When apologies work: How matching apology components to victimsв€™ self-construals facilitates ...

14 Pages·2010·0.36 MB·English

by  Ryan Fehr

OrganizationalBehaviorandHumanDecisionProcesses113(2010)37–50
ContentslistsavailableatScienceDirect
Organizational Behavior and Human Decision Processes
journal homepage: www.elsevier.com/locate/obhdp
When apologies work: How matching apology components to victims’
self-construals facilitates forgiveness
Ryan Fehr*, Michele J. Gelfand
DepartmentofPsychology,1147Biology–PsychologyBuilding,UniversityofMaryland,CollegePark,MD20742,UnitedStates
a r t i c l e i n f o a b s t r a c t
Articlehistory: Apologiesareusefulsocialtoolsthatcanactascatalystsintheresolutionofconflictandinspireforgive-
Received9January2009 ness.Yetasnumerousreal-worldblundersattest,apologiesarenotalwayseffective.Whereasmanylead
Accepted7April2010 toforgivenessandreconciliation,otherssimplyfallondeafears.Despitethefactthatapologiesdifferin
Availableonline13May2010
theireffectiveness,mostresearchhasfocusedonapologiesasdichotomousphenomenawhereinavictim
either(a)receivesanapologyor(b)doesnot.Psychologicalresearchhasyettoelucidatewhichcompo-
AcceptedbyPaulLevy
nentsofapologiesaremosteffective,andforwhom.Thepresentresearchbeginstoaddressthisgapby
testingthetheorythatperpetrators’apologiesaremostlikelytoinspirevictimforgivenesswhentheir
Keywords:
components align with victims’ self-construals. Regression and hierarchical linear modeling analyses
Apology
fromtwostudiessupporttheprimaryhypotheses.Aspredicted,victimsreactedmostpositivelytoapol-
Self-construal
Forgiveness ogiesthatwerecongruentwiththeirself-construals.
Socialaccounts (cid:2)2010ElsevierInc.Allrightsreserved.
Introduction improve customer experiences, and enhance leader effectiveness
(Liao, 2007; Tomlinson, Dineen, & Lewicki, 2004; Tucker, Turner,
Conflict is a ubiquitous social phenomenon that transcends Barling,Reid,&Elving,2006).Nonetheless,scholarsandlaypeople
people,cultures,andcontexts(DeDreu&Gelfand,2009).Although alikehaverecognizedthepotentialforapologiestofail.Apologies
conflictisinevitable,thequestionofhowconflictismanagedholds havebeenreferredtoas‘‘Highlyriskystrategies...[that]canmake
criticalimplicationsforitsconsequences.Inaworldofglobaloppor- a bad situation worse” (Kellogg, 2007, p. 21). Empirical research
tunitiesandglobalthreats,thewayconflictismanagedcanhave similarlysupportstheideathatapologiesarenotalwayseffective
such diverse consequences as escalation and war or forgiveness (Skarlicki,Folger,&Gee,2004),drawingattentiontothequestion,
and peace. Fortunately, a sizable literature in social psychology ‘‘Whyisitthatsomeapologiessucceedwhereasothersfallondeaf
andorganizationalbehaviorhasdevelopedtounderstandandex- ears?”.
plaintheconditionsunderwhichconflictleadstodestructiveversus Centraltothequestionofwhysomeapologiessucceedwhere
productiveoutcomes(DeDreu,2006;DeDreu&VandeVliert,1997; othersfailisarecognitionthatallapologiesarenotcreatedequal.
DeDreu&Weingart,2003;Peterson&Behfar,2003). Rather,apologiescancontaindifferentsetsofelements–referred
Onefactorthathasbeenarguedtoplayavitalroleinhowconflict tohereascomponents–thatmayaffecthowvictimsreacttothem.
is managed by offenders is apology (Darby & Schlenker, 1982; Forinstance,someapologiesmightfocusonthecompensationofa
Ohbuchi,Kameda,&Agarie,1989;Sitkin&Bies,1993).Intheintro- victim while others might focus on showing empathy. Apology
ductiontohislandmarkbookOnApology,psychiatristAaronLazare components have received theoretical attention in Sociology
opensbyreferringtoapologyas‘‘Oneofthemostprofoundhuman (Tavuchis,1991),Law(Wagatsuma&Rosett,1986),andPsychiatry
interactions”(Lazare,2004,p.1).Onanationalscale,apologieshave (Lazare,2004),yetwithfewexceptions(Darby&Schlenker,1982;
helpedtohealthewoundsoftheHolocaust,theNankingMassacre, Santelli, Struthers, & Eaton, 2009; Schmitt, Gollwitzer, Forster, &
andmanyotheratrocities(Brooks,1999).Amongindividuals,apol- Montada, 2004) they have received scant empirical attention in
ogieshavehelpedtorepaircountlessrelationshipsandrestorehar- Psychology.Rather,researchinPsychologyhastraditionallyexam-
mony(Scher&Darley,1997;Tavuchis,1991).Withinorganizational inedapologyasastrictlydichotomousphenomenonwhereinavic-
contexts,apologiescanbeusedtoresolveinterpersonaldisputes, tim either (a) receives an apology or (b) does not. For instance,
Brown, Wohl, and Exline (2008) examined apology effectiveness
byassigningparticipantsto‘‘apology”versus‘‘noapology”condi-
tions.Similarly,Liao(2007)testedforthepresenceofapologyfol-
* Correspondingauthor.
lowing a customer service failure by asking respondents if an
E-mail addresses: [email protected] (R. Fehr), [email protected]
(M.J.Gelfand). apology was or was not received (see also Frantz & Bennigson,
0749-5978/$-seefrontmatter(cid:2)2010ElsevierInc.Allrightsreserved.
doi:10.1016/j.obhdp.2010.04.002
38 R.Fehr,M.J.Gelfand/OrganizationalBehaviorandHumanDecisionProcesses113(2010)37–50
2005; McCullough et al., 1998; Ohbuchi et al., 1989; Struthers, socialsciences(Avruch&Wang,2005;Cunningham,2004;Goffman,
Eaton, Santelli, Uchiyama, & Shirvani, 2008; Tomlinson et al., 1967; Govier & Verwoerd, 2002; James, 2006; Lazare, 2004;
2004).Intheseandotherstudies,theprototypicalmethodologyis Scher&Darley,1997;Tavuchis,1991). AcrosssuchfieldsasLaw,
tocontrastasimple‘‘I’msorry”againsta‘‘noapology”condition, Sociology,andPsychiatry,scholarshavefocusedonthreeapology
rather than to consider the implications of the specific apology componentsthatareparticularlyrelevanttothecurrentresearch:
componentsthatvictimsreceive.Thistendencytodichotomizeis offersofcompensation,expressionsofempathy,andacknowledg-
systemicnotonlyforapologiesbutalsoforjustifications,denials, ments of violated rules/norms. Despite considerable consensus
and related social accounts. For instance, in their meta-analysis regardingthesecomponentsacrossfields,fewattemptshavebeen
on the effectiveness of justifications versus excuses, Shaw, Wild, made to integrate the relevant research. To this end, a review of
and Colquitt (2003) model the contexts surrounding social ac- thesecomponentsispresentedindetailbelow.
