .87; present tense α = .91).

Results

We hypothesized that there would be a main effect of Cyberballcondition, such that individuals in the excluded condition would reportsignificantly less basic needs satisfaction and state self-esteem (as mea-sured by Leary andHeatherton's scales), relative to those in the includedand control conditions. More importantly, we hypothesized an interac-tion between Cyberball condition and tense condition, such that partic-ipants in the past tense condition would exhibit the largest differencesacross Cyberball conditions. Specifically, these individuals should reportthe lowest self-evaluations when thinking back on their Cyberball ex-clusion experience, but would report the highest self-worth scoreswhen thinking back on their Cyberball inclusion experience. In termsof the effects of tense, we expected a large difference between pastand present reports among excluded participants but we did not havea specific prediction for how tense would affect responses in the inclu-sion and control conditions. To test the hypotheses, we conducted a 3(Cyberball: exclusion, inclusion, control) × 2 (tense: past, present) anal-ysis of variance (ANOVA) on each of the three measures.

Basic needs change2

Change in scores on the basic needs scale (BNS) was submitted toa 3 × 2 ANOVA.3 Both main effects of Cyberball, F(2,69) = 24.01,p b .001, η2 = .41, and tense, F(1,69) = 31.68 p b .001, η2 = .32,were significant. However, thesemain effects were qualified by a signif-icant interaction, F(2,69) = 15.47, p b .001, η2 = .31 (see Fig. 1).

Simple effect analyses revealed that basic needs dropped significant-ly more in the past tense condition compared to the present tense con-dition for excluded participants, Mdifference = −1.53, SE = .20, p b .001(Mpast = −1.74, SD = .84; Mpresent = − .21, SD = .45), with includedparticipants also showing a significant difference based on tense,Mdifference = − .40, SE = .20, p = .05 (Mpast = − .39, SD = .63;Mpresent = .01, SD = .18), and control participants showing no differ-ence, Mdifference = − .02, p N .9 (Mpast = − .08, SD = .36 l Mpresent =− .05, SD= .22).

When comparing Cyberball conditions within each tense condition,basic needs assessed in the present tense did not differ across thethree Cyberball conditions (ps N .28). In contrast, for the past tense con-dition, excluded participants exhibited a large reduction in basic needsrelative to included (Mdifference=−1.35, SE= .20, p b .001) and control(Mdifference=−1.67, SE= .20, p b .001) participants, who did not signif-icantly differ from each other (p = .12).4

Self-esteem (Leary)Similarly, a 3 (Cyberball condition) × 2 (tense) ANOVA was per-

formed on the Leary state self-esteem scale, which was administeredonly after the Cyberball manipulation (i.e., there was no pre-test mea-sure). Paralleling the results with change in basic needs, there were

2 All variables were normally distributed. Levene's test of equality of variance was non-significant for all cells except for Basic Needs change. It is anticipated that variancewill notbe equally distributed across cells for Basic Needs change as some cells exhibited largerchange over time than others. Additionally, ANOVA is robust to violations of normalitywith large violations rarely influencing the interpretation of results (Gravetter &Wallnau,1999; Kennedy & Bush, 1985).

3 Degrees of freedom are different for the analysis with BNS scores because 8 partici-pants' Time 1 BNS data were not recorded due to experimenter error.

4 When analyzing only Time 2 BNS instead of the change between Time 1 and Time 2,the results are nearly identical. Included and excluded participants both report less BN sat-isfaction in the past tense than the present tense (ps b .01; control participants, p N .54). Inthe past tense, excluded participants differ from both included and control partici-pants (ps b .01). In the present tense, excluded participants don't differ from eithercontrol or included participants (ps N .17), but included participants' report more ba-sic need satisfaction than control participants (p b .05).

main effects of Cyberball condition, F(2,77) = 16.70, p b .001, η2 = .30,and tense, F(1,77) = 40.61, p b .001, η2 = .35, which were qualified bya significant interaction, F(2,77) = 9.18, p b .001, η2 = .19. The resultswith Leary's self-esteem scale mimicked those with the BNS (seeFig. 2). In the exclusion condition, participants' self-esteem was lowerwhen past tense was used (Mpast = 2.69, SD = 1.19; Mpresent = 5.29,SD = .82), Mdifference = −2.61, SE = .38, p b .001. Similarly, includedparticipants reported lower self-esteem in the past tense (Mpast = 4.88,SD = 1.15; Mpresent = 5.96, SD = .94), Mdifference = −1.08, p b .01.Control participants showed no difference (Mpast = 4.98, SD = .96;Mpresent = 5.40, SD = .76), Mdifference =− .41, SE= .38, p N .25.

Described another way, when self-esteemwasmeasured in the pasttense excluded participants had lower self-esteem than included,Mdifference = −2.19, SE = .38, p b .001, and control participants,Mdifference=−2.30, SE= .37, p b .001. Control and included participantsdid not differ, p N .76. When self-esteem was measured in the presenttense, none of the Cyberball conditions significantly differed fromeach other, although excluded participants reported marginally lowerself-esteem than included participants, Mdifference = − .66, SE = .37,p = .08 (for excluded versus control, Mdifference = − .10, SE = .37,p = .13).

