Going to College and Unpacking Hazing: A Functional Approach toDecrypting Initiation Practices Among Undergraduates

Caroline F. Keating, Jason Pomerantz, Stacy D. Pommer, Samantha J. H. Ritt,Lauren M. Miller, and Julie McCormick

Colgate University

Initiation practices likely support group functioning by promoting group-relevant skillsand attitudes, reinforcing status hierarchies, and stimulating cognitive, behavioral, andaffective forms of social dependency. In field tests of these propositions, 269 under-graduates from same-gender organizations rated their initiation experiences. As pre-dicted, athletes reported relatively more physical challenge and pain, whereas membersof Greek-letter organizations reported more social deviance and embarrassment. Hier-archy was positively associated with initiations featuring social deviance but unexpect-edly negatively related to physically and psychologically harsh initiations. Harshtreatment and fun independently predicted group identity. Laboratory experiments onmale (n � 74) and female (n � 37) undergraduates found that discomforting inductionsincreased social dependence on group opinion and, for women, increased additionalforms of dependence (proximity to induction agents and negative mood when leftalone). The results across studies suggested that hazing’s task masters are 3: schoolingskills and attitudes, conveying hierarchy, and promoting social dependency.

Keywords: initiation, hazing, conformity, social identity, attachment to groups

If those above love ritual, then the common people willbe easy to manage.

—Confucius, 551–479 BCE (translated by E. Sling-erland)

The chain of motivational, cognitive, and so-cial processes that bind individuals to collec-tions of others is forged early in life. Infants andyoung children inherently form affectionalbonds and seek proximity with those who be-come familiar to them (Ainsworth, 1972;Bowlby, 1969/1982; Cassidy, 1999). The needto belong burgeons, fueling a desire for inter-personal attachments that older children, ado-lescents, and adults may satisfy through groupmembership (Baumeister & Leary, 1995;Smith, Coats, & Murphy, 2001). A cognitiveproclivity to distinguish in-groups from out-

groups, and to self-categorize, transforms indi-vidual identity into in-group social identity (Taj-fel & Turner, 1979; Turner, 1985). Striving tobelong to a particular group, especially duringritualized initiations, may result in the justifica-tion of that effort, thereby inoculating individ-uals against any dissonant cognitions they mayharbor concerning the consequences of groupmembership (Aronson & Mills, 1959; Brehm,1960; Cialdini, 2001). These linkages of moti-vational, cognitive, and social elements prepareindividuals to seek group affiliations and adoptsocial identities throughout their lifetime(Baumeister & Leary, 1995).

Affiliation and identification with a groupand its leader cultivate feelings of security, es-pecially under conditions of uncertainty orthreat (Baron, 2000; Baumeister & Leary, 1995;Festinger, 1954; Gump & Kulik, 1997; Hogg,2001; Rosenblatt, 1964; Schachter, 1959;Smith, Murphy, & Coats, 1999; Tajfel &Turner, 1979). For this reason, group leaderswho are perceived as powerful and influential inmanaging threat are likely to enjoy member-ships with relatively strong collective identities(Baron, 2000; Hogg, 2001; Rosenblatt, 1964).Theoretically, it matters little where the threat

Caroline F. Keating, Jason Pomerantz, Stacy D. Pommer,Samantha J. H. Ritt, Lauren M. Miller, and Julie McCor-mick, Department of Psychology, Colgate University.

Lauren M. Miller is now at the Department of Psychol-ogy, Virginia Commonwealth University.

Correspondence concerning this article should be ad-dressed to Caroline F. Keating, Department of Psychology,Colgate University, Hamilton, NY 13346. E-mail:ckeating@mail.colgate.edu

Group Dynamics: Theory, Research, and Practice Copyright 2005 by the Educational Publishing Foundation2005, Vol. 9, No. 2, 104–126 1089-2699/05/$12.00 DOI: 10.1037/1089-2699.9.2.104

104

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

originates. In humans and other primates, feel-ings of threat may direct affiliative behavior andaffectional bonds toward those suffering thesame fate or even toward the agents of the threat(Baron, 2000; Baumeister & Leary, 1995;Bowlby, 1969/1982; Rajecki, Lamb, & Obmas-cher, 1978). Therefore, events simulating threatmay be used to trigger the psychological pro-cesses that connect individuals to groups and totheir leaders.

