Moral Identity and Adolescent Prosocial and Antisocial Behaviors

13 Pages·2014·0.33 MB·English

JYouthAdolescence
DOI10.1007/s10964-014-0172-1
EMPIRICAL RESEARCH
Moral Identity and Adolescent Prosocial and Antisocial
Behaviors: Interactions with Moral Disengagement
and Self-regulation
Sam A. Hardy • Dallas S. Bean • Joseph A. Olsen
Received:15July2014/Accepted:13August2014
(cid:2)SpringerScience+BusinessMediaNewYork2014
Abstract Moral identity has been positively linked to Keywords Moral identity (cid:2) Moral disengagement (cid:2) Self-
prosocial behaviors and negatively linked to antisocial regulation(cid:2)Self-control (cid:2)Moralmotivation(cid:2)Interactions (cid:2)
behaviors; but, the processes by which it is linked to such Prosocial (cid:2) Antisocial
outcomesareunclear.Thepurposeofthepresentstudywas
toexaminemoralidentitynotonlyasapredictor,butalsoasa
moderatorofrelationshipsbetweenotherpredictors(moral Introduction
disengagement and self-regulation) and youth outcomes
(prosocialandantisocialbehaviors).Thesampleconsistedof Socialscientistshavelongbeeninterestedinprosocialand
384 adolescents (42 % female), ages 15–18 recruited from antisocial behaviors. Prosocial behaviors are helpful to
acrosstheUSusinganonlinesurveypanel.Latentvariables relationships, communities, and society, while antisocial
were created for moral identity, moral disengagement,and behaviors are harmful. A number of individual and con-
self-regulation. Structural equation models assessed these textual predictors of prosocial and antisocial behaviors
latentvariables,andinteractionsofmoralidentitywithmoral have been identified, and one individual-level predictor of
disengagementandself-regulation,aspredictorsofprosocial increasing interest is moral identity (Hardy and Carlo
(charity and civic engagement) and antisocial (aggression 2011). Evidence is mounting that moral identity may
andrulebreaking)behaviors.Noneoftheinteractionswere motivate people to engage in prosocial behaviors and
significant predicting prosocial behaviors. For antisocial abstain from antisocial behaviors, but the nature of links
behaviors,theinteractionbetweenmoralidentityandmoral between moral identity and such behaviors remains
disengagement predicted aggression, while the interaction unclear. Perhaps moral identity not only directly predicts
betweenmoralidentityandself-regulationwassignificantin behaviors, but also moderates links between other social
predicting aggression and rule breaking. Specifically, at cognitions (e.g., moral disengagement and self-regulation)
higher levels of moral identity, the positive link between andbehavior.Thus,thepurposeofthepresentstudywasto
moral disengagement and aggression was weaker, and the assess moral identity as a moderator of relationships
negative link between self-regulation and both antisocial between predictors (moral disengagement and self-regula-
behaviors was weaker. Thus, moral identity may buffer tion) and outcomes (prosocial and antisocial behaviors)
againstthemaladaptiveeffectsofhighmoraldisengagement among adolescents.
andlowself-regulation.
Moral Identity
S.A.Hardy(&)(cid:2)D.S.Bean
In general, moral identity is the degree to which being a
DepartmentofPsychology,BrighamYoungUniversity,Provo,
UT84602,USA moral person is important to an individuals’ identity.
e-mail:[email protected] However, there are a number of possible ways of opera-
tionalizing this construct. While some researchers use
J.A.Olsen
sophisticated qualitative methods for capturing moral
CollegeofFamily,Home,andSocialSciences,BrighamYoung
University,Provo,UT84602,USA identity (e.g., Frimer and Walker 2009), most use
123
JYouthAdolescence
self-report questionnaire measures that more efficiently (Blasi1993).Participantsrateasmallsetofmoraltraits,oneat
assessmoralidentityinlargesamples.However,therearea atime,intermsofhowimportantitisforthemtobeaperson
number of self-report formats in use that may capture withthattrait.Collegestudentshigheronmoralself-relevance
moral identity from different angles. We will review sev- engageinmoreprosocialbehaviors(Hardy2006).Similarly,
eralofthemorecommonwaysofmeasuringmoralidentity adolescentswithhighermoralself-relevancehavelowerself-
using self-report scales, and then use these various serving cognitive distortions (e.g., misattributions of hostile
approaches to assess moral identity in the present study. intent;Barrigaetal.2001),lowerlevelsofantisocialbehavior
First, the most widely used moral identity scale was (Barrigaetal.2001),moreinternalmotivations to bemoral
developed by Aquino and Reed (2002), and contains two (Krettenauer 2011), and heightened feelings of moral
subscales:moralidentityinternalizationandmoralidentity responsibility(Krettenauer2011).
symbolization.AquinoandReeddefinemoralidentityasthe Third, a new approach to capturing moral identity taps
extent towhichbeingapersonwith moraltraits isasocial the extent to which moral traits are a part of a person’s
identity that is salientto one’s self-concept,with internali- idealself(Hardyetal.2014).Basedonthepossibleselves
zationbeingthecentralityofthemoralpersonsocialidentity literature (Oyserman and James 2011), participants are
toself-concept,andsymbolizationbeingthedegreetowhich askedtothinkaboutthetypeofpersontheyideallywantto
the moral person social identity is expressed in action. As be,andthenrateasetofmoraltraitsintermsofhowmuch
such,participantsarepresentedasmallsetofmoraltraitsand each traits describes that person (i.e., their ideal self).