counts, but do not codify or model the content of the accounts Apologiesasoffersofcompensationarefocusedontherestora-
themselves. Thus, dichotomization appears to transcend research tionofequitythroughexchange.Thatis,theyarefocusedoncor-
onsocialaccounts. rectingthebalanceofarelationshipthroughsometypeofaction,
The literature’s limited conceptualization of apologies as eitherspecificorgeneral.Forinstance,offenderscanoffertopro-
dichotomous is problematic for a number of reasons. First, this vide their victims with specific, tangible goods (e.g. ‘‘[I] could go
conceptualizationistoobroadandatheoreticaltoallowforapre- and see if I can get you another...”; Schmitt et al., 2004, p. 470)
ciseunderstandingofwhyapologieswork.Second,adichotomous oroffermoregenerallytotakewhateveractionisneeded(e.g.‘‘If
viewfailstoconsiderthefactthatapologiesareofferedtospecific there is any way I can make it up to you please let me know.”;
victimswholikelydifferintermsofwhattheyexpecttohear,and Scher & Darley, 1997, p. 132). In many qualitative studies from
thusdoesnotintegratevictimpsychologicalstatesintotheapol- Law, Sociology, and Psychology, compensation is mentioned as a
ogy process. Third, the dichotomous approach to apology limits vital component of the apology process (Goffman, 1967; Lazare,
the specificity with which apology interventions can be recom- 2004;O’Hara&Yarn,2002;Schlenker&Darby,1981;Wagatsuma
mendedinresponsetoconflict.Aconsiderationofapologycompo- & Rosett, 1986). A number of quantitative studies support this
nents, for instance, could allow managers, conflict mediators, claim. Offers of compensation have been shown to relate to vic-
spouses,parents,andotheroffenderstotargettheirapologieswith tims’ impressions of their offenders, impressions of the conflict,
meaningful statements such as expressions of empathy or andemotionalstates(Blum-Kulka,House,&Kasper,1989;Conlon
acknowledgmentsofviolatednorms. & Murray, 1996; Scher & Darley, 1997; Schmitt et al., 2004). The
In examiningthe psychology of apologycomponents,the cur- importanceofoffersofcompensationcanbesummedupbyBishop
rentresearchfocusesontheintegralroleofvictims’self-construals DesmundTutu,whooncenotedthat‘‘Ifyoutakemypenandsay
inthelinkbetweenapologiesandforgiveness.Brieflydefined,self- you are sorry, but don’t give me the pen back, nothing has hap-
construal relates to how individuals perceive their relationships pened” (Tutu, 2004). The importance of offers of compensation
withotherpeople.Recentconceptualizationsofself-construalhave canbeextendedtoorganizationalcontextsaswell,wherecompen-
described a tripartite model, consisting of the independent, rela- sation is frequently offered as a form of apology to alleviate the
tional, and collective selves (Kashima et al., 1995). Self-construal negative effects of organizational injustice (Okimoto & Tyler,
hasbeenimplicatedinawiderangeofcognitions,emotions,and 2007).Itshouldbenotedthatalthoughoffersofcompensationpro-
behaviors (e.g. Brewer & Gardner, 1996; Cross, Gore, & Morris, totypically reference tangible goods, socio-emotional compensa-
2003; Cross & Madson, 1997; Johnson & Chang, 2006; Johnson, tion presents a parallel opportunity for the restoration of equity.
Selenta, & Lord, 2006). It has also recently been shown to affect Forinstance,anemployeewhoapologizesforsubversivebehavior
perceptions of interpersonal conflict (Gelfand et al., 2001). How- duringameetingcouldoffertoshowovertrespectforabossatthe
ever,itsimplicationsforperceptionsofandreactionstoapologies nextmeeting.Thus,compensationcanreferenceeithertangibleor
have yet to be examined. Forgiveness is examined as a key moreemotionallydrivencompensatoryoffers.
outcome variable for both its ubiquity in the apology literature While offers of compensation focus on equity and exchange,
and its implications for important individual and interpersonal expressionsofempathyfocusonrelationalissues.Theydemonstrate
outcomes such as well-being (Bono, McCullough, & Root, 2008), offenders’ recognition of and concern for their victims’ suffering,
stress (Witvliet, Ludwig, & Vander Laan, 2001), and helping bothsocio-emotionallyandcognitively(cf.Davis,1983).Fromaso-
behavior(Karremans,VanLange,&Holland,2005). cio-emotionalperspective,offendersmightdemonstrateempathy
Thegeneralpropositionofthispaper,discussedatlengthbelow, byexpressingwarmthtowardtheirvictimsorcompassionfortheir
isthatforgivenesswillbeenhancedwhenoffenders’apologiesare suffering.Fromacognitiveperspective,offendersmightdisplayan
consistentwithvictims’self-views.Specifically,itishypothesized understandingofthevictim’spointofviewortheconsequencesof
that victims who emphasize the independent, relational, and theoffenseforthevictim’swell-being.Aswithoffersofcompensa-
collective self-construals will be most likely to forgive their tion,researchonexpressionsofempathycanbefoundinSociology
offendersfollowingoffersof compensation,expressionsofempathy, (Goffman, 1967), Psychology (Lazare, 2004; Schlenker & Darby,
and acknowledgments of violated rules/norms, respectively. Two 1981), and many other fields (Cohen, 1999; Tavuchis, 1991;
studies have been conducted to test these hypotheses. In Study Wagatsuma&Rosett,1986).Recentadvancesinmanagementthe-
1,adirectassessmenttechniqueisusedtoexaminetherelation- oryhavelikewiseemphasizedtheimportanceofexpressingempa-
shipbetweenself-construalandparticipants’perceptionsofwhat thyforemployees,notinghowrelationalitycanhelptoeaseconflict
constitutesa‘‘goodapology.”InStudy2,apolicycapturingmeth- andfacilitatecooperation(Gelfand,Major,Raver,Nishii,&O’Brien,
odologyisused toconfirmandextendthefindingsfromStudy1 2006).Schmittetal.(2004)empiricallylinkedexpressionsofempa-
usinghierarchicallinearmodelinganalyses. thytovictims’perceptionsoftheirperpetrators,andoperational-
ized the component through the phrase ‘‘I feel really sorry for
Apologycomponents what I have done. I know how you feel now” (p. 469). Kotani
(2002)providedadditionaltheoreticalsupportforthiscomponent
Apology components have received only scattered empirical byemphasizingitsintegralroleinnon-Westerncontexts.Thesig-
consideration within Psychology (e.g. Darby & Schlenker, 1982; nificanceofexpressionsofempathywasrecentlyevidencedduring
Schmitt et al., 2004). However, they have received greater atten- the trial of a Catholic Bishop for charges of abuse. The plaintiffs
tion from a number of theorists throughout the humanities and were awarded $23.4 million dollars, but demanded that the
R.Fehr,M.J.Gelfand/OrganizationalBehaviorandHumanDecisionProcesses113(2010)37–50 39
settlement be stalled pending a direct apology. The Bishop ex- Self-construals have been shown to influence a wide range of
pressed ‘‘with very profound and deep compassion” empathy for individualperceptions,processes,andoutcomesincludinggeneral
the‘‘immensesuffering”thathehadcaused,afterwhichtheplain- well-being, reactions to injustice, motivations for goal pursuit,
tiffsdroppedtheirmonetaryrequest(Blaney&Dooley,1998,p.15). socialcomparisonprocesses,leaderperceptions,andevenaesthetic
Whereas offers of compensation and expressions of empathy preferences(Crossetal.,2003;Gore&Cross,2006;Guimondetal.,
arefocusedonthevictim–offenderdyad,acknowledgmentsofvio- 2007; Holmvall & Bobocel, 2008; Johnson et al., 2006; Van
latedrules/normsexpandthescopeofanapologytothegroupcon- Knippenberg,VanKnippenberg,DeCremer,&Hogg,2004;Zhang,
text. In essence, this component involves a recognition that Feick,&Price,2006).Gelfandetal.(2001)havefurthermoredirectly
interpersonalbehaviorisboundbyrulesandnorms,eitherimplicit demonstrated that self-construal impacts victims’ perceptions of
orexplicit,thatmustbefollowed.Forinstance,duringanapology otherwise identical conflict episodes. Nonetheless, research has
for the mistreatment of civilians, a soldier offered the following yettoelucidatehowself-construalmightimpactapologypercep-
referencetotherulesandnormsofhergroup:‘‘Ifailedmyduties. tions and thus the process of forgiveness. To this end, it is
Ifailedmymissiontoprotectanddefend...Iletdowneverysingle hypothesized that apologies will be most effective when they
soldierthatservestoday”(Stevenson,2005,p.23).Thus,acknowl- are congruent with victims’ self-construals – that is, when they
edgments of violated rules/norms may be particularly important containcomponentsthatalignwithvictims’mostcentralattitudes
within organizational and group contexts wherein strong behav- and beliefs. This hypothesis is consistent with the tenants of
ioral norms are prevalent. On a broader level, acknowledgments self-verificationtheory,whichemphasizeindividuals’preferences
ofviolatedrules/normscanalsoreferencepeoples’dutiesasmem- for information that is consistent with and verifies their own
bersofanentiresocietyorculture.AccordingtoWagatsumaand self-conceptualizations(Swann,1987).Whenapologiesarealigned
Rosett (1986), ‘‘The act of apologizing can be significant for its with victims’ self-construals, they verify victims’ beliefs about
ownsakeasanacknowledgmentoftheauthorityofthehierarchi- interpersonalinteractionsandthetypesofapologiesthatmustbe
cal structure upon which social harmony is based” (p. 473). offeredinthepursuitofforgiveness.