Self-esteem (Heatherton & Polivy)A 3 (Cyberball condition) × 2 (tense) ANOVAwas conducted on self-

esteem as measured by the Heatherton and Polivy (1991) scale, whichwas administered after the Cyberball manipulation. The interactionand main effect of tense were not significant (ps N .27). However,there was a main effect of Cyberball condition, F(1,77) = 5.86, p b .01,η2 = .13. LSD post hoc tests indicated that participants in the excludedcondition (M = 4.23, SD = .76) reported lower state self-esteem thanthose in the included (M = 5.01, SD = .91, p b .01) and control condi-tions (M = 4.73, SD = .86, p b .05). There was no difference in stateself-esteem between the included and control participants (p = .21).Although the interaction was non-significant, for exploratory purposessimple main effects tests were conducted on the relationship betweentense and Cyberball condition. When examining the past tense con-dition, excluded participants (M = 4.05, SD = .83) exhibited signifi-cantly lower state self-esteem than both included (M = 4.93, SD =.70; p b .05) and control participants (M = 4.90, SD = .84, ps b .05),who did not differ from each other (p = .93). In the present tensecondition, excluded participants (M= 4.39, SD= .71) had significantlylower state self-esteem than included participants (M = 5.09, SD =1.10, p b .05), and included participants showed marginally greaterstate self-esteem than those in the control condition (M = 4.54, SD =.87, p = .09). There were no differences between excluded and control

5 Several participants met two or more of the exclusion criteria. Thirty-two is the totalnumber of unique participants who did not meet the inclusion criteria.

1

2

3

4

5

6

7

Exclusion Inclusion Control

Past Present

Fig. 2.Mean Leary et al. (1995) state self-esteem scores as a function of Cyberball conditionand question tense.

45A.M. Garczynski, C.M. Brown / Journal of Experimental Social Psychology 53 (2014) 40–50

participants (p = .66). In other words, as seen with both the BNS andLeary self-esteem scale, excluded participants reported feeling signifi-cantly worse than control and included participants when asked in thepast tense. Unlike the BNS and more similar to the Leary self-esteemscale, excluded participants also reported feeling worse than included(but not control) participants in the present tense.

Discussion

Study 1 supported our hypothesis: Excluded participants' self-reported basic needs satisfaction and self-esteem were significantlylower when reported in the past tense relative to the present tense. In-cluded participants also reported lower self-esteem and less basic needssatisfaction in the past tense relative to the present tense. Importantly,excluded participants did not report feeling significantly worse thancontrol and included participants on the BNS when responding in thepresent tense. However, excluded participants did report feeling signif-icantlyworse than included (but not control) participants in the presenttensewhen their feelingswere assessedwith the Heatherton and Polivy(1991) state self-esteemmeasure, an effect that wasmarginal (p= .08)with Leary et al.' (1995) state self-esteem scale. This suggests that stateself-esteemmeasuresmay be better able to distinguish between currentfeelings of personal distress after Cyberball than the BNS. This would re-fute the explanation that excluded participants have already recoveredfrom their feelings of rejection, instead suggesting that the non-significant difference on the BNS is the result of social desirability con-cerns or a less sensitive measure.

Interestingly, participants who were included during Cyberball alsoreported feeling lower self-esteem and less basic needs satisfaction inthe past tense (during the game) than in the present. It may be thatCyberball is not an enjoyable experience even if the individual is beingincluded, making current post-Cyberball feelings more positive thanhow participants remembered feeling during Cyberball. In Study 2, weused a different manipulation of inclusion and exclusion to assesswhether the effect of tense is limited to experiences with Cyberball.

A last point we would like to address about Study 1 is the non-significant interaction on our third measure of personal distress,the Heatherton and Polivy (1991) state self-esteem scale. The Learystate self-esteem measure correlated strongly with change on the BNS(r = .82), whereas correlations with the Heatherton and Polivy stateself-esteem scale were weaker in magnitude (with basic needs change,r = .42, p b .01; with Leary self-esteem scores, r = .63, p b .001). Al-though the interaction was not significant, the simple effect analyseswith theHeatherton and Polivy state self-esteem scale replicated the re-sults of the Leary measure, with excluded participants consistentlyreporting lower self-esteem in the past tense (i.e., significantly different

from both included and control), whereas in the present tense they onlyreported lower self-esteem than included participants (a significant dif-ferent with the Heatherton and Polivymeasure, marginalwith the Learymeasure, andnon-significantwith the BNS). Despite producing differentpairwise comparisons, all three measures of personal distress consis-tently showed a larger main effect of Cyberball condition in the pasttense (BNS, η2 = .54; Leary, η2 = .38; Heatherton & Polivy, η2 = .11)than the present tense (BNS, η2 = .02; Leary, η2 = .05; Heatherton &Polivy, η2 = .06).

Experiment 2

A second experiment was conducted to replicate the finding thatquestion tense and exclusion condition work together to predict basicneeds satisfaction and state self-esteem. Importantly, we used a differ-ent manipulation of social exclusion to test whether the effect of tenseis specific to the Cyberball manipulation.Williams' (2007a, 2007b) tem-poral theory of ostracism is based primarily on his research usingCyberball, making it is possible that tense effects are limited to that ma-nipulation. A very different, yet commonly used, exclusion manipula-tion is to prime feelings of exclusion by having participants writeabout an autobiographical memory of rejection (vs. social acceptanceor a neutral experience; see Brown et al., 2009; Gardner, Pickett, &Brewer, 2000). Given that thismanipulation actually necessitates think-ing about exclusion in the past (whereas Cyberball creates annewexpe-rience), we used three tense conditions: distant past (participants reporthow they felt when initially experiencing the event they wrote about),recent past (participants report how they felt when writing the essayabout their memory), and present tense (participants report how theyfeel at the present moment).

Method

Participants

Data cleaning/quality assurance. Participants were 228 Mechanical Turkusers who completed the study as an online survey. Because thisstudy was completed online, several inclusion criteria were added forquality assurance of the data. The first was an attention check itemthat read, “Select answer 5 for your response,” followed by a 1–7 re-sponse scale. Participants who did not select 5 (indicating they werenot paying attention) were excluded from analyses (n = 19). The sec-ond check was a question that read, “Is there any reason your datashould be excluded from this study? For instance, were you distracted,or did you press buttons randomly?” Participants who answered affir-matively to this question were excluded (n = 6). The third was anitem that asked participants whether English was their first language(the response options were, “Yes,” “No,” and “No, but I am fluent in En-glish”). Given the verbal nature of the manipulation, participants whoindicated English was not their first language were excluded from anal-yses unless they also indicated that they spoke English fluently (n=4).Lastly, participants who did not follow instructions during the essaytask or wrote nonsense answers were excluded (for example, one par-ticipant simply wrote, “I feel great” five times; n = 10).