asked to image a person with those traits while rating a Moral ideal self motivates moral action because the ideal
numberofstatementsabouttheimportanceofthosetraitsto self serves as a goal people strive to approach. Indeed,
themselves. However, most evidence shows that symboli- studies have shown that adolescents with a stronger moral
zation is more about self-presentation concerns than moral idealselfshowmoreempathy,engageinmorealtruismand
concerns, and thus it was not used in the present study. environmentalism, evidence more moral personality traits,
Accordingtothenotionofmoralidentityinternalization,the have fewer symptoms of internalization and externalizing,
moreimportantpeopleseethemoralpersonsocialidentityas and are less aggressive (Hardy et al. 2012, 2014).
beingtotheirself-concept,themoremotivatedtheyaretobe Inadditiontothesescalesspecificallydesignedtoassess
amoralperson,andthemorecognitivelyaccessiblemoral- moral identity, there are other measures that can be lev-
person concepts are for processing information in social eraged for capturing moral identity. For instance, Crocker
situations.Assuch,studieshaveshownthatadultshigheron and colleagues (Crocker et al. 2003) developed a scale for
moralidentityinternalizationhavehigherlevelsofsympathy assessing the extent to which self-worth is contingent on
and moral reasoning (Aquino and Reed 2002), are more various factors. Onesuch factorisaperson’smoralvirtue.
likely to volunteer in the community and donate food Peoplewanttomaintainapositivesenseofself-worth,and
(Aquino andReed 2002),have heightenedmoral elevation are thus motivated to live consistent with their contingen-
(i.e., positive reactions to witnessing ‘‘uncommon acts of cies of self-worth (Crocker et al. 2003). College students
moralgoodness,’’suchaspositiveemotions,positiveviews for whom moral virtue is a stronger contingency for their
ofhumanity,anddesiretobeabetterperson;Aquinoetal. self-worth are more agreeable and conscientious, more
2011),havegreatermoralconcernsforout-groupmembers likely tospend time volunteering andengaginginspiritual
(ReedandAquino2003),aremorelikelytomakeprosocial activities, and less likely to spend time grooming and
business decisions (Aquino et al. 2009) and less likely to partying (Crocker et al. 2003). Similarly, they engage in
makedishonestbusinessdecisions(Aquinoetal.2009),and less alcohol use (Lewis et al. 2007).
morelikelytohavegreaterself-esteemandmeaning,lower Lastly, Cheek and colleagues (Cheek et al. 1985)
levelsofanxietyanddepression,andlowerratesofhazard- developed a measure for assessing the relative importance
ous alcohol use and sexual risk-taking (Hardy et al. 2013). ofvariousaspectsofidentitytoaperson’soverallsenseof
Similarly,adolescentshigheronmoralidentityinternaliza- identity. One subscale captures aspects of identity associ-
tionhavegreatermoralconcernforout-groupmembersand atedwithpersonalidentity,andincludesanitemregarding
less social dominance attitudes (i.e., attitudes of being the importance of personal values and moral standards.
superiortoothers;Hardyetal.2010). Again, people want to live consistent with their sense of
Second, another scale for assessing moral identity was identity (Blasi 1993), so the more important various iden-
developedbyGibbsandcolleagues(Barrigaetal.2001)and tity aspects are to a person’s overall sense of identity, the
adapted by others (e.g., Hardy 2006). This measure taps more motivation they have in that domain. To our
‘‘moralself-relevance,’’whichistheimportancepeopleplace knowledge, no studies have specifically focused on the
on seeing themselves as someone with moral traits. Moral importance of personal values and moral standards as an
self-relevance motivates moral action because people are aspect of identity. However, the more people see personal
inherently driven to live consistent with their self-concept identity issues (a broader category of aspects of identity
123
JYouthAdolescence
that includes values and standards) as important to their Whereas moral disengagement is a maladaptive social
identity, the greater integrity they show (Schlenker 2008). cognition, self-regulation is an adaptive social cognition.
Self-regulation is defined as the psychological capacity to
Moral Identity as a Moderator refrain from or override short-term, selfish motives and
impulsive actions, and enact behaviors that are consistent
In addition to moral identity motivating behavior, it might with one’s long-term goals (Tangney et al. 2004). Addi-
also moderate links between other social cognitive factors tionally, self-regulation is a multi-facetted construct, con-
andbehaviors.Inotherwords,itmayhaveadirecteffectas sisting of the ability to inhibit undesired responses and
well as a synergistic interaction effect with other social activate desired responses (Capaldi and Rothbart 1992).
cognitions. Two such social cognitions examined in the Indeed, youth with greater self-regulatory capacity engage
literature are moral disengagement and self-regulation. in more prosocial behaviors (Carlo et al. 2012), fewer
Moral disengagement entails strategies people con- externalizing behaviors (e.g., aggression and delinquency;
sciouslyorunconsciouslyusethatenablethemtoengagein Doan et al. 2012) and less risky drinking and sex (Quinn
immoralactionswhileretainingtheirviewofthemselvesas and Fromme 2010). Thus, self-regulation is an important
moralpeople(Banduraetal.2001).Inotherwords,people part of adaptive psychosocial functioning.