Tavuchis (1991) expands upon the logic of this component by In the sections below, the tripartite model of self-construal is
conceptualizing apologies as vital to the social order, which reviewed, and it is proposed that (a) individuals who emphasize
‘‘depends... on some measure of commitment to norms dealing theindependentselfwillreactmostpositivelytoapologiesthatin-
with standards of behavior and institutional arrangements” clude offers of compensation, (b) individuals who emphasize the
(p.12).Insomuchasanapologyreferencesthesocialorder,it‘‘di- relationalselfwillreactmostpositivelytoapologiesthatinclude
rectsattentiontorulesandmeta-rules,thatis,rulesaboutrules” expressions of empathy, and (c) individuals who emphasize the
(p. 13). These acknowledgments are at once socio-emotional and collectiveselfwillreactmostpositivelytoapologiesthatinclude
cognitive, entailing a recognition that a rule was broken and an acknowledgmentsofviolatedrules/norms.Whereasapologycom-
appealtothenegativeemotionalimpactofthebreach.Cross-cultural ponents that are congruent with victims’ self-construals should
theory suggests that acknowledgments of violated rules/norms stronglyimpactparticipants’reactionstoapologies,apologycom-
might be particularly important to many non-Western cultures ponents that are incongruent with the self should ‘‘fall on deaf
andsubgroups,whichhavebeenshowntoweighsocialrulesand ears”andexhibitweakereffects.
norms more heavily than Western cultures (Gelfand et al., 2001;
Markus&Kitayama,1991). Theindependentself-construalandoffersofcompensation
Apologythroughthelensoftheself Victims who possess strong independent self-construals view
themselvesasuniqueandautonomousentitieswhoare‘‘separated
Giventheinherentvalueofeachoftheabove-mentionedapol- fromothers”(Cross&Madson,1997,p.7).Theyarehighlyconcerned
ogycomponents,itstandstoreasonthateachcomponentwillex- withtheirpersonalrightsandentitlements,generallypursueself-
hibit a positive effect on forgiveness. Yet previous research has relevantgoals, and view theirrelationships as exchange-oriented
nonetheless failed to move beyond a conceptualization of apolo- (Bresnahan, Levine, & Chiu, 2004). In their actions, independent
gies as dichotomous. One purpose of the current research, then, people tend to treat relationships as exchange-oriented, wherein
istoexaminethedirecteffectsofapologycomponentsonvictim they expect to receive specific benefits from what they provide
forgiveness. Moreover, it is argued that people can differ in their others (Bresnahan et al., 2004; Downie, Koestner, Horberg, &
reactionstootherwiseidenticalstimuli,suchasapologiesthatin- Haga,2006;Hara&Kim,2004).Thus,wheninteractingwithothers,
clude the same components. Thus, apologies might also exhibit people with independent self-construals demonstrate a focus on
contingent effects, influenced by victim individual differences. competitionovercooperation,exchangeovercommunality,rights
Oneindividualdifferencethatislikelytoaffectvictims’reactions over duties, and individual achievement over group consensus
tospecificapologycomponentsisself-construal. (Shtyenberg,Gelfand,&Kim,2007;Wagner,1995).
Converging evidence in the psychological literature suggests As a consequence of their beliefs and attitudes about the self,
thattheselfcanactasapowerfulregulatorofindividuals’behav- individuals with strong independent self-construals are likely to
ior,influencinghowinformationisperceived,processed,andacted focusonissuesrelatedtotheirautonomy,individuality,andenti-
upon(Fiske&Taylor,1991).Whenspecificattitudesorbeliefsare tlementsfollowinganoffense.Theyshouldbelessconcernedwith
particularly central to the self, individuals tend to interpret the offenders’expressionsofempathyandmoreconcernedwithapol-
worldthroughthelensoftheseviewsandpayparticularattention ogiesthataddresswhattheoffenderwilldotorestoreequitytothe
to information that is consistent with them (Markus, Smith, & relationship.Morethanotherapologycomponents,offersofcom-
Moreland,1985).Recentresearchhassuggestedatripartitemodel pensationarecloselyalignedwiththeseconcernsandshouldthere-
oftheself,consistingoftheindependent,relational,andcollective fore be particularly effective. Offers of compensation emphasize
selves (Brewer & Gardner, 1996; Cross & Madson, 1997; Johnson the importance of re-establishing equity and restoring what the
etal.,2006;Kashimaetal.,1995).Whereastheindependentself- victim lost, be it physical or emotional. They establish the legiti-
construalinvolvesaconceptualizationoftheselfasauniqueand macyofthevictim’sclaimsand,indoingso,allowthevictimtofeel
autonomous entity, the relational and collective self-construals that he or she has ‘‘won” the moral competition between them,
emphasizecloseinterpersonalrelationshipsandgroupcategoriza- providinginformationthatiscongruentwiththevictim’sconcep-
tions,respectively. tualization of interpersonal relationships as competition-based
40 R.Fehr,M.J.Gelfand/OrganizationalBehaviorandHumanDecisionProcesses113(2010)37–50
(Brewer & Gardner, 1996; Markus & Kitayama, 1991). Previous predictive of forgiveness than self-oriented variables, such as
research supports this hypothesis. Ohbuchi, Fukushima, and self-esteem,inthehighlyrelationalcountryofChina.
Tedeschi (1999) found that Americans, who generally emphasize
the independent self-construal, tend to focus on the restoration H2: Individuals who emphasize the relational self-construal
of fairness following an offense. Wagatsuma and Rosett (1986) willreactmostpositivelytoapologiesthatincludeexpressions
similarlynotedthatAmericansarelikelyto‘‘considerthatpaying ofempathy.
thedamagesoracceptingpunishmentendsfurtherresponsibility
andthatthereisnoneedforpersonalcontrition”inlawcontexts Thecollectiveself-construalandacknowledgmentsofviolated
(p.462).Theimportanceofcompensationtotheindependentself rules/norms
isfurtherevidencedintheorganizationalliterature,whichshows
that the independent self is related to a strong concern over the Whereastherelationalselffocusesonclose,personalized,and
just distribution of resources (i.e. distributive justice; Johnson generally dyadic relationships, the collective self is focused on a
et al., 2006). Relatedly, Mattila and Patterson (2004) found that broader, more impersonal identification with groups and social
highlyindependentpeopleweremorelikelytoreactpositivelyto categories (Brewer & Gardner, 1996). When the collective self-
compensation following a service failure than less independent construal is dominant the salience of one’s group identity is en-
people. hanced,shiftingtheconceptualizationoftheselffrom‘‘I”to‘‘we”
(Gardner, Gabriel, & Hochschild, 2002). In interpreting external
H1:Individualswhoemphasizetheindependentself-construal behavior,groupdifferencesbecomehighlysalient,asdotherules
will react most positively to apologies that include offers of thatguidethesegroups,includinggroupduties,norms,andcom-
compensation. mitments(Johnson&Chang,2006;Marx,Stapel,&Muller,2005).