At times when threat is low, groups and theirleaders may instill threat in potential membersby invoking perceived enemies (Hogg, 2001;Landau et al., 2004; Rosenblatt, 1964) or per-haps by staging initiation rituals that activatefeelings of threat (Baron, 2000; Galanter, 1999).Contrived threats are likely to empower leadersand groups by stimulating the kinds of attitudes,social perceptions, and affiliative bonds thatbreed group identity and inspire obedience anddevotion among new members of face-to-facegroups. Processes such as these may be un-leashed among constituencies as large as na-tions (Landau et al., 2004; Rosenblatt, 1964) oras small as universities and social clubs(Baumeister & Leary, 1995; Hermanowicz &Morgan, 1999; Moreland & Levine, 1989).

The premise of our research was that threat-ening initiation practices such as hazing ritualsfunction to support and maintain groups in atleast three ways: by promoting group-relevantskills and attitudes, by reinforcing the group’sstatus hierarchy, and by stimulating cognitive,behavioral, and affective forms of social depen-dency in group members. Field and laboratoryresearch probed these propositions in samplesof college students. The rationale for these stud-ies is articulated next, and then the studies aredescribed.

Initiations as Contrived Threat

Preliminary welcoming receptions designedto attract new group members typically entailmildly enjoyable social experiences such as in-formal, conversational gatherings surroundingparties or dinners. However, the character ofevents often changes after these early meetings,consistent with the “foot in the door” technique.That is, initial requirements to attend groupevents may consist of simple efforts that areonly mildly arousing, such as turning out inparticular attire for an occasion, spending time

engaged in prescribed, social exchanges withgroup members, or waiting for extended periodsof time before being interviewed by representa-tives of the group (Pomerantz, 1995). Initialcompliance with low-cost efforts like thesemakes subsequent, costlier forms of compliancemore likely (Cialdini, 2001). Subsequent initia-tion events invite increasingly more effort andmay ultimately require newcomers to endurevarious types of discomfort and threat. For ex-ample, cleaning up after parties is a commonassignment given to newcomers, and it can be-come ritualized into particularly degradingforms: In some cases, pledges clean up garbage,vomit, and broken glass in the aftermath ofnights of binge drinking using bare hands andwearing “uniforms” that they are not allowed towash from party to party (Pomerantz, 1995).Yet the heightened negativity of the experienceis apparently transformed as devotion to thecollective.

Case studies reveal that a variety of initiationpractices characterize groups with highly cohe-sive and intensely loyal memberships (e.g., Ga-lanter, 1999; West, 1993). From what is knownabout real-world groups, initiation and indoctri-nation protocols are orchestrated so as to stim-ulate sequences of emotional, perceptual, andcognitive experiences over time (Baron, 2000;Galanter, 1999; Moreland & Levine, 1989;Singer & Ofshe, 1990; West, 1993). Initiationsmay include activities perceived to be fun andrewarding, physically and emotionally demand-ing, embarrassing, socially deviant, degrading,painful, and sometimes dangerous or brutal.The activities designed for newcomers typicallyadhere to highly prescribed protocols defined bygroup traditions (Robbins, 2004). Establishedgroup members may be given specialized roles,such as “pledge master,” and charged withmaintaining the integrity of initiation protocols.Leaders and group members commonly sub-scribe to the belief that newcomers should un-dergo induction procedures resembling thosethey experienced. Initiations are thought to pro-duce “better” group members. It is not unusualfor initiates to perceive induction activities asfun and rewarding, and initiation practices may,in fact, be fun. However, there is also evidencethat participants cognitively skew discomfort-ing initiation experiences in ways that makethem seem more enjoyable than they actuallywere (Cialdini, 2001).

105UNPACKING INITIATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

In some cases, the initiation activities newmembers endure clearly induce threat by incor-porating physical challenge or pain. For exam-ple, athletic teams and gang members have beenknown to ritualistically paddle incoming mem-bers; these episodes at times degenerate intobeatings or sexual assaults (Finkel, 2002).Pledges to Greek-letter organizations may sub-mit to ingesting concoctions meant to inducenausea; in some instances, pledges have chokedto death (Finkel, 2002). Pledges may be directedto wait for hours in dark, cold, uncomfortableplaces or to maintain painful physical positionsfor extended periods of time. They may becharged with the exhausting task of paintingceilings in the fraternity house all night long.Branding with lit cigarettes has been practicedas an initiation ritual in some college fraternitiesand sororities (Finkel, 2002).