execute various mechanisms (e.g., moral justification and Withsomuchridingonself-regulation,peoplewithlow
advantageous comparison) in order to reconstruct the self-regulation,eitherconsistentlyorsituationally,areatrisk
meaning of their conduct in a way that justifies their sup- for poor outcomes. However, just as moral identity can
portfororperpetrationofimmoralactswhilemaintaininga moderate the negative effects of moral disengagement, it
positive self-image. Indeed, youth more likely to morally mightalsobufferagainstself-regulatorydeficitsorboostthe
disengage are also more likely to participate in antisocial effectsofadequateself-regulation.Toourknowledge,only
behaviors (Hyde et al. 2010), including aggression (Gini one study has specifically examined interactions between
et al. 2014), bullying (Thornberg and Jungert 2013), and self-regulationandmoralidentityinpredictingactions(Gino
alcohol use (Newton et al. 2014). Further, teens higher on etal.2011).Ginoandcolleaguestheorizedthatpeoplewith
moral disengagement are less likely to help others if it is stronger identity commitments to morality would need to
not in their own personal interest to do so (Paciello et al. expend less cognitive resources to regulate their ethical
2013). Thus, moral disengagement is not conducive to behavior. They examined the effects of self-regulation on
positive psychosocial development and functioning. unethicalbehavioramongadultsandfoundthatwhenself-
Bandura and colleagues argue that the processes of regulatory resources are depleted by prior exertion of self-
moral disengagement largely result from situational pres- control, unethical behavior increases. More importantly,
sures rather than personality. This emphasis on the power highlevelsofmoralidentityweakenedthelinkbetweenself-
of the situation resonates with well-known social psycho- control depletion and unethical behavior. In other words,
logical research on obedience and authority by Stanley peoplewithastrongsenseofmoralidentitysawlessnegative
Milgram,PhillipZimbardo,andothers,suggestingthatbad sequella from a temporary deficit in self-regulation; they
things happen due to bad situations, not bad people dependedlessonself-regulationtoactmorally.
(Zimbardo 2007). Nevertheless, there is evidence that
personality factors can mitigate such situational pressures Moral Identity During Adolescence
(Burger2009).Inotherwords,evenifsituationalpressures
fan the flames of moral disengagement, there are likely Although studies have examined adolescent outcomes of
contravening personality characteristics that counterac- moral identity (Hardy and Carlo 2011), no studies to our
tively dampen the flames. For instance, among adults, knowledgehavebeenconductedonadolescentstoassessthe
Aquino et al. (2007) tested whether moral identity would role of moral identityas a moderator between social cogni-
moderate the degree to which moral disengagement pre- tionsandbehaviors.But,thelackofadolescentstudiesinthis
dictedaggressiontowardthoseresponsiblefor9/11attacks. areadoesnotinandofitselfwarrantitsstudy.Rather,thereare
They reported that moral identity weakened the negative other more empirical and substantive reasons to want to
link between moral disengagement and aggression. Simi- unpacktherolesofmoralidentityduringadolescence.First,
larly, in another study of adults, attachment security theteenyearsseeheightenedratesofprosocialandantisocial
weakened the positive association between moral disen- behaviors(Veenstra2006).Thus,highmoraldisengagement
gagement and stealing behaviors (Chugh et al. 2014). and low self-regulation during adolescence can exacerbate
Therefore,moralidentity(apersonalitycharacteristic)may ratesofantisocialbehaviorandsquelchpotentialprosociality.
moderate links between moral disengagement and behav- Moral identitymay,therefore, helpteens tobethebestthat
iors such that moral disengagement has less maladaptive they can be, by minimizing the negative sequella of moral
effects for people higher on moral identity. disengagementandself-regulatorydeficits.Second,although
123
JYouthAdolescence
theseedsofmoralidentitymaybeplantedearlyinchildhood, (N = 384; ages 15–18 years, M = 16.28, SD = .97; 58 %
moral identity per sedoesnot emergeuntil adolescents and Male; 70 % European American, 12 % African American,
youngadulthood(HardyandCarlo2011).Itissuspectedthat 10 % Hispanic, and 8 % other; 27 % Protestant, 20 %
thistimingisduetothematurationofmoralityandidentityin Catholic, 20 % Non-Denominational Christian, 19 % no
becomingmoreideologically-basedduringadolescence,thus affiliation,8 %agnosticoratheist,6 %other).Thefamilies
peoplebegintodefinethemselveswiththeirmoralidealsand inourstudycamefrom45ofthestatesintheUS,53 %had
commitments. So, this potentially important capacity (i.e., annualhouseholdincomesunder$50,000,and53 %ofthe
moral identity) to mitigate problems with moral disengage- teens were living with both of their biological or adoptive
ment and self-regulation becomes more available during parents.Oftheparentswhoprovideddata(n = 325),54 %
adolescence,justwhenitisneededmost. were mothers, and 56 % did not have a college degree.
AdultsintheUSwithadolescentchildrenbetween15and
18 yearsoldwereinvitedtoparticipatethroughanemailsentby
The Present Study Survey Sampling International (SSI; www.surveysampling.
com). SSI recruits participants from websites, social media,
The purpose of the present study was to examine whether surveypanels,andothersources.Parentswhowereinterested
moralidentitywouldmoderateassociationsbetweensocial weredirectedtoawebpageprovidinginformationaboutthe
cognitive predictors and youth outcomes, in addition to studyandaskingforconsentfortheiradolescenttoparticipate.
being a predictor of such outcomes. For behavioral out- Ifparentalpermissionwasgiven,theparentwasaskedtohave
comes, we examined two prosocial behaviors (charity and theiradolescenttakehisorherportionofthesurveyinprivate.
civic engagement) and two antisocial behaviors (aggres- Uponcompletionofthesurvey,theywereaskedtohavetheir
sion and rule-breaking), to provide multiple indexes of parent(theinitialcontact)taketheparentportionofthesurvey
positive and negative behaviors. This was because the (theparentsdidnothaveaccesstotheadolescents’responses).
processes may be different for prosocial and antisocial Families who reached the end of the survey received com-
behaviors. As reviewed above, moral identity, moral dis- pensationroughlyintheamountof$4perfamily,butthetypeof
engagement, and self-regulation have all been linked to compensation varied depending on the how the participants
variousprosocialandantisocialbehaviorsinpriorresearch. wererecruitedbythesurveypanelandtheparticipants’pref-
We assessed three hypotheses. First, moral identity and erences for mode of compensation (e.g., cash, points, prizes,
self-regulation will positively predict the prosocial youth sweepstakes,orcharitabledonationsintheirname).Thepresent
outcomes and negatively predict the antisocial youth out- studyincludedbothadolescentandparent-reportmeasures.