Thus, individuals with strong collective self-construals are likely
Therelationalself-construalandexpressionsofempathy to pay particularly close attention to offenses against the group
andviolationsagainstgroupexpectations.Intheirbehavior,people
In contrast to the independent self-construal, victims with withhighlycollectiveself-construalsexhibitastrongbiastoward
highlyrelationalself-construalsconceptualizethemselvesasfun- theirowngroups,actinwaysthatdemonstrateacommitmentto
damentally connected to other people (Cross & Madson, 1997). the group, and focus on confirming the positive qualities of the
Theydo notview themselves as separatefrom others, but rather group. For instance, they might defend the morality of a group
aslinkedtoanddefinedbytheirrelationships.Inbothperception member’sactions,worktowardsolvingagroupproblem,orextol
and action, highly relational people focus on the quality of their thevirtuesofthegrouptooutsiders.
relationships and direct their actions toward maintaining and For the collective self, acknowledgments of violated rules/
developingsuchrelationships(Gelfandetal.,2006).Forexample, normsshouldbevital.Aboveandbeyondoffersofcompensation
recent research has demonstrated the predictive validity of the or expressions of empathy, acknowledgments of violated rules/
relational self for information disclosure among roommates, in normsshiftthefocusofanapologyfromthevictim–offenderdyad
turn leading to higher levels of dyadic intimacy (Gore, Cross, & tothebroadersocialcontextinwhichtheviolationisembedded.
Morris,2006).Furthermore,therelationalselfhasbeenshownto Acknowledgments of violated rules/norms demonstrate an
correlate with relational motivations for goal pursuit (Gore & understanding of the importance of the norms and rules that
Cross, 2006), elaborate cognitivenetworks for interpersonal rela- define victims’ groups. They further establish a belief in the
tionships(Cross,Morris,&Gore,2002),andtheaccuracyofindivid- legitimacy of the group’s expectations and a concern for the
uals’ perceptions of their friends’ values and beliefs (Cross & emotional implications of violating those expectations, as in
Morris, 2003). In sum,these findings pointto theintegral roleof thecaseofthesoldierwhoacknowledgedthatsheshouldnothave
relationships in the lives of highly relational people. They are violatedthenormsandexpectationsofthemilitarygrouptowhich
highly attentive to relational cues, deeply concerned with the shebelonged.
status of their relationships, and motivated to act in ways that
fosterthem. H3: Individuals who emphasize the collective self-construal
Morethanoffersofcompensationoracknowledgmentsofvio- willreactmostpositivelytoapologiesthatincludeacknowledg-
lated rules/norms, expressions of empathy should be particularly mentsofviolatedrules/norms.
efficacious in eliciting forgiveness from victims that emphasize
therelationalself.Suchexpressionsarehighlyrelational,insomuch Takentogether,thehypothesespresentedaboveformamodel
astheyaddresstheemotionalstateofthevictimandimplyacog- wherein victims’ emphasized self-construals determine which
nitiveunderstandingofthevictim’sperspective.Theysuggestfeel- apology components will be reacted to most positively. These
ingsofcloseness,interdependence,andinterpersonalrelatedness, hypotheses are tested across two studies. In Study 1, a direct
all of which are important when the relational self-construal is assessment technique is utilized to provide preliminary evidence
strong(Cross&Madson,1997).Recentresearchfurthermoresug- for the hypotheses. In Study 2, the findingsfrom Study 1 are ex-
gests that empathy entails a literal embodiment of others’ emo- pandedby(a)examiningtheeffectsofapologycomponentswithin
tional experiences, highlighting the other-oriented nature of the context of an interpersonal conflict and (b) measuring the
empathic experience (Niedenthal, Barsalou, Winkielman, direct relationship between apology components and victims’
Karuth-Gruber, & Ric, 2005; Niedenthal, Winkielman, Mondillon, forgiveness of their offenders. Furthermore, Study 2 examines
& Vermeulen, 2009). Johnson et al. (2006) provide some support whether apology effectiveness varies according to the situational
for the idea that relational people value expressions of empathic contextinwhichitisembedded.Aswithallconflictmanagement
concernandunderstanding.Intheirstudy,interpersonaljustice– strategies,apologiesdonotoccurinavacuumbutrathercanoccur
whichreflectsanemotionalconcernforothers–wasfoundtobe acrossanarrayofcontexts.Asthesecontextschange,sotoomight
particularly important when the relational self was strong. theeffectivenessofanapology.Onekeysituationalfactorshownto
Fu, Watkins, and Hui (2004) provide further evidence for the impactawiderangeofconflictoutcomesisharmseverity(Boon&
importance of expressions of empathy among relational people Sulsky,1997;Ohbuchietal.,1989;Zechmeister&Romero,2002).
in a cross-cultural setting. Specifically, they found that ‘‘other- Thus, a final goal for Study 2 is to examine the impact of harm
oriented” variables, such as relationship orientation, were more severityonapologyandself-construaldynamics.
R.Fehr,M.J.Gelfand/OrganizationalBehaviorandHumanDecisionProcesses113(2010)37–50 41
Study1 assessedwithfiveitems(1=StronglyDisagree,5=StronglyAgree).
Respective examples of items from the independent, relational,
Participantsandprocedure and collective subscales include ‘‘I have a strong need to know
how I stand in comparison to my classmates or coworkers,”
Participants were 175 undergraduate students (73.1% women, ‘‘Caring deeply about another person such as a close friend or
26.9% men) at a large Mid-Atlantic University who participated relative is very important to me,” and ‘‘I feel great pride when
inthestudyinexchangeforcoursecredit.Theaverageageofpar- myteamorworkgroupdoeswell,evenifI’mnotthemainreason
ticipantswas19.7.Thestudyitselfwascompletedacrosstwoses- for success.” Alphas for the scales were .77, .67, and .63, respec-
sions. In the first session, participants completed a self-report tively, consistent with previous studies (Johnson & Chang, 2006;
measureofself-construalalongwithbasicdemographicinforma- Johnsonetal.,2006).
tion. In the second session, participants completed the primary
apology measure. The two sessions were completed an average
Results
of29daysapart.Thedistancebetweensessionsexhibitednoasso-
ciationswithanyofthestudy’svariables,noranyinteractionswith
Descriptivestatisticsandcorrelations
thestudy’sprimaryfindings.
Table2presentsmeans,standarddeviations,andcorrelationsfor
Apologycomponentscaledevelopment theprimarystudyvariables.Meanratingsofoffersofcompensation,
expressions of empathy, and acknowledgments of violated rules/
Aliteraturereviewidentifiednopreexistingscalesthattapinto norms as indicative of ‘‘good apologies” were all above the scale
thespecificcomponentsofapologies.Therefore,theauthorsdevel- midpoint, indicating that participants seemed, on average, to
oped a set of items for this purpose. Initial item generation was recognizeeachapologycomponentasindicativeofagoodapology.
conductedonthebasisofacontentanalysisoftheapologyliterature. For offers of compensation, M=3.59, t(174)=13.26, p<.001. For
Fromthisinitialpoolofapproximately30items,15wereidentified expressions of empathy, M=4.39, t(174)=31.91, p<.001. For
viadiscussionbetweentheauthorsasmostrepresentativeofthe acknowledgmentofaviolatedrule/norm,M=3.86,t(174)=13.57,
coreofeachapologycomponent.Toensurethattheseitemsindeed p<.001.Moreover,correlationsamongthethreescalesweremod-
tappedintotheirintendedconstructsagroupoffivegraduatestu- est(.15–.31)providingfurtherevidenceforthedistinctivenessof
dents,blindtothehypotheses,wereaskedtoconductaQ-sortof thethreefocalcomponents.
theitemsbaseduponaprovidedsetofdefinitionsofeachapology
component. The students displayed 100% agreement in assigning Hierarchicalregression
theitemstotheappropriateapologycomponents.
Witheachapologyitem,participantswereaskedtoexpresstheir Hypotheses1–3weretestedviahierarchicalregression.Gender,
agreementordisagreementthatagoodapologyshouldincludea whichhasbeenshowntorelatetobothindividuals’self-construals
specificcomponent.Theoffersofcompensationscaleincludesitems (Cross&Madson,1997)andreactionstooffenses(Tomlinsonetal.,
suchas‘‘Ingeneral,agoodapologyshouldincludeanoffertocom- 2004),wascontrolledforinallanalyses.Alltheorizedrelationships
pensate me for what happened” and ‘‘In general, a good apology were tested by regressing the individual apology components on
shouldincludeasuggestionthathe/shereimbursemeinsomeway.” gender(Step1)andthethreeself-construals(Step2).