College campuses are but one setting inwhich newcomers to groups may face physicalextremes. Canadian Air Force initiates performdemeaning tasks, binge drink, and submit tointense, painful, physical activity (Winslow,1999). During “blood pinning” initiations, U.S.Marine paratrooper trainees earning goldenwing pins had them slammed and punched intotheir bodies by fellow marines (Gleick, 1997).New cult members have been subjected to in-adequate diets while performing repetitive tasksrequiring physical endurance (Galanter, 1999;West, 1993). On campus or off, the threat pro-voked by these types of experiences combinesthe physical and psychological consequences ofsubmitting to effortful, painful, or self-injuriousbehavior. Having endured it, individuals mayfind the consequence of such physically harshtreatment to entail feelings of mastery over thechallenge.

Other types of threatening inductions featuresocial deviance, embarrassment, or humiliation(Cialdini, 2001). For instance, initiation ritualsin some fraternity houses include “lard sliding,”where pledges strip, slather their bodies withlard, and slide along the floor (Pomerantz,1995). Initiates may be assigned to while awayhours concocting original nicknames for fellowrecruits that have secret, shared meanings. Newmembers may be asked to reveal personallyembarrassing information and to memorize andkeep secret special songs and histories meant todistinguish the group from the larger society(Galanter, 1999; Pomerantz, 1995; Robbins,

2004). Telling and keeping secrets is an effec-tive way to establish social bonds between un-familiar individuals (Wegner, Lane, & Dimitri,1994). Initiates to all-female organizations maybe assigned to sing bawdy songs in front ofall-male groups. In some sororities, pledges dis-robe and have areas of body fat identified andcircled with markers while being observed byfellow recruits and group leaders and members(Broadbent, 2003). Binge drinking and cigarsmoking, sexual conquest assignments, per-forming tedious tasks, and running fools’ er-rands are socially deviant, silly, embarrassing,or demeaning activities characterizing the initi-ations of many groups. Though not alwaysphysically demanding, initiation activities suchas these harness the power of effortful engage-ment in norm violation to create arousal anddistress, thereby inducing threat of a social na-ture. At the same time, these practices seemgeared toward creating and elaborating distinc-tions between the in-group and normativegroups (Robbins, 2004).

To summarize, initiations comprising physi-cal and psychological hardships may be used toinduce arousal and threat in initiates. Theseconditions leave individuals more susceptible togroup influence and indoctrination (Baron,2000). But beyond that, certain types of initia-tion activities seem orchestrated to achieve par-ticular effects. Experiencing physical extremesmay train initiates to withstand physical duress.Engaging in social deviance may primarilyfunction to etch distinction between the in-group and normative groups in the minds andemotions of initiates. Maltreatment may elicitcognitive, behavioral, and emotional symptomsof social dependency. From a functional per-spective, different types of initiation experi-ences seem designed to preserve group featuresand cultivate group allegiance in particularways.

Initiations Cultivate Group-Relevant Skillsand Attitudes

We tested the proposition that initiation prac-tices preserve critical features of group life by“unpacking” the initiation practices of collegeathletic teams, sororities, and fraternities inhopes of identifying basic elements. We rea-soned that physical challenges would predomi-nate the induction practices of athletic groups

106 KEATING ET AL.

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

whereas activities highlighting social deviance(and therefore social distinctiveness) wouldcharacterize the initiations of groups dedicatedto creating exclusive social networks, such asGreek-letter organizations. On the basis of afunctional perspective, athletes were predictedto report relatively greater degrees of physicalduress in their initiations than were members ofGreek-letter organizations. Members of Greek-letter organizations were expected to report ini-tiation activities entailing more social deviancethan were members of athletic groups.

Initiations Reinforce Group Hierarchy

A second function of initiations may be toattune new members to the group’s hierarchicalstructure (Moreland & Levine, 1989). Initia-tions provide early opportunities for group lead-ers to establish power over newcomers to theorganization. Getting recruits to follow induc-tion protocols essentially trains them to complywith the group’s authority structure. Groupscharacterized by relatively complex, vertical,formalized hierarchies maintain greater socialdistances between leaders and new membersthan do groups with flatter, informal hierar-chies. Therefore, hierarchical groups were ex-pected to rely on the long, psychological reachof stressful initiations as a means of establishingsocial control more than were groups with lesshierarchy. We predicted that members of groupswith more structured hierarchies, operationallydefined by greater role diversity and greaterpower differences between leaders and newmembers, would report more severe initiationpractices and more frequent engagement in ini-tiation activities than groups with lesshierarchy.