comes,whiletheassociationswiththeoutcomeswillbethe
inverse for moral disengagement. This is to validate and Measures
extendpriorworkonthemaineffectsofmoralidentity,self-
regulation, and moral disengagement on behaviors in ado- The three social cognitive variables (moral identity, moral
lescents and adults. Second, moral identity will moderate disengagement, and self-regulation) were assessed using
relationships between moral disengagement and the youth adolescent self-report measures. This is because adoles-
outcomes.Specifically,weanticipatethatindividualshigher cents are likely the best source of information regarding
on moral identity will see weaker links between moral dis- their own social cognitions (Waters et al. 2003). On the
engagement and the outcomes because moral identity can other hand, all of the outcome behaviors were assessed
potentiallymitigatetheinfluenceofsocialpressuresthatignite usingparent-reportmeasures.Usingparentreportsreduces
moral disengagement. Third, moral identity will moderate concerns over common method variance between predic-
associationsbetweenself-regulationandtheyouthoutcomes. tors and outcomes (Lewis et al. 2012).
Specifically, teens higher on moral identity will see weaker
linksbetweenself-regulationandtheoutcomes,presumably Moral Identity
becausetheyarelessdependentonself-regulatorycapacities
toactivateprosocialactionsandinhibitantisocialactions. Moral identity was assessed using five different measures.
For each of the four multi-item scales, the items were
averaged tocreatescalescores.Thesefourscalescores,as
Method wellasthefifthmeasureofmoralidentitythatwasasingle
item, were used as five indicators (a = .82) of a latent
Sample moral identity construct.
The sample consisted of adolescents from across the US Moral Internalization The first measure was Aquino and
recruited online through Survey Sampling International Reed’s (2002) five-item (a = .80) moral identity
123
JYouthAdolescence
internalization scale, which assesses the extent to which statements (sample item: ‘‘Kids cannot be blamed for
beingsomeonewithmoraltraitsiscentraltoone’spersonal misbehavingiftheirfriendspressuredthemtodoit’’)from
identity. For this measure, participants were prompted to 1(stronglydisagree)to7(stronglyagree).Eightcomposite
envisionapersonwithmoraltraits(caring,compassionate, subscale scores were created by averaging the items for
fair, friendly, generous, helpful, hardworking, honest, and eachoftheeightfacetsofmoraldisengagement,andthose
kind) and asked to statements about those traits (sample composites were used as eight indicators (a = .95) of a
item: ‘‘Being someone who has these characteristics is an latent moral disengagement variable.
importantpartofwhoIam’’)onascalefrom1(completely
disagree) to 7 (completely agree). Self-regulation
Moral Self-relevance The second measure included 8 Self-regulationwasassessedusing10items(a = .83)from
items (a = .94) adapted from Gibbs and colleagues’ (Bar- the Early Adolescent Temperament Questionnaire—
rigaetal.2001)moralself-relevancescale,whichcaptures Revised (EATQ-R; Ellis and Rothbart 2001; see also,
theimportanceofvariousmoraltraitstoself-concept.Each Capaldi and Rothbart 1992)—five-items from the Activa-
itementailedhavingparticipantsrateamoraltraitintermsof tion Control subscale (sample item: ‘‘I can stick with my
importancetotheselfonascalefrom1(notimportanttome) plans and goals’’) and five items from the Inhibitory
to 5 (extremely important to me; sample item: ‘‘How Control subscale (sample item: ‘‘I have a hard time fin-
importanttoyouisitthatyouarehonest’’). ishing things on time’’; reverse-coded). Adolescents rated
the truthfulness of each statement about their ability to
Moral Ideal Self The third measure was Hardy and col- overrideimmediateimpulsestoachievelong-termgoalson
leagues’(Hardyetal.2014)20-item(a = .97)MoralIdeal a five-point scale ranging from 1 (almost always untrue of
Self Scale, which assesses the extent to which a person’s you)to5(almostalwaystrueofyou).Theitemswereused
ideal self is moral. Participants rated 20 moral traits (e.g., as indicators of a latent self-regulation variable.
generous, truthful, follows values, respectful, and good
example)intermsoftheextenttowhichthosetraitsdescri- Charity
bedthetypeofpersontheywanttobe(i.e.,theiridealself),
usingascalefrom1(notatall)to7(verymuch). Involvement in charitable activities was measured using 6
items (a = .86) from the Youth Inventory of Involvement
Moral Contingencies of Self-worth The fourth measure scale(Panceretal.2007).Parentsratedthefrequencyatwhich
included5items(a = .81)fromtheContingenciesofSelf- their adolescenthadengaged in various charitableactivities
Worth Scale (Crocker et al. 2003) that pertain to the (sampleitem:‘‘Ivisitedorhelpedoutpeoplewhoweresit’’)in
importance of living virtuously to one’s self-esteem thepastyearonascalefrom1(youradolescentneverdidthis)
(sampleitem:‘‘Myself-esteemdependsonwhetherornotI to5(youradolescentdidthisalot).Theaverageoftheseitems
follow my moral/ethical principles’’), rated from 1 wasusedasanobservedvariableintheanalyses.
(strongly disagree) to 5 (strongly agree).