Final sample. Following data cleaning, 32 participants failed to meetthe inclusion criteria and were therefore excluded from analyses.5 Thefinal sample consisted of 196 participants. The average age was32.53 years, with a standard deviation of 13.23. The samplewas primar-ily female (65.8%). A majority of participants ethnically identified asWhite/Caucasian (69.9%) followed by Asian/Pacific Islander (9.2%),then Black/African American (7.7%), and then Hispanic/Latino (6.6%).The remaining participants (6.6%) were of other ethnicities.

1

2

3

4

5

6

7

Exclusion Inclusion Control

Distant Past Recent Past Present

Fig. 3.Mean basic needs scale (BNS) scores as a function of essay condition and questiontense.

46 A.M. Garczynski, C.M. Brown / Journal of Experimental Social Psychology 53 (2014) 40–50

ProcedureThe present study used a 3 (essay: exclusion, inclusion, control) × 3

(tense: distant past, recent past, present) between-participants design.Participants completed this study online using Amazon's MechanicalTurk (see Buhrmester, Kwang, & Gosling, 2011, for a review of thisplatform). Participants received $0.15 USD as compensation for partici-pation. Participants began by completing the essay manipulation. Theywere randomly assigned to the inclusion, exclusion, or control condi-tion. Participants were provided with the following instructions basedon essay manipulations used in past research (e.g., Brown et al., 2009).The instructions were:

We are interested in people's memories and how people recall de-tails. We would like you to spend a few moments trying to recallas vividly as possible an experience you have had in your past. Spe-cifically, we wish for you to remember an experience during whichyou felt accepted or included [INCLUSION CONDITION]/an experienceduring which you felt rejected or excluded [EXCLUSION CONDITION]/your experiences waking up yesterday morning [CONTROL CONDI-TION]. We don't just want you to remember sensory details but alsohow you felt. In the space below, write as detailed a memory as youcan, describing the event and how it made you feel. Please write atleast one full paragraph about the experience, using as much detailas possible.

Following the essay task, participants completed the BNS and Learyet al.'s (1995) self-esteem scale. The order in which participants com-pleted these scales was randomized. These scales were either wordedin the distant past, recent past, or present. In the distant past condition,the instructions were phrased so that participants indicated how theyfelt during the actual experience from their past. Participants wereinstructed, “Please respond to the following questions based on howyou felt when you initially experienced the event you wrote about.” Inthe recent past condition, participants were instructed, “Please respondto the following questions based on how you felt whilewriting about yourmemory.” The recent past instructions are comparable to the past tensecondition in Study 1 because participants are writing about their affec-tive experience just moments before. In the present tense condition,the instructions read, “Please respond to the following questionsbased on how you feel at this moment.”

Materials

Basic needs. This was the same scale used in Experiment 1 and partici-pants responded on a scale of 1 (strongly disagree) to 7 (stronglyagree). It demonstrated adequate internal consistency in the presentresearch (α = .96; distant past α = .96; recent past α = .96; presentα = .95).

Self-esteem (Leary). The scale was also identical to the one used in Ex-periment 1, and it demonstrated adequate internal consistency in thepresent research (α = .98; distant past α = .98; recent past α = .97;present α = .97).

Results

As in Experiment 1, we hypothesized that there would be amain ef-fect of essay condition, such that individuals in the excluded conditionwould have significantly lower scores on post-manipulation basicneeds satisfaction and state self-esteem, relative to those in the includedor control condition. We also hypothesized an interaction betweenessay condition and tense condition, such that participants in the distantand recent past tense conditions would exhibit the largest differencesacross essay conditions. Specifically, these individuals should reportthe lowest self-evaluationswhen thinking back on their exclusion expe-rience, replicating the findings of the first experiment. To test the

hypotheses, we conducted 3 (essay: exclusion, inclusion, control) × 3(tense: distant past, recent past, present) ANOVAs on the two self-evaluation measures.

Basic needsA 3 (essay condition) × 3 (tense condition) ANOVA was per-

formed on the basic needs scale. Bothmain effects of essay condition,F(2, 187) = 74.18, p b .001, η2 = .44, and tense, F(2, 187) = 10.76,p b .001, η2 = .10, were significant. However, these main effects werequalified by a significant interaction, F(4, 187)= 9.09, p b .001, η2= .16(see Fig. 3).

Simple effect analyses with participants in the excluded conditionrevealed that basic needs were significantly lower for the distant pasttense condition compared to the present tense condition, Mdifference =−2.16, SE = .36, p b .001 (MDistantPast = 2.04, SD = .61; MPresent =4.19, SD= 1.38). Basic needs were also lower in the recent past condi-tion relative to the present tense condition for excluded participants,Mdifference = −1.96, SE = .35, p b .001 (MRecentPast = 2.23, SD = .85;MPresent = 4.19, SD = 1.38). Excluded participants in the distant andrecent past conditions did not differ, Mdifference = .20, SE = .36, p N .59(MDistantPast = 2.04, SD = .61; MRecentPast = 2.23, SD = .85). In otherwords, participants' recollections (past tense) of their basic needs satis-faction were the same for the actual exclusion experience (a distantmemory) as for their experience writing about the memory just mo-ments ago (recent past). However, mere seconds after finishingwritingabout the experience, they reported greater current (present) basicneeds satisfaction when responding in the present tense.