Initiations Stimulate Social Dependency

Initiations also function to promote cognitive,behavioral, and affective forms of social depen-dency among group members (Moreland & Le-vine, 1989). But what psychological mecha-nisms explain why initiation experiences thatinduce threat, duress, or discomfort rally ratherthan discourage the loyalties of those who en-dure them? One possibility stems from attach-ment theory (Ainsworth, 1972; Bowlby, 1969/1982). For a decade or so, researchers havenoted parallels between the attachment pro-

cesses characteristic of parent–infant bonds andthose underlying adult bonds (e.g., Hazan &Zeifman, 1999; Shaver, Hazan, & Bradshaw,1988). More recently, attachment theory hasbeen applied to individual attachments togroups (Smith et al., 2001). Smith and his col-leagues noted, for example, that the self-regu-latory processes that tie individuals to one an-other resemble those that bind individuals togroups (Smith et al., 1999, 2001). Smith et al.(2001) argued that because the biosocial–emo-tional system for human attachment is evolu-tionarily old and ingrained in human nature, itcould have been coadapted to configure individ-ual relationships with collectives as well as withother individuals. Using similar thinking, wepropose that a unique aspect of the attachmentsystem, maltreatment effects, applies to humanconnections with groups and may cast light onhow group initiations function to promote be-havioral, cognitive, and affective forms of so-cial dependency.

Maltreatment effects describe the phenome-non whereby harsh conditions trigger goal-di-rected responses in organisms seeking refugefrom the duress (Bowlby, 1969/1982). Hencethe infant seeks out a caregiver when hungry,fearful, tired, or in pain. Under mundane cir-cumstances, manifesting such social depen-dency would be adaptive, contributing fitness toboth kin caregivers and offspring. However, theresponse can be elicited even when the care-giver is the source of the discomfort; that is,social dependency may be directed toward thevery agent of the mistreatment (Bowlby, 1969/1982; Rajecki et al., 1978).

Maltreatment effects have been investigatedin a variety of nonhuman species, includingmonkeys, dogs, sheep, and chickens (Cairns,1979; Rajecki et al., 1978). In one study, pup-pies were either indulged or alternately in-dulged and punished by their human handler(Fisher, 1955, as cited by Rajecki et al., 1978).When tested for attachment to him, puppies thatwere both indulged and punished showed morethan twice the desire to maintain close proxim-ity to their handler than did the indulged groupof puppies. The stimulation of abuse can bepotent: Monkeys often form stronger bondswith periodically abusive agents than with non-abusive agents (Rajecki et al., 1978).

The implications of maltreatment effects forhuman attachments to groups are important. In-

107UNPACKING INITIATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

dividuals who undergo mental and physical du-ress can become dependent on and even at-tached to those inflicting the treatment, espe-cially when a clear power differential is presentand alternative social options are unavailable(Auerbach, Kiesler, Strentz, Schmidt, & Serio,1994; Moreland & Levine, 1989; Rajecki et al.,1978). From a functional perspective, treatmentthat includes punishment as well as reward maybe most effective in creating social dependencyon its agents (Dolinski, Ciszek, Godlewski, &Zawadzki, 2002; Galanter, 1999).

Anecdotal evidence seems to support the ideathat severe treatment sometimes stimulates so-cial bonds in humans. The Stockholm syndromedescribes the ideological support and affectionthat hostages have been known to develop fortheir captors (West, 1993). It was named for thereactions of employees taken hostage by twoex-convicts during an attempted bank robberyin Sweden in 1973. During captivity, the hos-tages came to identify with their captors’ causeand resisted efforts to be rescued. Affectionalbonds also developed; after being freed by po-lice, two hostages eventually became engagedto their captors. Patty Hearst experienced a sim-ilar transformation of ideology, affection, andidentity after being kidnapped, terrorized, andheld hostage by the Symbionese LiberationArmy in the 1970s (Hearst, 1988). More re-cently, an American kidnapping victim takenhostage by Muslim rebels in the Philippinesseemed to suffer a similar, psychological fate;he was described as having “fallen for his cap-tors” (Macfie, 2000, p. 1). And when Auerbachet al. (1994) simulated a highly stressful captiv-ity situation in which airline employees were“taken hostage” by “terrorists,” liking for the“terrorists” increased over time. Related psy-chological processes presumably explain whyabused spouses and children often prefer to staywith rather than leave their abusers (Boulette &Andersen, 1986; Dutton & Painter, 1993; Ra-jecki et al., 1978).

When maltreatment is connected to involve-ment with a defined group, the social depen-dency it fuels may be manifested cognitively,emotionally, and behaviorally. Cognitively andemotionally, the need to maintain and defendthe sense of self in the face of the threat anduncertainty can be remedied by transforming anindividuated identity into a group identity(Hogg, 2001). Behaviorally, the dependency

fostered by maltreatment is likely to be ex-pressed as compliance with group norms andattraction to group members (Baron, 2000;Moreland & Levine, 1989). We tested whethersocial dependency in the wake of relativelysevere initiations was manifested in these waysby assessing group and individuated identityand by tracking conformity and attraction togroups. Identity was assessed among the mem-bers of campus groups with varying initiationpractices. Conformity and attraction to thegroups was measured in laboratory experimentsthat simulated mild or severe inductions to con-trived groups.