Civic Engagement
Moral Aspects of Identity The fifth measure of moral
identitywasasingleitem(‘‘Mypersonalvaluesandmoral
Involvement in civic engagement was measured using 11
standards’’)regardingtheimportanceofpersonalvaluesand
items (a = .90) from the Youth Inventory of Involvement
moralstandardsastheyrelatetoone’sidentity,takenfrom
scale (Pancer et al. 2007). Parents rated the frequency at
Cheek,UnderwoodandCutler’s(1985)AspectsofIdentity
which their adolescent had engaged in various political or
Scale,andratedfrom1(notatallimportanttomysenseof
community involvement activities (sample item: ‘‘I par-
whoIam)to5(extremelyimportanttomysenseofwhoIam).
ticipated in a political party, club, or organization’’) in the
pastyearonascalefrom1(youradolescentneverdidthis)
Moral Disengagement
to 5 (your adolescent did this a lot). The average of these
items was used as an observed variable in the analyses.
Moral disengagement was measured using a 32-item
measure developed by Bandura and colleagues (Bandura
et al. 1996) that assesses eight facets of moral disengage- Aggression
ment: moral justification, euphemistic language, advanta-
geous comparisons, displacement of responsibility, Involvement in aggressive acts was measured using 18
diffusion of responsibility, distorting of consequences, items (a = .92) from the Child Behavior Checklist
attributionofblame,anddehuminization.Participantsrated (School-Age version; Achenbach and Rescorla 2001).
123
JYouthAdolescence
Parents rated the extent to which various aggressive study variables (latent variables for moral disengagement,
behaviors (sample item: ‘‘Gets in many fights’’) reflect the self-regulation and moral identity, along with observed
behaviorsoftheiradolescentonascalefrom1(nottrue)to variables for charity, civic engagement, aggression, and
3(very true oroften true).Theaverage ofthese itemswas rule breaking), and all of the covariances among these
used as an observed variable in the analyses. variables. Given that several study variables were skewed
(i.e., the advantageous comparisons moral disengagement
Rule Breaking subscale, the aggression composite, and the rule breaking
composite were all positively skewed above 2.0), this
Involvementinrulebreakingwasmeasuredusing17items model was estimated using MLR (maximum likelihood
(a = .86) from the Child Behavior Checklist (School-Age estimation with robust standard errors) which accounts for
version; Achenbach and Rescorla 2001). Parents rated the non-normal variable distributions. Two of the 10 observed
extent to which various rule breaking behaviors (sample indicators of the self-regulation latent variable had factor
item: ‘‘Lying or cheating’’) reflect the behaviors of their loadingsbelow.4andwerethussubsequentlydropped.The
adolescent on a scale from 1 (not true) to 3 (very true or revised model fit the data moderately well, v2
often true). The average of these items was used as an (258) = 639.58, p = .0001, CFI = .91, RMSEA = .06.
observed variable in the analyses. Standardized factor loadings for the latent variables were
Analysis Plan
Table1 Estimated means and standard deviations of observed
variables
The hypotheses were assessed via structural equation
modeling(SEM)withlatentvariablesusingMplus(version Observedvariables Range M SD
7.11)statisticalsoftware.Modelparameterswereestimated
Moralidentityindicators
using Full Information Maximum Likelihood estimation,
Moralidentityinternalization 1–7 6.10 .94
which includes in the analyses all cases with data on at
Moralself-relevance 1–5 4.09 .76
leastonevariable.Asindicatorsofmodelfit,weusedRoot
Moralidealself 1–7 6.22 .82
Mean Squared Error of Approximation (RMSEA: values
Contingenciesofself-worth 1–7 5.38 1.04
below.05indicategoodfit,below.08indicatemoderatefit,
Aspectsofidentity 1–5 4.28 .79
and below .10 indicate mediocre fit), and the Comparative
Moraldisengagementindicators
Fit Index (CFI; values above .95 indicate good fit, and
Moraljustification 1–5 2.34 .87
values above .90 indicate moderate fit).
Euphemisticlanguage 1–5 1.68 .75
First, the measurement model was established by esti-
Advantageouscomparisons 1–5 1.41 .71
matingaconfirmatoryfactoranalysis(CFA)modelwithall
Displacementofresponsibility 1–5 1.90 .84
study variables. Second, to examine the study hypotheses
Diffusionofresponsibility 1–5 1.81 .82
we estimated a series of full structural models. Initial
Distortingconsequences 1–5 1.60 .72
structural models examined main effects of moral disen-
Attributionofblame 1–5 1.86 .80
gagement, self-regulation, and moral identity on the out-
Dehumanization 1–5 1.72 .85
comes. These were followed by models including
Self-regulationindicators
interactions between the predictors (moral disengagement
Inhibitorycontrol—item1 1–5 3.74 1.11
and self-regulation) and the moderator (moral identity).
Inhibitorycontrol—item2 1–5 3.78 .99
Specifically, the moderation hypotheses were assessed by
Inhibitorycontrol—item3 1–5 3.46 1.06
estimatingeightseparateinteractionmodels;eachpredictor
Activationcontrol—item1 1–5 3.48 1.19
interacted with moral identity in predicting each outcome.