Simple effect analyses with participants in the included conditionrevealed no significant differences in scores based on tense, ps N .18(MDistantPast = 5.36, SD= 1.47; MRecentPast = 5.44, SD = 1.13 MPresent =4.98, SD= .90). Pairwise comparisons among participants in the controlcondition revealed a significant difference between distant past(MDistantPast = 3.74, SD = 1.07) and present tense (MPresent = 4.63,SD = 1.21), Mdifference = − .89, SE = .30, p = .003, as well asbetween distant past and recent past tense (MRecentPast = 4.48, SD =.85),Mdifference =− .74, SE= .30, p b .02. However, there was no differ-ence between recent past and present tenseMdifference=− .15, SE= .30,p N .60. In otherwords, control participants reported feelingworsewak-ing up yesterday morning than they felt currently or when writingabout the morning.

When comparing essay conditions within each tense condition,when basic needswere assessed in the present tense, the only essay con-ditions that differed from one another were the exclusion and inclusionconditions, Mdifference = − .79, SE = .35, p b .03 (the control conditionfell in between the exclusion and inclusion condition and non-

47A.M. Garczynski, C.M. Brown / Journal of Experimental Social Psychology 53 (2014) 40–50

significantly different from both, ps = .18 and .28, respectively). Whenbasic needs were assessed in the distant past or the recent past, allessay conditions (e.g., inclusion, exclusion, control) significantly differedfrom one another (ps b .01).

Self-esteem (Leary)Similarly, a 3 (essay condition) × 3 (tense condition) ANOVA was

performed on responses to the Leary state self-esteem scale. Parallelingthe results with change in basic needs, there were main effects of essaycondition, F(2, 187)= 78.14, p b .001, η2 = .45, and tense, F(2, 187) =12.39, p b .001, η2 = .12, which were qualified by a significant interac-tion, F(4, 187) = 6.52, p b .001, η2 = .12 (see Fig. 4).

Simple effect analyses with participants in the excluded conditionrevealed that self-esteem was higher in the present (MPresent = 4.23,SD = 1.53) than in the distant past (MDistantPast = 1.86, SD = .83),Mdifference = 2.37, SE = .38, p b .001, or recent past (MRecentPast = 2.70,SD = 1.11), Mdifference = 1.53, SE = .37, p b .001. Self-esteem wasalso higher in the recent past than in the distant past, Mdifference = .84,SE = .38, p b .03. Pairwise comparisons with participants in the inclu-sion condition revealed no significant differences across distant past(MDistantPast = 5.70, SD = 1.52), recent past (MRecentPast = 5.66, SD =.87), or present tense (Mpresent = 5.44, SD = .83; ps N .48). Pairwisecomparisons using participants in the control condition revealed signif-icantly lower self-esteem in the distant past (MDistantPast = 3.79, SD =1.30) relative to the present (MPresent = 4.75, SD = 1.10), Mdifference =− .96, SE= .32, p b .01, and marginally lower self-esteem in the distantpast compared to the recent past (MRecentPast = 4.38, SD = 1.04),Mdifference=− .59, SE= .32, p= .07. Therewere no differences betweenpresent and recent past, Mdifference = − .373, SE= .32, p N .24.

When comparing essay conditions within each tense condition,when self-esteem was measured in the distant or recent past tense,all essay conditions were significantly different from one another(ps b .001). When self-esteemwasmeasured in the present tense, par-ticipants in the inclusion condition reported significantly higher self-esteem than participants in the exclusion (Mdifference = 1.21, SE = .37,p b .01) and control conditions (Mdifference = .69, SE = .35, p b .05).Therewere no significant differences between the exclusion and controlconditions, Mdifference = − .52, SE = .35, p = .13. See Fig. 4 for a visualrepresentation of the Study 2 results regarding Leary's self-esteemmeasure.

Discussion

Experiment 2 replicated Experiment 1's finding that social exclusionproduced lower basic needs satisfaction and self-esteem when feelings

1

2

3

4

5

6

7

Exclusion Inclusion Control

Distant Past Recent Past Present

Fig. 4. Mean Leary et al. (1995) state self-esteem scores as a function of essay conditionand question tense.

were measured in the past tense relative to the present tense. Impor-tantly, this was found using a differentmanipulation of social exclusion.With thismanipulation, therewere two types of past tense: Howpartic-ipants felt during the exclusion experience itself and how they feltwhenwriting about the experience just moments before. With the Cyberballmanipulation, participants in the past tense condition reported theirfeelings just moments before, which is also when the exclusion eventoccurred. Interestingly, excluded participants in Experiment 2 reportedfeeling worse in the past than the present regardless of whether thatpast was distant or just moments before. Excluded participants' basicneeds satisfaction was equally low in the distant and recent past condi-tions, but their self-esteem became increasingly negative as the refer-ence point was further in the past. Although non-significant, includedparticipants felt better in both past tense conditions than in the present,and control participants regularly reported feeling worse in the distantpast than in the other two tense conditions (although this is under-standable, as the distant past was when they woke up the previousmorning, which is an unpleasant time for many people). These patternsin the included and control conditions were the same regardless of themeasure used (basic needs or self-esteem).

Critically, in Experiment 1 excluded participants' self-reportedfeelings did not significantly differ from those of included and controlparticipants when the questions were worded in the present tenseand when the measure was the BNS (although they were marginallydifferent from included participants on the Leary self-esteem measureand significantly different on the Heatherton & Polivy measure). InExperiment 2, excluded participants reported significantly less self-esteem and basic needs satisfaction than included (but not control)participantswhen reporting their feelings in the present tense. This sug-gests that the overall effect of more intense feelings of exclusion report-ed in the past than the present is true for both exclusionmanipulations,whereas the finding that using present-tensewording eliminates exclu-sion effects appears to be limited to the combination of Cyberball andthe BNS.