Representativeitemsfromtheexpressionsofempathyscaleinclude Totesthypothesis1,theoffersofcompensationscalewerere-
‘‘Ingeneral,agoodapologyshouldincludetruesympathyforme” gressedongenderandthethreeself-construals.Asignificanteffect
and ‘‘In general, a good apology should include an expression of fortheindependentselfwastheorized.Insupportofthehypothe-
greatconcernformysuffering.”Sampleitemsfromtheacknowledg- sis,therewasasignificanteffectoftheindependentself-construal
mentsofviolatedrules/normsscaleinclude‘‘Ingeneral,agoodapol- onthebeliefthatagoodapologyshouldincludeanofferofcom-
ogy should include an acknowledgement that he/she violated an pensation, b=.19, p<.05. Next, to test hypothesis 2, the expres-
importantgrouprule”and‘‘Ingeneral,agoodapologyshouldin- sions of empathy scale were regressed on gender (Step 1) and
cludeanacknowledgementthathe/shedidn’tliveuptogroupstan- the three self-construals (Step 2). As predicted, the relational
dards.”Allapologycomponentsweremeasuredwith5-itemscales self-construalwasrelatedtothebeliefthatanexpressionofempa-
(1=StronglyDisagree,5=StronglyAgree). thyisindicativeofagoodapology,b=.32,p<.01.Finally,hypoth-
Toconfirmthedistinctivenessofthethreeapologyscales,explor- esis3wastestedbyregressingtheacknowledgmentofaviolated
atory factor analysis with varimax rotation was conducted via a rule/norm scale on gender (Step 1) and the three self-construals
maximumlikelihoodprocedure(Fabrigar,Wegener,MacCallum,& (Step2).Aspredicted,thecollectiveself-construalwasrelatedto
Strahan,1999).Resultssupportedthethreecomponentmodel,as thebeliefthatanacknowledgmentofaviolatedrule/normisindic-
evidencedbyhighloadingsoneachitem’stheoreticallyrelevantfac- ative of a good apology, b=.25, p<.01. In each case, only the
tor and low cross-loadings. An exception was one item from the relevantself-construalexhibitedasignificanteffect.Therelational
compensationscale,whichwasdeleted.Thefinal14-itemscalethus andcollectiveself-construalsdidnotpredictreactionstooffersof
included five items each for the empathy and acknowledgment compensation, the independent and collective self-construals did
scalesandfouritemsforthecompensationscale.Individualitems not predict reactions to expressions of empathy, and the
andtheirloadingsarepresentedinTable1.Thefinalthreescales independent and relational self-construals did not predict reac-
yieldedcoefficientalphasof.81forthecompensationscale,.85for tions to acknowledgments of violated rules/norms. Thus,Study 1
theempathyscale,and.92fortherules/normsscale. providedclearevidencethattheperceivedimportanceofdifferent
apologycomponentsis directlyrelated tovictims’self-construals
(seeTable3).
Self-construal
Self-construal was measured with the Levels of Self-Concept Study1discussion
Scale(LSCS;Selenta&Lord,2005),designedtomeasureindividual
differences in the chronic accessibilities of the independent, Taken together, the findings from Study 1 provide initial sup-
relational, and collective self-construals. Each self-construal was port for the theorized link between self-construal and apology
42 R.Fehr,M.J.Gelfand/OrganizationalBehaviorandHumanDecisionProcesses113(2010)37–50
Table1
Exploratoryfactoranalysisforapologycomponentitems,Study1.
Items Factor
Ingeneral,agoodapologyshouldinclude... Compensation Empathy Rule/norm
1...anoffertocompensatemeforwhathappened .785 (cid:2).006 .191
2...anoffertohelpmerecovermydamages .672 .014 .079
3...anoffertodosomethingspecifictomakeupforwhathappened .614 .133 .124
4...asuggestionthathe/shereimbursemeinsomeway .741 .032 .115
5...anexpressionofgreatconcernformysuffering .049 .823 .088
6...ashowofempathytowardme (cid:2).044 .710 .009
7...anindicationthathe/shetrulycaresabouthowIfeel (cid:2).073 .712 .139
8...anexpressionoftendernesstowardme .117 .669 .176
9...truesympathyforme .218 .661 .128
10...averbalrecognitionthathe/shefailedtoactasagoodgroupmember .051 .144 .745
11...anadmissionthathe/shedidnotliveuptothestandardsofthegroup .128 .094 .931
12...anacknowledgmentthathe/sheviolatedanimportantgrouprule .162 .118 .838
13...ashowofconcernforbreakinganimportantsocialnorm .163 .102 .716
14...anacknowledgmentthathe/shedidn’tliveuptogroupstandards .179 .126 .804
Eigenvalue 1.75 2.27 4.24
%Varianceexplained 12.51 16.21 30.28
Note.Boldfacevaluesindicatewhichfactorstheitemsloadon.Procedurewasmaximumlikelihoodestimationwithvarimaxrotation.
Table2
Descriptivestatisticsandzero-ordercorrelations,Study1.
M SD 1 2 3 4 5 6 7
1.Gendera 1.73 .44 –
2.Independentself 3.37 .83 (cid:2).07 (.77)
3.Relationalself 4.58 .44 .05 (cid:2).09 (.67)
4.Collectiveself 4.15 .53 .21** (cid:2).05 .28** (.63)
5.Compensation 3.59 .74 .08 .18* .01 .13 (.81)
6.Empathy 4.39 .57 .11 (cid:2).06 .36** .20** .15 (.85)
7.Rule/norm 3.86 .84 .14 (cid:2).06 .07 .27** .31** .27** (.92)
Note.N=175.
a Male=1,female=2.
* p<.05.
** p<.01.
Table3
Hierarchicalregressionanalyses–perceptionsofagoodapology,Study1.
Stepandindependentvariables Compensation Empathy Rule/norm
b TotalR2 DR2 b TotalR2 DR2 b TotalR2 DR2
Step1
Gender .08 .11 .14
.006 .011 .019
Step2
Independentself .19* (cid:2).02 (cid:2).05
Relationalself (cid:2).01 .32** (cid:2).01
Collectiveself .12 .10 .25**
.055 .049 .144 .133 .083 .064
Note.N=175.
* p<.05.
** p<.01.
components. Each apology component was seen, on average, as demonstrate the importance of examining apologies beyond the
indicativeofagoodapology.Scalecorrelationsandfactoranalysis yes/nodichotomiesthathavedominatedtheliterature.