Activationcontrol—item2 1–5 2.62 1.07
Interactions between latent variables were estimated using
the XWITH command and numerical integration in Mplus Activationcontrol—item3 1–5 3.19 1.24
(Muthe´n and Muthe´n 1998–2010). Activationcontrol—item4 1–5 3.49 1.21
Activationcontrol—item5 1–5 3.07 1.30
Scale-score(meansofitems)outcomes
Results Charity 1–5 2.41 1.02
Civicengagement 1–5 1.69 .78
Measurement Model Aggression 0–2 .21 .29
Rulebreaking 0–2 .16 .22
We first estimated a CFA model to establish the mea-
Estimatedmeansandstandarddeviationswereobtainedfromthefinal
surement model. This model included all of the primary confirmatoryfactoranalysismodel
123
JYouthAdolescence
Table2 Estimatedbivariate
1 2 3 4 5 6
correlationsbetweenstudy
variables 1 Moralidentity
2 Moral -.54***
disengagement
3 Self-regulation .53*** -.43***
4 Charity .39*** -.17* .39***
5 Civicengagement .27*** -.01 .40*** .68***
6 Aggression -.39*** .52*** -.57*** -.21*** -.19**
N=379;?p\.10;*p\.05; 7 Rulebreaking -.38*** .52*** -.47*** -.18** -.12? .77***
**p\.01;***p\.001
all statistically significant and sufficiently large in size civicengagement,aggressionandrulebreaking),usingthe
(ranging from .40 to .89). Estimated means and standard MLR estimator, and adding age and gender as covariates.
deviations for all observed variables (observed indicators Age was not significantly related to any of the study out-
of the latent variables as well as the four observed out- comes, and was thus dropped. Gender was predictive of
comes) are presented in Table 1, while estimated bivariate charity (in that girls were more charitable than boys), so it
correlations between study variables are reported in was retained as a control variable in all subsequent anal-
Table 2. Moral identity correlated negatively with moral yses. The model fit was very similar to the CFA reported
disengagement and positively with self-regulation, while above, v2 (279) = 666.24, p = .0001, CFI = .91,
moral disengagement and self-regulation were negatively RMSEA = .06. Standardized coefficients are in Table 4,
associated. Moral identity and self-regulation were both so we only present unstandardized coefficients, standard
positively associated with charity and civic engagement errors,andp-valuesinthetext.Mplusonlyreturnsp-values
and negatively associated with aggression and rule break- tothreedecimalplaces,soincaseswhereitreturned0.000,
ing.Theoppositewastrueformoraldisengagement,which we report it as p = .0001 in the text. In this main effects
associated negatively with charity (but not civic engage- regression model, self-regulation (b = .35, SE = .09,
ment)andpositivelywiththeaggressionandrulebreaking. p = .0001) and moral identity (b = .51, SE = .13,
As an additional preliminary analysis, we examined the p = .0001) positively predicted charity. Interestingly, in
correlations between each of the five moral identity indi- this model all three variables positively predicted civic
cators and the other study variables. This was not in engagement (moral disengagement b = .32, SE = .12,
preparation for the primary analyses but was merely to p = .01; self-regulation b = .39, SE = .06, p = .0001;
provide information regarding the potential relative moralidentityb = .25,SE = .12,p = .04).Giventhatthe
importanceofvariousindicatorsofmoralidentitytomoral bivariate correlation of moral disengagement with civic
functioning. We estimated this model as we did the prior engagement was not significant, this suggests a suppressor
CFA, but included the indicators of moral identity as effect.Specifically,inthecontextofthisregressionmodel,
additional observed variables rather than specifying a the variability in moral disengagement that is not over-
moralidentitylatentvariable(seeTable 3).Thepatternsof lapping with variability in self-regulation and moral iden-
relationshipswithotherstudyvariablesweresimilaracross tity (i.e., the unique variability) is actually positively
all five moral identity indexes. However, to provide a related to civic engagement. Moral disengagement was a
comparisoninTable 3wealsopresentaveragecorrelations positive predictor of both aggression and rule breaking
of each moral identity index with the other moral identity (b = .16, SE = .05, p = .001; b = .13, SE = .05,
indexes, with the two social cognitive variables, and with p = .004), while self-regulation was a negative predictor
thefouroutcomes.Moralself-relevancewasmoststrongly (b = -.15, SE = .03, p = .0001; b = -.08, SE = .02,
correlated with the other moral identity indexes, with the p = .0001). Moral identity did not predict the antisocial
social cognitive variables, and with the outcomes. outcomes.
Structural Models Interactions
Main Effects To test for interactions between the social cognitive pre-
dictors (moral disengagement and self-regulation) and
A single full structural model was estimated with moral moral identity we estimated eight additional structural
disengagement, self-regulation, and moral identity speci- equation models. Each model included one of the two
fied as predictors of the four outcome variables (charity, predictors (moral disengagement or self-regulation), moral
123
JYouthAdolescence
Table3 Estimatedbivariate
Moralidentity Moralself- Moral Contingenciesof Aspectsof
correlationsofthemoral
internalization relevance idealself self-worth identity
identityindicatorswithother
studyvariables 1.Moralidentity 1 .49*** .44*** .46*** .50***
internalization
2.Moralself-relevance .49*** 1 .63*** .50*** .58***
3.Moralidealself .44*** .63*** 1 .42*** .45***
4.Contingenciesofself- .46*** .50*** .42*** 1 .49***
worth
5.Aspectsofidentity .50*** .58*** .45*** .49*** 1
Averageofrows1–5 .47 .55 .49 .47 .51
(moralidentity)
6.Moraldisengagement -.47*** -.39*** -.37*** -.40*** -.34***
7.Self-regulation .32*** .47*** .32*** .34*** .36***
Averageofrows6–7 .40 .43 .35 .37 .35
(socialcognitions)
8.Charity .22*** .34*** .25*** .23*** .30***
9.Civicengagement .14* .24*** .12* .21** .20***
10.Aggression -.25*** -.35*** -.26*** -.29*** -.22***
11.Rulebreaking -.24*** -.33*** -.23** -.30*** -.23***
Averageofrows8–11 .21 .31 .21 .26 .24
N=379;?p\.10;*p\.05;
(outcomes)
**p\.01;***p\.001
Table4 Maineffectsstructuralmodel Table5 Interactioneffectsstructuralmodels
Predictors Outcomes Predictors Outcomes
Charity Civic Aggression Rule Charity Civic Aggression Rule
engagement breaking engagement breaking
b b b b b b b b
Moral .14? .28** .36*** -.38** Moral .07 .21 .36*** .34**
disengagement disengagement
Self-regulation .29*** .42*** -.43*** -.30*** Moralidentity .44*** .37*** -.15* -.16?