General discussion

The present study tested the idea that the tense phrasing usedwith self-report personal distress measures shapes responses to socialexclusion. Experiment 1 manipulated social exclusion using the virtualostracism Cyberball paradigm, while Experiment 2 utilized an autobio-graphical memory task. In both experiments participants reported theirbasic needs satisfaction and self-esteem, and we manipulated whetherthese questions were asked in the past tense (referring to how theyfelt during the exclusion experience) or the present tense (referring tohow they feel while completing the measure). For comparison, we in-cluded social inclusion and control conditions.

In Experiment 1, excluded participants reported significantly lowerbasic needs and state self-esteem in the past tense, relative to thepresent tense. In Experiment 2, excluded participants demonstratedlower basic needs and state self-esteem in distant past and recentpast, relative to present tense. Temporally, the present and recent pastare actually closer to each other, yet participants' responses weremore similar in the distant and recent past conditions. In other words,participants report remembering feeling personal distresswhenwritingabout an exclusion experience, but claim to be feeling less personal dis-tress justmoments later. Despite this, their current personal distress stillsignificantly differed from that reported by included participants. Thismakes it clear that excluded participants are still feeling the impact ofremembered exclusion mere seconds after writing about the memory,although they report those current feelings as being less intense thantheir feelings moments before and their feelings during the event itself.This could represent a response bias (i.e., exaggerating past feelings orunderreporting present distress) or actual emotional dissipation thathas occurred over a span of just a few seconds.

48 A.M. Garczynski, C.M. Brown / Journal of Experimental Social Psychology 53 (2014) 40–50

The effect of temporal framing in the control and inclusion conditions

We conducted this research to understand temporal framing on self-reported responses to social exclusion, butwe also found some effects oftense in the control and inclusion conditions. In Experiment 1, includedparticipants reportedmarginally lower basic needs satisfaction and sig-nificantly lower state self-esteem (using Leary et al.'s, 1995, scale) in thepast tense relative to the present tense; control participants showed notense effects on any measure. In Experiment 2, included participantsshowed no difference in basic needs or state self-esteem across tense;control participants exhibited lower basic needs satisfaction and self-esteem in the distant past relative to the present past and present tense.

The inconsistent findings of the effects of tense on the inclusion andcontrol conditions across experiments could be a function of themanip-ulations themselves. For instance, theCyberball control condition essen-tially included no manipulation at all—control participants onlycompleted the samemeaningless mental visualization survey as partic-ipants in the other condition, but without playing the ball toss game—which may be a truer control condition than the autobiographicalmemory task. The autobiographical memory control condition askedparticipants to write about their experiences waking up yesterdaymorning. A reading of the essays indicated that many people's experi-ences of waking up in the morning were mildly negative, as people re-ported feeling tired and having difficulty waking. Therefore, the effectof tense, with present and recent past conditions being similar to eachother and different from the distant past, can be explained by people ac-curately remembering that they felt less pleasant in the morning thanthey feel currently or felt a fewmoments ago. However, an implicit the-ory explanation—that people overestimate the intensity of their pastemotions—may also apply to this effect. Another point that should benoted is that the basic needs and self-esteemmeasures aremeant to re-flect affect caused by social experiences, but the results of the controlessay condition suggest that they can capture general affective statesas well. Lastly, we find it interesting that people report feeling upsetwhile reliving exclusion (the recent past condition), but they do not re-port feeling upset while reliving the apparently negative experience ofwaking up the previous morning.

The inclusion condition also exhibited inconsistent findings acrossthe twomanipulations. Tense did not affect responseswith the autobio-graphical memory manipulation, but with Cyberball inclusion partici-pants reported lower state self-esteem and marginally lower basicneeds in the past tense. It is also important to note that, relative to thecontrol condition, inclusion boosted basic needs and state self-esteemvia thememorymanipulation, but not through the Cyberball manipula-tion. Itmay be that the playing a game of Cyberball is unpleasant even inthe inclusion condition. Because the task involves long delays betweentosses, even individuals in the inclusion condition may experiencesome negative affect (i.e., boredom) during the activity. A majorityof past Cyberball research has not included a no-Cyberball controlcondition (see Brown et al., 2009, for an exception) so the experienceof playing Cyberball—whether excluded or included—is not well under-stood. Therefore, future research should further explore the experienceof inclusion in Cyberball relative to not playing the game.

The mechanism of temporal framing effects

It is possible that comparing the effect of question tense on excludedparticipants' reported feelings relative to participants in social inclusionand control conditions can shed light onwhich of the threemechanismsis likely to be responsible: Williams' theory that non-significant differ-ences between excluded and included participants represent actual re-covery from exclusion (an emotion dissipation explanation), a socialdesirability explanation, or an implicit theory explanation. In Experi-ment 1, there were no significant differences on the BNS measure be-tween the control, inclusion, and exclusion conditions when phrasedin the present tense. However, the present tense condition did produce

a significant difference between excluded and included participants ontheHeatherton andPolivy (1991)measure and amarginal (p=.08) dif-ference on the Leary et al. (1995) measure. In addition, in Experiment 2excluded participants reported lower basic needs satisfaction and self-esteem in the present tense than did included participants. These find-ings, combined with other research showing that excluded participantsstill show physiological and behavioral effects of their rejectionminutesafter the manipulation exclusion (e.g., Stroud et al., 2000; Zhong &Leonardelli, 2008; Zwolinski, 2012), lead us to believe that the null ef-fect of Cyberball condition on BNS scores does not represent excludedparticipants' recovery from the rejection experience just momentsbefore. Therefore, when participants who were excluded in Cyberballreport that their basic needs are currently just as satisfied as participantswho were included (as they did in Experiment 1), it seems likely thatthis is the result of a response bias rather than indicating actualrecovery.