confirmedthateachcomponentwasseenasdistinct.Inlinewith DespitethefindingsfromStudy1,somequestionsremain.First,
theprimaryhypotheses,apologycomponentsweremostlikelyto it is not known if individuals’ perceptions of what should be in-
be seen as effective when they weretightly aligned with partici- cluded in a ‘‘good apology” are indicative of victims’ reactions to
pants’ self-construals. Standardized betas for analyses regressing apologycomponentsfollowingconflict.Relatedly,theimplications
apologycomponentperceptionsoncongruentself-construalswere of these perceptions for conflict outcomes such as forgiveness
allsignificant,rangingfrom.19to.32.Ontheotherhand,standard- remain untested. In Study 2, these concerns were addressed by
izedbetasforanalysesregressingapologycomponentperceptions examining the impact of apology components and self-construal
on incongruent self-construals were all non-significant, ranging on forgiveness. Hierarchical linear modeling was used to isolate
from (cid:2).01 to .12. Empirically, these findings highlight the thecross-leveleffectsofself-construalontherelationshipbetween
divergent validity of the focal apology components and their apology content and forgiveness. Forgiveness was examined as a
relationships to victim self-construals. Theoretically, the findings construct of principle interest to conflict management scholars
R.Fehr,M.J.Gelfand/OrganizationalBehaviorandHumanDecisionProcesses113(2010)37–50 43
(e.g.Aquino,Grover,Goldman, &Folger,2003;McCulloughetal., organizational phenomena, including job performance ratings
1998) that has been shown to predict a range of important out- (Rotundo & Sackett, 2002), reputation perceptions (Cable &
comes including interpersonal reconciliation (Fincham, Beach, & Graham, 2000), revenge (Tripp, Bies, & Aquino, 2002), job choice
Davila, 2007), self-esteem (Karremans, Van Lange, Ouwerkerk, & (Judge & Bretz, 1992), fairness perceptions (Hemingway& Conte,
Kluwer,2003),andphysicalhealth(Witvlietetal.,2001)toname 2003),employeeselection(Graves&Karren,1992),andconflictres-
afew. olution(Tomlinsonetal.,2004).Inatypicalpolicycapturingstudy,
Inexaminingtheinteractiveeffectsofapologycomponentsand theexperimenterpresentsparticipantswithaseriesofprofilesthat
self-construal on apology effectiveness, an important question is manipulateafocalsetofvariablesandmeasuretheimpactofthese
how the situational context might bound these effects. One key manipulationsonanoutcomeofinterest.Thecurrentstudyapplied
situationalconstructisharmseverity.Awiderangeofstudieshave this methodology by manipulating apology components across a
demonstrated the importance of harm severity for conflict seriesofconflictscenariosandmeasuringtheimpactoftheapology
outcomes (Boon & Sulsky, 1997; Exline, Baumeister, Bushman, componentsonparticipants’forgivenessratings.
Campbell,&Finkel,2004;Fincham,Jackson,&Beach,2005;Frantz With policy capturing, it is also possible to explore between-
& Bennigson, 2005; Fukuno & Ohbuchi, 1998; Itoi, Ohbuchi, & subject differences in policy profiles (e.g. Rotundo & Sackett,
Fukuno, 1996; Ohbuchi et al., 1989; Zechmeister & Romero, 2002).Inthisstudy,foursuchvariableswereexplored.First,totest
2002).Datafromtheseandotherstudiesindicateseveralcompet- theprimaryhypotheses,theindependent,relational,andcollective
inghypothesesforthecurrentresearch. self-construals were measured as individual differences. Second,
Onehypothesisisthatharmseveritywillstrengthentheimpact harmseveritywasmanipulatedbetweensubjectstotesttherobust-
ofawell-tailoredapologyonforgiveness,resultinginathree-way nessofself-construal’seffects.Thus,thecurrentstudyallowedfora
interaction between harm severity, apology, and self-construal simultaneousexaminationoftheeffectsofself-construalandharm
such that harm severity strengthens the effects of the two-way severityonindividuals’weightingsofspecificapologycomponents
apology/self-construal interactions. According to this hypothesis, intheirforgivenessdecisions.
the antecedents of forgiveness are strongest under conditions of
severe harm because severity strengthens the need for and rele- Participantsandprocedure
vance of actions that promote forgiveness (Pronk, Karremans,
Overbeek, Vermulst, & Wigboldus, 2010). A mild offense such as One hundred and seventy-one undergraduate students from a
beinglateforameetingislesslikelytofosternegativevictimreac- largeMid-AtlanticUniversityparticipatedintheprimarystudyin
tionsthanasevereoffensesuchasviolence,andthusmaylessen exchangeforcoursecredit(75.4%women,24.6%men).Thestudy
theperceivedneedforaconflictinterventionlikeapology.Consis- was conducted across two sessions, with participants filling out
tent with this idea, harm severity has been found to predict vic- theself-construalmeasureinsessiononeandthepolicycapturing
tims’ expectations that an apology will be offered (McLaughlin, experimentinsessiontwo.Participants’averageagewas20.2.The
Cody,&O’Hair,1983;Ohbuchietal.,1989)andtostrengthenthe twosessionswerecompletedanaverageof15daysapart.Aswith
positive effect of apology on emotional alleviation (Fukuno & Study 1, the distance between sessions exhibited no associations
Ohbuchi, 1998). A competing hypothesis, however, is that harm with any of the study’s variables, nor any interactions with the
severity will weaken the impact of self-construal and apology study’sprimaryfindings.
components on forgiveness, resulting in a three-way interaction
betweenharmseverity,apology,andself-construalsuchthatharm Policycapturingdesign
severityweakenstheeffectsofthetwo-wayapology/self-construal
interactions. According to this hypothesis, harm severity lessens Uponenteringthelab,participantsfirstreadsomebackground
the effectiveness of apologies by eliciting negative feelings that informationregardingtheirrelationshipwithafriend.Theprimary
aretoostrong,thusleadingvictimstorebukealleffortsatreconcil- purposesofthebackgroundinformationwereto(a)establishthe
iation (Schlenker, 1985; Schoenbach, 1990). Consistent with this participant’srelationshipwiththefriendand(b)detailthecontext
hypothesis, previous research has demonstrated that when vic- ofthesituationinwhichtheconflictoccurred.Theinformationwas
tims’pre-apologyimpressionsoftheir offendersare highlynega- consistentacrossparticipantsandincludedthefollowing:
tive, apologies are ineffective (Struthers et al., 2008). A final
Pat lives down the hall from you in your dorm here at Local
possibility,however,isthatharmseveritymightexertanulleffect
University. You see each other frequently, and are in the same
on apology effectiveness, resulting in two-way apology/self-
co-edfraternity.You’vegonetomanypartiesandeventstogether
construal interactions that are not moderated by harm severity.
and often eat at the same table in the dining halls. Recently, the
Given the centrality of the self in the regulation of emotions,
two of you were working next to each other in the library when
cognitions, and behavior across a range of social contexts (e.g.
Pataskedtoborrowyourlaptoptoquicklywriteupanassignment.
Cross&Madson,1997;Markus&Kitayama,1991),theinteractive
Youagreed.Later,Patreachedforadisktosavetheproject.
effects of apologycomponentsand the self on victim forgiveness
mayprovetobeconsistentacrosslevelsofharmseverity.Totest Afterreadingthebackgroundinformation,theparticipantsread
thesecompetinghypotheses,Study2thereforeexaminestheinter- 10conflictscenariosthatmanipulatedtheapologiesthatweregi-
activeeffectsofapologiesandself-construalacrossthreelevelsof ven following the conflict. The conflict event was based upon
harmseverity. Gonzales, Manning, and Haugen’s (1992) ‘‘disk” conflict scenario,
as published in the Journal of Personality and Social Psychology,
whereintheoffenderusesadiskthatcausesthevictim’scomputer
Study2 tocrash.Allthreeapologycomponentswerefullycrossedinafull
factorialdesign,allowingfortheexaminationoftheindependent
Study2utilizedapolicycapturingmethodology–anapplication effects of each component. Two repeat scenarios were also in-
of Brunswik’s approach to studying human decision making that cludedtoallowforreliabilityanalysis.
uses statistics to quantitatively describe ‘‘the relations between Harmseveritywasmanipulatedbetweensubjects,andsowas
someone’s judgment and the information... used to make that consistentacrossscenariosforeachparticipant.Inthemildcondi-
judgment” (Stewart, 1988, p. 41). Previous research has utilized tion,participantsweretoldthattheylost‘‘asmallamountofwork”
policycapturingtoexploredecisionpatternsrelatedtoarangeof that would take 1h to re-enter. In the control condition, the
44 R.Fehr,M.J.Gelfand/OrganizationalBehaviorandHumanDecisionProcesses113(2010)37–50
severityofharmwasnotdetailed.Intheseverecondition,partici- respectively. A series of t-tests confirmed significant differences
pants were told that they lost ‘‘a significant portion of work”, inseverityperceptionsacrossmanipulations.
wouldneedseveralweekstore-enterit,andreceivedlowgrades
onseveralimportantassignmentsasaresult.Aftereachscenario,
Results
participantswereaskedtoratetheirforgivenessoftheoffender.