Moralidentity .30*** .20* .03 -.02 Interaction -.04 .02 -.18* -.21
Self-regulation .26*** .35*** -.53*** -.41***
Genderwasincludedasacontrolvariable
N=379.?p\.10;*p\.05;**p\.01;***p\.001 Moralidentity .26*** .13 -.09 -.13?
Interaction .05 .14? .27*** .29**
Genderwasincludedasacontrolvariable
identity, and one of the four outcomes (charity, civic
N=379formoraldisengagementmodels,N=374forself-regula-
engagement, aggression, and rule breaking). These latent
tionmodels
interactionmodelswereestimatedusingtheMLestimation
? p\.10;*p\.05;**p\.01;***p\.001
with numeric integration (the MLR estimator is not avail-
ablewithnumericintegrationintheMplussoftware).With aggression, moral disengagement was a positive predictor
numeric integration in Mplus, most fit indexes are not (b = .16, SE = .04, p = .0001), moral identity was a
available. Further, the standardized coefficients are not negative predictor (b = -.07, SE = .03, p = .034), and
available, so those were calculated manually and are the interaction was significant (b = -.13, SE = .07,
reported in Table 5. In the text we report only the p = .049). A plot of this interaction (see Fig. 1) shows a
unstandardized coefficients, standard errors, and p-values. strongerpositiveassociationbetweenmoraldisengagement
First, we will present results from the four models and aggression at lower levels of moral identity. Lastly,
examining the interaction of moral disengagement with only moral disengagement was a significant predictor of
moralidentity.Forthemodelspredictingcharity(b = .73, rule breaking (b = .11, SE = .04, p = .01).
SE = .13, p = .0001) and civic engagement (b = .47, Next, we will present results from the four models
SE = .12, p = .0001), moral identity was a positive pre- examining the interaction of self-regulation with moral
dictor,whilemoraldisengagementandtheinteractionterm identity. For the model predicting charity, self-regulation
were not significant predictors. For the model predicting (b = .32, SE = .08, p = .0001) and moral identity
123
JYouthAdolescence
3 One 3 One
Standard Standard
2 Deviation 2 Deviation
Above Above
ession 1 MIdeenanti tMyoral gression 1 MIdeenanti tMyoral
ggr 0 Ag 0
Level of A --21 -3 -2 -1 0 1 2 3 OSDtneavenidaatirodn Level of --21 -3 -2 -1 0 1 2 3 OSDtneavenidaatirodn
Below
Below
Mean Moral
Mean Moral -3
-3 Identity Latent Level of Self-Regulation Identity
Latent Level of Moral Disengagement
Fig.2 Plot of the interaction between self-regulation and moral
Fig.1 Plot of the interaction between moral disengagement and identitypredictingaggression
moralidentitypredictingaggression
(b = .46, SE = .13, p = .0001) were positive predictors, 3 One
Standard
but the interaction was not significant. On the other hand, 2 Deviation
g Above
wpohseintivperepdriecdtiicntgorci(vbic=e.n3g4a,gSeEme=nt.,0s6e,lfp-r=egu.0la0t0io1n), wmaosraal Breakin 1 MIdeenanti tMyoral
identity was not significantly related, and the interaction ule 0
was not significant (b = .23, SE = .12, p = .05; of R -3 -2 -1 0 1 2 3
t = 1.961). Lastly, for both aggression and rule breaking, vel -1 One
e Standard
L
moral identity was not a significant predictor, but self- -2 Deviation
Below
regulationwasnegativelypredictive(b = -.18,SE = .03, Mean Moral
-3
p = .001; b = -.11, SE = .02, p = .0001), and the Latent Level of Self-Regulation Identity
interaction term (b = .16, SE = .04, p = .0001; b = .13,
Fig.3 Plot of the interaction between self-regulation and moral
SE = .04, p = .001) was significant. For both antisocial
identitypredictingrulebreaking
outcomes,theplotsoftheinteractionswith(seeFigs. 2,3)
show that self-regulation was a weaker negative predictor
of the antisocial outcome at high levels of moral identity. breaking. The pattern of the interactions was such that
moralidentitydampenedrelationshipsbetweenthesesocial
cognitions (moral disengagement and self-regulation) and
Discussion the antisocial outcomes (aggression and rule breaking).