As for the type of response bias—either underreporting currentfeelings or overestimating past feelings—this is more challenging toestablish. An implicit theory explanation is that excluded partici-pants who are currently feeling distress believe that they musthave felt worse during the actual event, hence they exaggeratetheir past memories relative to the present. In contrast, a social desir-ability explanation is that excluded participants are comfortableacknowledging that they once felt distressed, but do not want toadmit they are currently feeling intense distress. The implicit theoryexplanation could potentially be refuted if excluded participants' cur-rent feelings did not differ from that of participants in the other condi-tions (i.e., excluded participants must feel some current distress inorder for an implicit theory to exaggerate their memory of past dis-tress). Because excluded participants did differ from included partici-pants on some present tense measures (both self-esteem and basicneeds in Experiment 2, and significant or marginal effects on the twoself-esteem measures in Experiment 1), an implicit theory explanationis still tenable. In fact, it is quite possible that both response bias expla-nations are true.

However, evidence from other researchers lends support for a so-cial desirability interpretation. Eisenberger et al. (2003) had partici-pants experience exclusion through Cyberball while activity in theiranterior cingulate cortex (ACC) was measured; afterwards, partici-pants completed self-report questionnaires about their feelings of ex-clusion and social distress during the exclusion event (past tense). TheACC serves as a neural alarm system for pain, and activity in thisarea during the exclusion experience was strongly correlated withself-reported memory for the distress they felt during the experience(a past tense measure). This convergence with a neural measure sug-gests that participants' reports of their past feelings are unlikely to be bi-ased, and instead the effect of tense is most likely the result of biased(underreported) current feelings due to social desirability concerns. Asecond piece of converging evidence comes from Williams et al.(2000), in which participants completed the study from their ownhome and were included or excluded in Cyberball via the Internet; inthis study, which is one of the few to do so, participants completedthe BNS in the present tense. These researchers did observe significantdifferences in self-reported current distress, and we believe that thiscan be explained by participants completing both Cyberball and theBNS in the privacy of their own home, where self-report concernsshould be at a minimum.

Of course, our Experiment 2 participants also completed the studyfrom their own homes, and it is also in this study where we see a differ-ence between included and excluded participants on the BNS in thepresent tense. Therefore, we believe that non-significant BNS differ-ences between participantswhowere included or excluded in Cyberballcan be explained by social desirability concerns. However, it remainspossible that implicit theories of emotion dissipation over time also con-tribute to participants reporting that their affect a few moments agowas more intense than it is now.

49A.M. Garczynski, C.M. Brown / Journal of Experimental Social Psychology 53 (2014) 40–50

Implications, limitations, and future directions

The present study has several implications for future research. First,it suggests that minor changes to the procedure of a study (such as thetense of dependent variable measures) can have large effects on thefindings of that experiment. This has implications for at least two stud-ies that examined effects of social exclusion on responses to the BNSover time while confounding tense (e.g., Goodwin et al., 2010; Zadroet al., 2006). To elaborate, Zadro and colleagues administered the BNSin the past tense immediately following rejection and then adminis-tered it again in the present tensemeasure 45min later; they concludedthat differences in responses at the two time points reflect differentstages of recovery, yet time was confounded with tense. Goodwin andcolleagues similarly confounded tense and time; their participants firstcompleted a past tensemeasure of distress (intended tomeasure reflex-ive responses) administered immediately after an exclusion experience,and later they completed a present tense measure meant to capturetheir reflective response (in between these two measures was the op-portunity to appraise the exclusion event, which was intended toshape how participants coped in the reflective stage). The purpose ofthese experiments was to establish moderators of recovery from exclu-sion over time. In both studies, a non-significant difference betweenexclusion and inclusion participants on the delayed present tense mea-sure was interpreted as evidence that excluded participants had recov-ered from their rejection. Both of these studies used the Cyberballmanipulation and BNS, and in this paper we found that this particularcombination of manipulation and measure produces non-significantdifferences in the present tense but significant differences in the pasttense. Importantly, wewant to emphasize that themoderators of recov-ery from exclusion identified in these papers (e.g., social anxiety inZadro et al., 2006) may still be true moderators. However, the magni-tude of their effect may be overestimated by the confound of time andtense. In addition, inconsistencies in question tense might also help ex-plain why social exclusion sometimes lowers self-esteem and mood(Gerber & Wheeler, 2009) while producing null effects in other studies(Blackhart et al., 2009).

Although the present research provides evidence that the temporalframing of self-report measures moderates responses to social exclu-sion, it is not without limitations. Our research highlights how self-report measures can be biased, but the very nature of this bias makesit challenging to conclude the direction, cause, and even degree ofbias. Future research should assess physiological indicators of distress,such as cortisol levels or blood pressure, both during the exclusionevent and when participants self-report their feelings. If self-reports inthe past or present tense differentially correlatewith physiological reac-tions during the event and self-report, this could shedmore light on thepresence or direction of self-report bias. A second method to shed lighton the possible mechanism of the tense effects could be to measure ormanipulate participants' self-presentation concerns and beliefs aboutemotional intensity across time. For instance, participants could beinstructed that lower self-esteem and basic needs will remain intenseafter exclusion or that they should expect to have a short-lived andslight negative response to social exclusion. Another interesting direc-tion for future researchwould be to ask participants to report their feel-ings both immediately and after a delay while counterbalancing thetense of the measures (i.e., present tense immediate and past tensedelay, and vice versa). If tense influences responses more than timing,this would be particularly persuasive evidence that the effect of tenseis caused by a response bias.6 We suspect that both social desirabilityconcerns and implicit theories can contribute to temporal framing ef-fects, although we think social desirability exerts a stronger influence.