Asamplescenario,includingallthreeapologycomponentsand
Thereliabilityofparticipants’forgivenessratingswasassessed
amildseveritycondition,islistedbelow.Eachcomponentislisted
via the two repeat profile pairs, which yielded reliability coeffi-
inbracketsafteritappears.
cients of .91 and .92, respectively. All remaining analyses were
Pathurriedlygrabbedthewrongdisk,marked‘‘donotuse”,froma computed via hierarchical linear modeling, with Level 2 defined
pileonthetable.Ithadavirusonit,andwhenPatinsertedthedisk asbetween-personandLevel1aswithin-person.TheLevel2cor-
into your laptop, the laptop crashed. You lost a small amount of relation matrix is provided in Table 4; the Level 1 correlation
schoolwork,whichwilltakeyouanhourtore-enter.[mildharm] matrixisprovidedinTable5.
Upon realizing what happened, Pat said ‘‘Sorry”. Pat then Level1analysis
expressed concern for your suffering, saying ‘‘I feel sick to my
stomach thinking about how upset you must be over this.” As initial evidence for the value of different apology compo-
[expressionofempathy]Then,Patadmittedtonotbeingagood nents in eliciting forgiveness, the amount of within-person vari-
groupmember,saying‘‘I’veletthewholegroupdown.I’vefailed ance in the sample at Level 1 was examined (i.e. the amount of
in my duties to our fraternity and the campus community.” variance accounted for by the apology manipulation; Hofmann,
[acknowledgment of violated rule/norm] Lastly, Pat suggested Griffin, & Gavin, 2000). The calculated effect size indicated that
thepossibilityofcompensatingyoubysaying‘‘Icanfindsomeone the Level 1 predictors (offers of compensation, expressions of
tofixthecomputerforyou.”[offerofcompensation] empathy, and acknowledgments of violated rules/norms) ac-
counted for 59.8% of the variance in forgiveness. To assess the
Measures amount of variance specifically accounted for by each apology
component,afullLevel1equationwasthencomputedbysimulta-
Forgiveness neously regressing forgiveness on all three apology components.
Forgiveness was measured using two items adapted from the For offers of compensation, b=.36, p<.01. For expressions of
19-Item Transgression-Related Interpersonal Motivations Scale, empathy, b=.22, p<.01. For acknowledgment of a violated rule/
selected on the basis of brevity and face validity (TRIM-19; norm,b=.16,p<.01.Theseresultsindicatethatallthreeapology
McCullough&Hoyt,2002;McCulloughetal.,1998).Thefirstitem componentssignificantly and positively affected victims’ forgive-
was ‘‘Given this situation, I would forgive Pat.” The second item ness, consistent with previous theory and research (e.g. Darby &
was ‘‘Given this situation, I would trust Pat in the future.” The Schlenker,1982;Schmittetal.,2004).
items were combined to form an overall measure of forgiveness,
a=.84. Cross-levelanalysis
Self-construal Toassesstheinteractiveeffectsofself-construalandapologies
As in Study 1, self-construal was measured with the LSCS onforgiveness,aslopes-as-outcomesapproachwasdeveloped.In
(Selenta & Lord, 2005). Alphas for the independent, relational, slopes-as-outcomesmodels, aLevel 2 variableishypothesizedto
and collective self-construals in this study were.79, .71, and .62, moderate the effects of a set of Level 1 variables on a given out-
respectively,consistentwithpreviousresearch(Johnson&Chang, come of interest. In the current research, the independent, rela-
2006;Johnsonetal.,2006). tional,andcollectiveself-construalsatLevel2werehypothesized
to moderate the effects of the three apology components on for-
Manipulationcheckforapologycomponents givenessatLevel1.Threeseparatemodelswerebuiltandtested.
Following previous policy capturing research (Rotundo & Ineachmodel,forgivenesswasregressedonallthreeapologycom-
Sackett,2002),eachapologycomponentwasratedforitsdistribu- ponentsatLevel1,withgender(asacontrol)andoneofthethree
tionalequivalencetoensurethatthecomponentstappedintotheir self-construalsenteredatLevel2.Ineachmodel,asignificantmod-
underlying constructs with equal strength, but did not overlap erating effect of self-construal (e.g. the independent self) on the
with the other components’ constructs. At the end of the study, relationship between the theoretically relevant apology compo-
participantswereaskedtoratethedegreetowhicheachapology nent (e.g. an offer of compensation) and forgiveness should be
manipulationrepresentedanofferofcompensation,anexpression observed(seeTable6).
of empathy, and an acknowledgment of a violated rule/norm Consistentwiththefindingsfromthedirectassessmentstudy,
(1=NotAtAll,7=Completely).Eachcomponentwasratedsimilarly supportwasfoundforeachprimaryhypothesis.Inmodel1,victim
intermsofitsstrength.Fortheofferofcompensation,M=6.17;for independentself-construalwasenteredalongwithgenderatLevel
theexpressionofempathy,M=6.19;fortheacknowledgmentofa 2 and each apology component at Level 1. As predicted, victim
violatedrule/norm,M=6.30.Aseriesoft-testsconfirmedthateach independent self-construal significantly predicted the effect of
manipulation tapped into its intended construct more than the offenders’ offers of compensation on victim forgiveness c=.15,
othercomponents’constructs. t(168)=2.01, p<.05. In model 2, victim relational self-construal
was entered along with gender at Level 2 and each apology
Manipulationcheckforseverity component at Level 1. Consistent with hypothesis 2, victim rela-
Subjective offense severity was assessed with four items tionalself-construalpredictedtheimpactofoffenders’expressions
(1=NotAtAll,7=Completely;a=.93).Anexampleitemis‘‘Howse- of empathy on victim forgiveness, c=.20, t(168)=2.17, p<.05.
vere were the consequences of Pat’s transgression against you?” Finally,inmodel3,thecollectiveself-construalwasenteredalong
Theresults supportedtheeffectiveness of theseveritymanipula- with gender at Level 2, and the apology components were again
tion, with participants indicating subjective levels of severity enteredatLevel1.Consistentwithhypothesis3,victimcollective
of 3.20, 4.32, and 5.19 for mild, unspecified, and severe harm, self-construal significantly predicted the effect of offenders’
R.Fehr,M.J.Gelfand/OrganizationalBehaviorandHumanDecisionProcesses113(2010)37–50 45
Table4
Level2descriptivestatisticsandzero-ordercorrelations,Study2.
M SD 1 2 3 4 5 6
1.Gendera 1.75 .43 –
2.Independentself 3.10 .77 (cid:2).07 (.79)
3.Relationalself 4.43 .47 .20* (cid:2).12 (.71)
4.Collectiveself 3.98 .53 .26** .03 .48** (.62)
5.Harmseverityb 2.00 .83 (cid:2).03 .02 .16* .11 –
6.Forgiveness 4.51 1.09 (cid:2).07 (cid:2).01 (cid:2).03 (cid:2).15* (cid:2).43** (.84)
Note.N=171.
a Male=1,female=2.
b Mildharm=1,controlcondition=2,severeharm=3.
* p<.05.
** p<.01.
Table5
Level1descriptivestatisticsandzero-ordercorrelations,Study2.
M SD 1 2 3 4 5 6 7 8 9
1.Gendera 1.75 .43 –
2.Independentself 3.10 .77 (cid:2).07** (.79)
3.Relationalself 4.43 .47 .20** (cid:2).12** (.71)
4.Collectiveself 3.98 .52 .26** .03 .48** (.62)
5.Compensationb .50 .50 .00 .00 .00 .00 –
6.Empathyb .50 .50 .00 .00 .00 .00 .20** –
7.Rule/normb .50 .50 .00 .00 .00 .00 (cid:2).20** (cid:2).20** –
8.Harmseverityc 2.00 .82 (cid:2).03 .02 .16** .11* .00 .00 .00 –
9.Forgiveness 4.51 1.49 (cid:2).05* (cid:2).01 (cid:2).03 (cid:2).11** .37** .27** .05 (cid:2).31** (.84)
Note.N=1710.
a Male=1,female=2.
b 0=cuenotincluded,1=cueincluded.
c Mildharm=1,controlcondition=2,severeharm=3.