Identity

Two ratings of membership importance wereconstrued as measures of identity. FollowingEthier and Deaux (1994), our first measuretapped group identity by asking participants torate the importance of the group to them. Oursecond, relatively novel measure of individu-ated identity invited participants to rate theirown importance to the group. Identity, there-fore, had dual, self-related aspects: (a) howmuch the individual was psychologically in-vested in belonging to the group and (b) howmuch the individual perceived the group to in-vest in his or her unique, individuatedbelonging.

The distress-provoking aspects of initiationswere expected to augment ratings of the impor-tance of group membership by the individual asindividuals sought refuge from threatening cir-cumstances in the psychological safe house ofgroup belonging (Baron, 2000; Baumeister &Leary, 1995; Hogg, 2001; Rosenblatt, 1964).Threatening conditions were not, however, ex-pected to make individuals feel more importantto the group; if anything, there was reason tobelieve that being subjected to severe initiationprocedures was likely to diminish rather thanincrease an initiate’s sense of individuatedpower and importance relative to the group(Baron, 2000; Moreland & Levine, 1989). Forthe members of campus groups we surveyed,our specific prediction was that those who ratedgroup initiations as severe would also rate thegroup’s importance to them as greater thanwould individuals belonging to groups with lesssevere initiation practices. In contrast, mem-bers’ perceptions of their individuated impor-

108 KEATING ET AL.

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

tance to the group were not expected to relate tosevere treatment.

Conformity

Because the behavioral expression of socialdependency—conformity—was difficult tomeasure in the field, we resorted to experimen-tal simulations of initiation protocols to testwhether harsh treatment evoked compliance.There is some behavioral evidence from labo-ratory studies that has linked social dependencyin the form of compliance or conformity toexperimentally induced discomfort. For exam-ple, harsh treatment in the form of public failurewas found to increase social compliance (VanDuuren & Di Giacomo, 1996, 1997). Studies ofinitiation severity per se rarely test for socialdependency, although when Schopler and Bate-son (1962) did, they found that severe induc-tions increased initiates’ dependency on theopinion of the experimenter. These researchersargued that the severe initiation generated de-pendency by establishing, in the perceptions ofinitiates, the experimenter’s power over them.Consistent with Schopler and Bateson (1962),we predicted that the social dependency engen-dered by relatively severe treatment would bemanifested in high rates of conformity to groupopinion.

Affective Social Bonds

According to attachment theory, discomfortreadies the emotional system to seek protectionby establishing affective bonds with potentialcaregivers (Bowlby, 1969/1982). When thesources of abuse are the only available outlet,affiliative responses may be directed towardthem (Rajecki et al., 1978). Thus, another pos-sible outcome of severe initiations is the devel-opment of interpersonal attraction toward itsagents. This hypothesis was tested in two labo-ratory experiments in which confederatesstaged either discomforting or innocuous initi-ations. Discomforting initiations were predictedto enhance the attractiveness of confederates inthe eyes of participant “initiates” (Study 2). InStudy 3, we tracked two additional measures ofattachment, proximity maintenance and separa-tion anxiety (Bowlby, 1969/1982), and pre-dicted that inductees who experienced a rela-tively severe induction would maintain close

physical distances to confederates and experi-ence heightened anxiety when confederates leftthem alone.

Cognitive Dissonance: An AlternativeMechanism

The standard, social psychological explana-tion for why we come to value the group wesuffer for—and the greater the suffering, thegreater the value—is dissonance theory (Aron-son, 1995; Brehm, 1960; Cialdini, 2001; Fest-inger, 1954). Theoretically, harsh initiationsarouse dissonance in individuals who engage inactivities that violate conceptions of the self(Aronson & Mills, 1959). The more voluntarythe engagement, the greater the dissonancearoused and the more motivation there is toresolve it (Elliot & Devine, 1994). Dissonanceis mitigated by either cognitively diminishingthe negative aspects of initiation activities (e.g.,“It wasn’t degrading; it was fun!”), overvaluingthe group (e.g., “Belonging to this group is soworthwhile that it is worth suffering in order tobecome part of it”), or both (Aronson, 1995;Brehm, 1960; Cialdini, 2001). Thus, the mech-anism underlying the connection between se-vere initiations and enhanced evaluation of thegroup is ostensibly dissonance reduction.