An intriguing direction for future research is whether the actof reporting one's personal distress after exclusion could produce a

6 We thank an anonymous reviewer for this suggestion.

self-perception effect (e.g., Bem, 1967) that shapes actual recoveryfrom rejection. To explain, it's possible that excluded participants whounderreport their current distress for self-presentation reasons maycome to believe that they truly do feel unaffected by the exclusion.This could be tested by replicating the present studies and assessing be-havioral or physiological indicators of distress following self-reports. Ifbiased self-reporting shapes actual recovery from rejection, excludedparticipants who are asked to consider their feelings in the presenttense should show fewer behavioral and physiological effects of rejec-tion than excluded participants who report their past feelings.

Lastly, it would be meaningful to explore the role of other relevantpersonality factors. As mentioned earlier, our results challenge the in-terpretation of Zadro et al.'s (2006) study showing that social anxietydid not moderate immediate responses to exclusion but did moderateresponses 45 min later. It would be interesting to replicate their studywhile removing the confound of question tense to examinewhether so-cial anxiety moderates responses to social exclusion in the presenttense. If personality does moderate immediate self-reports in the pres-ent tense, then itmay bemeasuring real recovery (i.e., non-anxious peo-ple recover faster), although it's also possible that socially anxiousindividuals differ in their self-presentation concerns and implicittheories of emotion dissipation as opposed to reflecting true differencesin emotion dissipation. Moment-to-moment measures of affect (e.g.,Wesselmann et al., 2012) may help disentangle these competingexplanations.

Conclusions

This research identifies temporal framing as a crucial moderator ofself-reported personal distress following social exclusion. Temporalframing could affect responses due to actual recovery from exclusion,socially desirability concerns, or implicit theories about how emotionschange over time. Our results and data from other researchers suggestthat the latter two explanations aremore plausible than actual emotion-al recovery. This research is important because self-reported feelings ofbelonging and self-esteem are often used as the sole determinant ofwhether an event was experienced as rejection and whether a personhas recovered from rejection (e.g., Bernstein et al., 2010; Gonsalkorale& Williams, 2006; Goodwin et al., 2010; Wirth et al., 2010; Zadroet al., 2006). We encourage other researchers to make use of this infor-mation when exploring the phenomenon of social exclusion in thefuture.

References

Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal at-tachments as a fundamental human motivation. Psychological Bulletin, 117, 497–529.

Baumeister, R. F., Twenge, J. F., & Nuss, C. (2002). Effects of social exclusion on cognitiveprocesses: Anticipated aloneness reduces intelligent thought. Journal of Personalityand Social Psychology, 83, 817–827. http://dx.doi.org/10.1037//0022-3514.83.4.817.

Bem, D. J. (1967). Self-perception: An alternative interpretation of cognitive dissonancephenomena. Psychological Review, 74, 183–200.

Bernstein, M. J., Sacco, D. F., Young, S. G., Cook, E., & Hugenberg, K. (2010). Being ‘in’ withthe in crowd: The effects of social exclusion and inclusion are enhanced by sharedingroup status. Personality and Social Psychology Bulletin, 36, 999–1009.

Blackhart, G. C., Eckel, L. A., & Tice, D. M. (2007). Salivary cortisol in response to acute so-cial rejection and acceptance by peers. Biological Psychology, 75, 267–276. http://dx.doi.org/10.1016/j.biopsycho.2007.03.005.

Blackhart, G. C., Nelson, B. C., Knowles, M. L., & Baumeister, R. F. (2009). Rejection elicitsemotional reactions but neither causes immediate distress nor lowers self-esteem:A meta-analytic review of 192 studies of social exclusion. Personality and SocialPsychology Review, 13, 269–309.

Brown, C. M., Young, S. G., Sacco, D. F., Bernstein, M. J., & Claypool, H. M. (2009). Social in-clusion facilitates interest in mating. Evolutionary Psychology, 7, 11–27.

Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon's Mechanical Turk: A newsource of inexpensive, yet high-quality data? Perspectives on Psychological Science, 6,3–5. http://dx.doi.org/10.1177/1745691610393980.

Crowne, D. P., &Marlowe, D. (1960). A new scale of social desirability independent of psy-chopathology. Journal of Consulting Psychology, 24, 349–354.

Eisenberger, N. I., Lieberman, M. D., &Williams, K. D. (2003). Does rejection hurt? An fMRIstudy of social exclusion. Science, 302, 290–292.

50 A.M. Garczynski, C.M. Brown / Journal of Experimental Social Psychology 53 (2014) 40–50

Fisher, R. J., & Dubé, L. (2005). Gender differences in responses to emotional advertising: Asocial desirability perspective. Journal of Consumer Research, 31, 850–858.

Gardner, W. L., Pickett, C. L., & Brewer, M. B. (2000). Social exclusion and selective mem-ory: How the need to belong influences memory for social events. Personality andSocial Psychology Bulletin, 26, 486–496.

Gerber, J., & Wheeler, L. (2009). On being rejected: A meta-analysis of experimental re-search on rejection. Perspectives on Psychological Science, 4, 468–488.

Gonsalkorale, K., &Williams, K. D. (2006). The KKKwon't let me play: Ostracism even by adespised outgroup hurts. European Journal of Social Psychology, 37, 1176–1186. http://dx.doi.org/10.1002/ejsp.392.

Goodwin, S. A., Williams, K. D., & Carter-Sowell, A. R. (2010). The psychological sting ofstigma: The costs of attribution ostracism to racism. Journal of Experimental SocialPsychology, 46, 612–618.

Gravetter, F. J., & Wallnau, L. B. (1999). Essentials of statistics for the behavioral sciences(3rd ed.). Pacific Grove, CA: Brooks/Cole.

Hart, W. (2012). Unlocking past emotion: Verb use affects mood and happiness.Psychological Science. http://dx.doi.org/10.1177/0956797612446351.