Intermsofourwayofcapturingmoralidentity,thiswas
Evidence is mounting suggesting that moral identity may the first study to demonstrate interrelations between mul-
be a salient part of moral personality development (Hardy tiple independent indexes of moral identity. In all the
and Carlo 2011). Thus, it is now fairly established that structural equation models, we specified a latent moral
moralidentityispredictiveofbehaviorsinadolescentsand identityvariablewithfivedifferentmoralidentitymeasures
adults.But,westillknowlittleaboutthedynamicsofhow as the observed factor indicators. All of the structural
moral identity is linked to behaviors. It is possible that, in equationmodelsfitthedatawell,thefactorloadingsofthe
addition to motivating moral action, moral identity also moralidentityindexesonthemoralidentitylatentvariable
acts as a moderator of relationships between other social werealllarge,allfivemoralidentityindexeswerestrongly
cognitions and behaviors. Therefore, the primary purpose intercorrelated, and all five moral identity indexes were
of this study was to examine whether and how moral dis- significantly associated with all of the other study vari-
engagement and self-regulation would interact with moral ables. This yields substantial evidence that the five mea-
identity in predicting prosocial (charity and civic engage- sures were capturing a common construct. This approach
ment) and antisocial (aggression and rule breaking) youth may have more fully captured the breadth of the moral
outcomes. Hypotheses regarding the moderating role of identity construct than isolated measures used in prior
moral identity were partially supported. More specifically, studies.Further,themoderatelysizedcorrelationsbetween
therewerethreesignificantinteractions,allintheexpected moral identity and the prosocial and antisocial behaviors
direction. Moral identity moderated the relationships further validate prior work on moral identity, suggesting
between moral disengagement and aggression, self-regu- that it may be an important component and perhaps a
lation and aggression, and self-regulation and rule facilitator of positive youth development and healthy teen
123
JYouthAdolescence
psychosocialfunctioning(forreviews,seeHardyandCarlo and rule breaking). In looking at the results for the inter-
2011; Lapsley 2008). action models, it appears that for predicting prosocial
Beyond these main effects of moral identity on the outcomes, the primary role of moral identity was as a
outcomes, three of the eight interactions tested were sig- predictor (i.e., the main effects for moral identity were
nificant, and all pertained to antisocial outcomes. As significant in three out of the four models predicting pro-
hypothesized, individuals higher on moral identity showed social outcomes), whereas for the antisocial outcomes the
weakerlinksbetweenthesocialcognitivepredictors(moral primaryrolewasasamoderator(i.e.,theinteractioneffects
disengagement and self-regulation) and the antisocial out- were significant in three out of the four models predicting
comes (aggression and rule breaking). This is in line with antisocialoutcomes).Thisalignswithpriorworkonmoral
prior studies that have similarly shown moral disengage- identity,moraldisengagement,andself-regulation.Mostof
ment (Aquino et al. 2007) and self-regulation (Gino et al. the research on outcomes of moral identity has focused on
2011) to interact with moral identity when predicting prosocial outcomes (for review, see Hardy and Carlo
negative outcomes. Thus, if causal relationships are at 2011).Ontheotherhand,mostoftheresearchonoutcomes
work(whichcannotbeestablishedwiththepresentdata),it of moral disengagement (e.g., Gini et al. 2014) and self-
may be that moral identity in some way buffers the nega- regulation(e.g.,Doanetal.2012)hasfocusedonantisocial
tive social effects of moral disengagement and self-regu- outcomes. Thus, moral identity may indeed play multiple
lation failure. Perhaps moral identity has sufficient roles. First, it may provide motivation to engage in pro-
motivational and self-regulatory power to make up for socialbehaviors.Second,itmightalsoserveasamoderator
maladaptive social cognitions. to help minimize the effects of high moral disengagement
Regarding moral disengagement, these interaction and low self-regulation on antisocial behaviors. Future
results echo those found previously by Aquino and col- research is needed to further elucidate these roles.
leagues (Aquino et al. 2007), who made a case for moral Lastly, as a preliminary analysis, we tested the inde-
identity as a buffer against high moral disengagement. pendent bivariate relationships between each of the five
However, these relative roles of moral disengagement and moral identity measures and the other study variables.
moral identity, with moral identity as the moderator, are Relationships between the moral identity indexes and the
countertodiscussionsofmoraldisengagementbyBandura other study variables were typically moderate in strength
etal. (2001).Rather,Bandura,whoproposedthe construct (on average), but varied across the moral identity indexes.
of moral disengagement, suggested that it is a way for Specifically,moralself-relevancewasmoststronglylinked
people who want to see themselves as moral to be able to to the other moral identity indexes, the social cognitions,
engage in immoral acts and still maintain their sense of and the youth outcomes. Further work is needed empiri-
moral identity. Thus, to Bandura, moral identity is at the callycomparingtheutilityofthesedifferentmoralidentity
mercy of moral disengagement, and moral disengagement indexestobetterevaluatewhichmeasuresarepreferablein
is at the mercy of the situation. In contrast, the empirical which situations.
findingsthusfarsuggestthatmoraldisengagementisatthe Despite the important and interesting patterns of find-
mercyofmoralidentity,inthatmoralidentitycandampen ings, the present study had a number of limitations. First,
the effects of moral disengagement. Future research might thedatawerecorrelationalandcross-sectional,limitingour
seek tofurther examinethe potential ofpersonalityfactors ability to ascertain the causal ordering of relationships
to mitigate negative situational pressures. among study variables. Although it was presumed that
For interactions between self-regulation and moral moral disengagement, self-regulation and moral identity
identity predicting the antisocial outcomes, the results are precedethefourbehavioroutcomes,thereverseordermay
in line with those of Gino and colleagues (Gino et al. also hold; those involved in civic engagement, charitable
2011), who argued that moral identity may buffer against giving, aggression and rule breaking may develop certain
depleted self-regulation. However, these relative roles of patterns of moral disengagements, self-regulation, and
self-regulation and moral identity, with moral identity as moral identity over time. Future research should employ
the moderator, seem counter to Rest’s (1983) four com- experimentalandlongitudinaldesigntobetterenableusto
ponentsmodelofmorality.Restpitchedself-regulationasa infer causal links across study variables. Second, some
capacity that would enable individuals to follow-through measures were self-report, which can lead to problems of
with moral motivations and intentions (i.e., self-regulation shared method variance and social desirability bias. How-
isamoderator).Incontrast,thepresentdatapositionmoral ever, Clarke, Lewinsohn, Hops and Seeley (1992) report
identity as the moderator of links between self-regulation thatpeopletendtobefairlyaccurateinreportingtheirown
and outcomes, at least for antisocial outcomes. behaviors and internal states. Further, the predictors and
It was interesting that the three significant interactions moderator were self-reported but the behaviors were par-
were for predicting the antisocial behaviors (aggression ent-reported, limiting the problems of social desirability
123