Support for the idea that relatively threaten-ing or severe initiations make groups more valu-able to newcomers comes from a few classic,laboratory experiments conducted in the 1950sand 1960s (i.e., Aronson & Mills, 1959; Gerard& Mathewson, 1966). These early studies es-sentially found that the greater the mental orphysical effort required for group membershipwas, the more inflated ratings of group attrac-tiveness became. For instance, participants inAronson and Mills’s (1959) study who under-went an embarrassing induction found fellowgroup members to be more attractive than didparticipants who experienced a mild induction.Similarly, participants given relatively intenseelectrical shock during induction exhibitedgreater liking for their group than did those whoreceived mild shock (Gerard & Mathewson,1966). These early studies were consistent withthe notion that newcomers reduce dissonanceby overvaluing the group for which theysuffered.

109UNPACKING INITIATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

Discrepancies Among Previous Studies ofInitiations

Both attachment theory and dissonance the-ory offer psychological mechanisms that pur-port to explain the connection between severeinitiation practices and group attraction. Yetlater studies have failed to support the basic ideathat severe initiations produce greater liking forthe group. In the laboratory, Hautaluoma, Enge,Mitchell, and Rittwager (1991) manipulated theseverity of induction and reported that mildrather than severe procedures heightened in-ductees’ attraction to the group. Field studieshave also cast doubt on the link between suf-fering and liking. Lodewijkx and Syroit (1997)surveyed newcomers to Dutch sororities andfound no evidence that (reportedly) severe in-ductions provoked either strong group affilia-tions or heightened group attraction. Specifi-cally, the Dutch sorority pledges who assessedinitiation experiences as “unpleasant” reportedfeelings of frustration, loneliness, and depres-sion, as well as diminished affiliation with thegroup (Lodewijkx & Syroit, 1997). In a recentrefinement of these analyses, Lodewijkx andSyroit (2001) reached a similar conclusion byfinding no evidence of dissonance effects. Thusformal evidence of hazing’s effects on social–emotional bonds is quite mixed. It appears thatharsh treatment does not reliably produce thekind of attraction to groups that some research-ers seek to construe as evidence of dissonancereduction or that we suggest may be a manifes-tation of maltreatment effects.

Limitations of Previous Research

Irrespective of theoretical orientation, mea-surement and design problems may account fordiscrepancies among the findings of laboratoryexperiments, field studies, anecdotal reports,and case studies of the effects of initiations.Laboratory experiments typically impose short-term “inductions” designed to invoke specific,psychological threats, such as social embarrass-ment (i.e., Aronson & Mills, 1959; Schopler &Bateson, 1962), fear of shock (i.e., Gerard &Mathewson, 1966), or anxiety produced by per-forming a repetitive task (Hautaluoma et al.,1991). What laboratory participants placed in“minimal” groups experience is short lived andvariously defined (e.g., embarrassment, fear of

pain, boredom). Meanwhile, initiates of real-world groups experience something verydifferent.

Anecdotal reports and case studies suggestthat real-world initiation activities arouse a con-fluence of feelings. Activities may compriseaspects that are fun, embarrassing, disgusting,boring, painful, brutal, and challenging. Unfor-tunately, researchers in the field typically mea-sure real-world hazing by packaging these ex-periences together and measuring them as asingle construct. For example, Lodewijkx andSyroit (1997) measured “severity” of initiationexperiences on scales for “unpleasantness.”Studies indexing initiations with a single scale(i.e., “severity”) constrain the complexity ofinitiations and initiate reactions into a unitaryresponse. We gathered multiple assessments ofinitiation experiences and submitted them to amultiple regression analysis in order to distin-guish among them.

The Present Research

We “unpacked” initiations and their conse-quences for individuals and groups on collegecampuses from a functional perspective. In thefield, we assessed whether initiation practicesrelated to the development of group-relevantattitudes and skills, to the reinforcement ofgroup hierarchy, and to group identity. In thelaboratory, we tested whether relatively severeinductions spawned conformity and attractionto group members as a manifestation of thesocial dependency generated by maltreatment.

In Study 1, we asked students belonging tocampus organizations (typically, sororities, fra-ternities, athletic teams and clubs, and a cap-pella groups) to retrospectively assess their im-pressions of “activities designed to initiate ortest new members.” Members rated the degreeto which these activities were fun, embarrass-ing, physically and psychologically challeng-ing, socially deviant or abnormal, and painful.Members also responded to rating scales assess-ing the relative power of group leaders over newmembers. Measures of group and individuatedidentity were taken; participants rated how im-portant the group was to them and how impor-tant they were to the group, respectively.