Heatherton, T. F., & Polivy, J. (1991). Development and validation of a scale for measuringstate self-esteem. Journal of Personality and Social Psychology, 60, 895–910.

Kanten, A. B., & Teigen, K. H. (2008). Better than average and better with time: Relativeevaluations of self and others in the past, present, and future. European Journal ofSocial Psychology, 38, 343–353.

Karniol, R., & Ross, M. (1996). The motivational impact of temporal focus: Thinking aboutthe future and the past. Annual Review of Psychology, 47, 593–620.

Kavanagh, P. S., Robins, S. C., & Ellis, B. J. (2010). The mating sociometer: A regulatorymechanism for mating aspirations. Journal of Personality and Social Psychology, 99,120–132. http://dx.doi.org/10.1037/a0018188.

Kennedy, J. J., & Bush, A. J. (1985). An introduction to the design and analysis of experimentsin behavioral research. Lanham, MD: University Press of America.

Kenrick, D. T., Griskevicius, V., Neuberg, S. L., & Schaller, M. (2010). Renovating the pyra-mid of needs: Contemporary extensions built upon ancient foundations. Perspectiveson Psychological Science, 5, 292–314.

Leary, M. R., Tambor, E. S., Terdal, S. K., & Downs, D. L. (1995). Self-esteem as an interper-sonal monitor: The sociometer hypothesis. Journal of Personality and Social Psychology,68, 518–530.

Mayer, J. D., & Gaschke, Y. N. (1988). The experience and meta-experience of mood.Journal of Personality and Social Psychology, 55, 102–111.

Schwarz, N. (1999). Self-reports: How the questions shape the answers. AmericanPsychologist, 54, 93–105.

Schwarz, N. (2007). Retrospective and concurrent self-reports: The rationale for real timedata capture. In A. A. Stone, S. S. Shiffman, A. Atienza, & L. Nebeling (Eds.), The scienceof real-time data capture: Self-reports in health research (pp. 11–26). New York: OxfordUniversity Press.

Smart Richman, L., & Leary, M. (2009). Reactions to discrimination, stigmatization, ostra-cism, and other forms of interpersonal rejection: A dynamic, multi-motive model.Psychological Review, 116, 365–383.

Smith, R. E., Leffingwell, T. R., & Ptacek, J. T. (1999). Can people remember how theycoped? Factors associated with discordance between same-day and retrospective re-ports. Journal of Personality and Social Psychology, 76, 1050–1061.

Stroud, L. R., Tanofsky-Kraff, M., Wilfley, D. E., & Salovey, P. (2000). The Yale InterpersonalStressor (YIPS): Affective, physiological, and behavioral responses to a novel interper-sonal rejection paradigm. Annuals of Behavioural Medicine, 22, 204–213.

Thomas, D. L., & Diener, E. (1990). Memory accuracy in the recall of emotions. Journal ofPersonality and Social Psychology, 59, 291–297.

Twenge, J. M., Baumeister, R. F., DeWall, C. N., Ciarocco, N. J., & Bartels, J. M. (2007). Socialexclusion decreases prosocial behavior. Journal of Personality and Social Psychology,92, 56–66.

Weik, U., Maroof, P., Zöller, C., & Deinzer, R. (2010). Pre-experience of social exclusionsuppresses cortisol response to psychosocial stress in women but not in men.Hormonal Behavior, 58, 891–897. http://dx.doi.org/10.1016/j.yhbeh.2010.08.018.

Wesselmann, E. D., Wirth, J. H., Mroczeka, D. K., & Williams, K. D. (2012). Dial a feeling:Detecting moderation of affect decline during ostracism. Personality and IndividualDifferences, 53, 580–586. http://dx.doi.org/10.1016/j.paid.2012.04.039.

Williams, K. D. (2001). Ostracism: The power of silence. New York: Guilford (282 pp.).Williams, K. D. (2007). Ostracism. Annual Review of Psychology, 58, 425–452. http://

dx.doi.org/10.1146/annurev.psych.58.110405.085641.Williams, K. D. (2007). Ostracism: The kiss of social death. Social and Personality

Psychology Compass, 1, 236–247.Williams, K. D., Cheung, C. K. T., & Choi, W. (2000). Cyberostracism: Effects of being ig-

nored over the Internet. Journal of Personality and Social Psychology, 79, 748–762.Wirth, J. H., Sacco, D. F., Hugenberg, K., & Williams, K. D. (2010). Eye gaze as relational

evaluation: Averted eye gaze leads to feelings of ostracism and relational devaluation.Personality and Social Psychology Bulletin, 36, 869–882.

Wirth, J. H., & Williams, K. D. (2009). “They don't like our kind”: Consequences of beingostracized while possessing a group membership. Group Processes: IntergroupRelations, 12, 111–127.

Zadro, L., Boland, C., & Richardson, R. (2006). How long does it last? The persistence of theeffects of ostracism in the socially anxious. Journal of Experimental Social Psychology,42, 692–697.

Zadro, L.,Williams, K. D., & Richardson, R. (2004). How low can you go? Ostracism by a com-puter is sufficient to lower self-reported levels of belonging, control, self-esteem, andmeaningful existence. Journal of Experimental Social Psychology, 40, 560–567.

Zhong, C. B., & Leonardelli, G. J. (2008). Cold and lonely: Does social exclusion literally feelcold? Psychological Science, 19, 838–842.

Zöller, C., Maroof, P., Weik, U., & Deinzer, R. (2010). No effect of social exclusion onsalivary cortisol secretion in women in a randomized controlled study.Psychoneuroendocrinology, 35, 1294–1298. http://dx.doi.org/10.1016/j.psyneuen.2010.02.019.

Zwolinski, J. (2012). Psychological and neuroendocrine reactivity to ostracism. AggressiveBehavior, 38, 108–125. http://dx.doi.org/10.1002/ab.21411.