* p<.05.
** p<.01.
which show the interactive effects of self-construal at Level 2
Table6
Hierarchicallinearmodelinganalyses–forgiveness,Study2.a and apologies at Level 1 on forgiveness. These graphs were
produced in HLM, which allows for the visual depiction of such
Variable Coefficient SE t
cross-leveleffects(e.g.Loi,Yang,&Diefendorff,2009).
Slopes-as-outcomesmodelforindependentself-construal A final set of analyses were conducted to test the impact of
Intercept (cid:2).018 .109 (cid:2).16
Compensation .152* .075 2.01 harmseverityonthemoderatingroleofself-construal.Specifically,
Empathy .029 .055 .52 threemodelstestedforpotentialthree-wayinteractionsbetween
Rule/norm .073 .059 1.23 harm severity, victims’ self-construals, and apology components.
Slopes-as-outcomesmodelforrelationalself-construal In each model, harm severity and a focal self-construal were en-
Intercept (cid:2).049 .183 (cid:2).27 tered at Level 2 along with the interaction term (with gender as
Compensation .060 .128 .47 acontrol),andallthreeapologycomponentswereenteredatLevel
Empathy .199* .091 2.17
1. Across all three models, the three-way interaction was non-
Rule/norm .230* .098 2.35
significant, indicating that harm severity neither accentuates nor
Slopes-as-outcomesmodelforcollectiveself-construal
mitigates the relationship between self-construal and apology
Intercept (cid:2).297 .164 (cid:2)1.81
component weightings. It should be noted that harm severity
Compensation .113 .116 .98
Empathy .084 .084 1.01 did,however,exhibitadirectnegativecorrelationwithforgiveness
Rule/norm .184* .089 2.07 acrossparticipants,r(171)=(cid:2).43,p<.01(seeTable4).Thus,while
the relationship between forgiveness and harm severity in this
Note.N=171.Degreesoffreedomforallresults=168.
aAllresultsarecontrollingforgender. studyisconsistentwithpreviousresearch,thedatadonotindicate
*p<.05. any three-way interaction between harm severity, self-construal,
andapology,suggestingthattheroleoftheselfintheapologypro-
cessisnotboundbyharmseverity.
acknowledgments of violated rules/norms on victim forgiveness,
c=.18, t(168)=2.07, p<.05. No anomalous effects were found,
with the exception of the relationship between the relational Generaldiscussion
self-construalandacknowledgmentofaviolatedrule/norm,which
wassignificantp<.05.Insum,theseresultsindicatethatvictims’ Asamethodofconflictresolution,apologieshaveperhapsnever
forgivenessratingsaredeterminednotonlybythecontentofthe been as popular as they are today. In 2007, The New York Times
apologiesthattheyreceive,butalsobythedegreethattheseapol- alonepublishedover500articlesonapology,coveringeverytopic
ogiesarecongruentwiththeirself-construals.Thus,theremaybe fromSouthwestAirline’s‘‘ChiefApologyOfficer”,whomails20,000
noone‘‘best”apology.Rather,theresultsindicatethatforgiveness apology letters to dissatisfied customers each year, to Rolling
ismaximizedwhenoffenders’apologiesaretailoredtoindividual StonesguitaristKeithRichard’sdemandthataSwedishnewspaper
victims’self-construals.ThesefindingsareillustratedinFigs.1–3, apologizeforitspoorreviewofoneofhisconcerts(Bailey,2007;
46 R.Fehr,M.J.Gelfand/OrganizationalBehaviorandHumanDecisionProcesses113(2010)37–50
5.5 5.5
No Compensation
Compensation
5 5
s
ess nes
n e
orgive 4.5 Forgiv 4.5
F
4 No Acknowledgement
4
Acknowledgement
3.5
3.5 1 3.5 5
1 3.5 5
Collective Self-Construal
Independent Self-Construal
Fig.3. Moderatingeffectofthecollectiveself-construalontherelationbetween
Fig.1. Moderatingeffectoftheindependentself-construalontherelationbetween
acknowledgmentsofviolatedrules/normsandforgiveness.
offersofcompensationandforgiveness.
beyond the basic question of if an apology has been offered and
5.5 ask ‘‘Which components of apologies are most effective, and for
No Empathy whom?”
Empathy InStudy1,therelationshipbetweenself-construalandapology
perceptions was examined via direct assessment. Factor analysis
5 providedstrongevidenceforthedistinctivenessofthethreefocal
apology components, while a series of regressions confirmed the
ss primary hypotheses. In Study 2, results from a policy capturing
e
n experiment converged with the findings from Study 1. Each self-
e
v 4.5
gi construal was shown to strengthen the effect of its congruent
or apology component on victim forgiveness. An exception was the
F
findingthatthe relational self-construalstrengthenstheeffect of
acknowledgmentsofviolatedrules/normsinStudy2.Onereason
4
for this effect may be that suchacknowledgments highlight feel-
ings of empathy and closeness to others within group contexts,
whererulesandnormsaremutuallysharedandemphasized.More
3.5 generally,theresultsalignwithalonghistoryofresearchshowing
1 3.5 5 thatpeoplepreferinformationthatisconsistentwithandverifies
Relational Self-Construal theirownself-views(e.g.Swann,1987).InStudy2,theimpactof
harmseveritywasalsotested.Competingevidencewaspresented
Fig.2. Moderatingeffectoftherelationalself-construalontherelationbetween
tosuggestthatharmseveritymaystrengthen,mitigate,orexerta
expressionsofempathyandforgiveness.
null effect on the impact of apologies and self-construal on
forgiveness.Harmseveritywasnotfoundtoexhibitanysignificant
‘‘Keith Richards,” 2007). Indeed, apologies have become popular moderating effect, thus providing evidence to suggest that the
enough to move at least one scholar to refer to the modern era interactiveeffectsofapologiesandself-construalarerobust.
as‘‘TheAgeofApology”(Brooks,1999).Despitethemanypotential
benefitsofapologies,itisclearthatpeoplearenotalwayssatisfied Practicalimplications
withtheapologiestheyreceive.Forinstance,whilesomeMuslims
acceptedPopeBenedict’sapologyforhiscriticalcommentsoftheir To date, the tendency to treat apologies as dichotomous has
faithin2006,othersdenounceditfornotconstitutinga‘‘fullapol- hamperedscholars’abilitytoofferpracticaladviceonhowtobest
ogy” (Cooperman, 2006). In China, Mattel executives received apologize to victims. The present research addresses this short-
widespread criticism for their apology following a recall of toys comingbyofferingamorepreciseaccountofhowapologiesshould
deemeddangerousfortheirleadcontent(‘‘PlentyofBlame,”2007). bestructured.Independently,allthreeexaminedapologycompo-
Inherentintheseexamplesisarecognitionthatapologiesmust nentsexhibitedsignificant andpositiveeffectsonvictimforgive-
includethe specific componentsthat the victim needs to hear to be ness.Thesefindingssuggestthatdetailedapologieswithmultiple
truly effective. By ignoring the specific components of apologies, components are in general more likely to touch upon what is
researchers risk an oversimplified understanding of the apology importanttoavictimthanbrief,perfunctoryapologies.Offenders
process.Shoulditbeinferredthatthephrase‘‘I’msorry,I feelso shouldthereforeofferapologieswithmultiplecomponentswhen-
upsetaboutwhathappened”isanalogoustoasimple‘‘I’msorry”? ever possible. However, apologies also exhibited effects that
What if an apologetic CEO, following an accounting scandal, also hinged upon victim self-construals. Victims with relational self-
offerstocompensatestakeholdersmonetarily?Willallvictimsre- construals, for example, were particularly forgiving following
spond to these apologies uniformly? In theory, the content of an expressionsofempathy.Thus,thereisalsoaclearneedforoffend-
apology should influence how effective it is, and who it is most ersto‘‘considertheiraudience”whenofferingapologies.Thisneed
effective for. The purpose of this paper, therefore, was to move tometa-cognizeaboutwhatavictimislookingforinanapologyis