In Studies 2 and 3, experimental confederatesserved as group members who “inducted” naiveparticipants into their group using either dis-

110 KEATING ET AL.

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

comforting or innocuous procedures. Social de-pendence was measured by assessing partici-pants’ conformity to confederates’ opinions andby having them rate the attractiveness of con-federates and the enjoyment of the experience.In Study 3, affiliative bonds were indexed bythe proximity participants maintained with con-federates and by the anxiety they reported whenconfederates left them alone, consistent with thethinking of Bowlby (1969/1982).

Gender was likely to add complexity to rela-tionships involving power. Male and femaleattitudes toward social dominance typically dif-fer (Pratto, Sidanius, Stallworth, & Malle,1994), and the nature of power structures inmale and female groups reflects these differ-ences. Power relationships among males aremaintained and supported by overt aggressionand are characterized by well-defined hierar-chies (Bjorkqvist, 1994; Buss, 2004; Crick &Grotpeter, 1995). Power bonds among femalesemerge from subtle, indirect, relational forms ofaggression and coincide with less obvious hier-archical dominance structures (Bjorkqvist,1994; Buss, 2004; Crick & Grotpeter, 1995).From early to late in life, male relationshipstypically promote status distinction whereas fe-male relationships generally project status dif-fusion (Ostrov & Keating, 2004; Pratto et al.,1994). Thus, our samples included only individ-uals initiated into same-gender organizations,and gender of participant was incorporated as avariable.

Limitations

Our field study (Study 1) was meant to launcha new look at the fundamentals of initiations.We investigated how initiations, as contrivedthreats, altered perceptions of self in relation-ship to groups and their mission.

There were several shortcomings to thisrough start. First, we relied on participant verbalreports of initiation experiences rather than ac-tual observation. Second, we studied only un-dergraduate volunteers who identified with asame-gender group, thereby restricting our abil-ity to generalize results beyond the sample wegathered. Third, our natural groups confoundedorganization with hierarchy: We relied predom-inantly on Greek-letter organizations for elabo-rated hierarchies and on athletic teams and clubsfor less hierarchical structures. Moreover, the

individuals attracted to different organizationsmay have had inherent proclivities toward oragainst particular forms of initiation activities.Though our experimental studies (Studies 2 and3) offered partial remedies to some of thesemethodological problems, they suffered fromthe kinds of shortcomings inherent in contrivedsettings.

Study 1

Method

Participants. An initial pool of 156 maleand 166 female undergraduate students partici-pated by volunteering to complete a short ques-tionnaire. All were enrolled in a small, highlyselective liberal arts college in the Northeast.They ranged in age from 18 to 22. Each re-ceived a coupon for a free slice of pizza for theirparticipation except for 39 participants whoearned introductory psychology laboratorycredit in lieu of the coupon.

Ninety-two percent of men and 87% ofwomen who participated reported membershipin a single-gender organization as their primaryaffiliation. Data from these participants formedthe basis of all analyses. Therefore, our finalsample included 138 men and 131 women whoindicated a same-gender group as their mostimportant campus membership.

Materials. Testing materials included asmall (6.5 � 3.5 in.), innocuous-looking book-let containing 21 pages. On the first page, par-ticipants reported their academic class year andgender. On the second page, they checked off anorganizational affiliation (if they had one).Printed instructions asked each participant tochose the one group he or she identified withmost, with the following options listed: a clubsport team, student newspaper, an a cappellagroup, a sorority, student government, pepband, a varsity sports team, a religious organi-zation, a fraternity, or “other.” Each subsequentpage presented a single question so that previ-ous responses were less salient as each newquestion was considered and subsequent re-sponses were made.

The questions appeared in a standard order.Question 1 asked, “Did you participate in initi-ation activities?”; response options were never,rarely, occasionally, frequently, and often. Thenext six questions assessed how psychologi-

111UNPACKING INITIATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

cally challenging, how embarrassing, howphysically demanding, how much fun, howpainful, and how deviant (defined as “not nor-mal, strange”) the activities were and presented7-point response scales labeled not at all (1),somewhat (4), and extremely (7).

Similar 7-point scales were used for the nexttwo identity-related questions. Similar to Ethierand Deaux (1994), we assessed the degree towhich individuals identified with groups by ask-ing, “How important is the group to you?” Thesecond identity-related question posed was“How important are you to the group?”

The next subsequent set of questions askedabout the existence of officers and their powerover members. The first query in this clusterwas “Is there a president, captain, or groupleader? (circle No or Yes). How much powerdoes the president, captain, or leader have overother members of the group?” Perceptions wererecorded on a 7-point scale labeled no power(1), some power (4), and lots of power (7).Subsequent pages presented similar questionsfor copresidents/cocaptains/coleaders, vicepresidents, secretaries, treasurers, and newmember activities chairpersons. In the end,questions about the power of these secondaryoffices were not analyzed because athletic teamsand clubs had no such offices. The final ques-tion presented the same 7-point scales andasked, “How much power do new membershave over other members of the group?”

Procedure. Contact was made by two fe-male experimenters stationed at various loca-tions on and around campus. These locationsincluded dining halls, dormitory lounges, in-door and outdoor student gathering places, li-braries, and lobby areas in academic buildingsand near athletic facilities. Students were in-vited to volunteer for a study of student groupactivities and group identity. They read a con-sent form stating that we were studying studentperceptions of the campus organizations inwhich they were involved. Students were in-formed that participation was voluntary andanonymous and that neither their personal iden-tities nor their specific group affiliations wouldbe revealed. After reading the short descriptionof the study, they signed a consent form thatwas detached from the survey and immediatelyplaced in a separate backpack. Experimentersmaintained an appropriate distance from partic-ipants as they independently completed the sur-

veys. These procedures were designed to en-courage accurate, honest responses.

As part of a pilot study, participants weregiven the option of completing the 10-itemNeed to Belong Scale (Baumeister & Leary,1995) after completing our survey. Approxi-mately 74% of participants completed the scale(these data are not reported here). When fin-ished with their chosen tasks, participants werethanked, debriefed, and given a coupon redeem-able for a free slice of pizza.

Results and Discussion

Description of the sample. Table 1 containsdescriptive statistics for male and female under-graduate participants who reported membershipin same-gender organizations. Men and womenwere generally similarly distributed in terms ofclass year and organizational affiliations. Mostidentified either athletic clubs or teams orGreek-letter organizations as the primary groupto which they belonged. The sizes of the groupsmen and women belonged to averaged be-tween 26 and 50 members. In these ways, maleand female participants were comparable. How-ever, male juniors were relatively underrepre-sented. More men than women were associatedwith a cappella groups. More women than menidentified with athletic groups (see Table 1).

Table 1Descriptive Statistics (Percentages) forUndergraduate Participants in Study 1

VariableMen

(n � 138)Women

(n � 131)

Class yearFirst 29 23Second 27 24Third 16 23Fourth 28 30

OrganizationAthletic teams 50 57Greek affiliations 38 38A cappella groups 12 05

Participation in initiationactivities

Never 0 0Rarely 17 16Occasionally 28 26Frequently 26 32Often 28 25

112 KEATING ET AL.

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

Initiation tactics of athletic and Greek-lettergroups. To test the hypothesis that differentgroups rely on different initiation tactics in or-der to reinforce group-relevant attitudes andskills, we sorted participants into two distinctmemberships: varsity and club athletics (n �146) and Greek letter (n � 102). Members of acappella groups (n � 21) were omitted fromthese analyses. Predictions were that membersof athletic groups would report relativelygreater degrees of physical challenge and painas part of initiations and that members of Greek-letter organizations would report relativelygreater degrees of embarrassment and devianceduring initiations. Group membership and gen-der were the independent variables in a 2 � 2multivariate analysis of covariance, with per-ceptions of the degree of physical challenge,pain, psychological challenge, embarrassment,and deviance entailed in initiations as depen-dent variables. Ratings of how much fun par-ticipants attached to initiations were covariedin order to control for possible dissonance ef-fects in which participants cognitively morphnegative experiences into positive ones. The

adjusted means for this analysis appear in Fig-ure 1.

The predicted multivariate main effect forgroup membership was significant, F(5,236) � 9.41, p � .01, and the pattern of resultsdepicted in Figure 1 fit predictions. Members ofathletic groups reported significantly morephysical challenge, F(1, 240) � 7.61, p � .01,and more pain, F(1, 240) � 6.13, p � .02, thandid members of sororities and fraternities.Meanwhile, members of Greek-letter organiza-tions reported more embarrassment, F(1,240) � 25.20, p � .01, and more social devi-ance, F(1, 240) � 15.16, p � .01, than didathletic group members. Differences for psy-chological challenge were not significant, F(1,240) � 1.96, p � .16. A similar pattern ofresults was obtained when the covariate, per-ceived fun, was omitted from the multivariateanalysis.

The omnibus analysis also yielded a maineffect for gender, F(5, 236) � 5.23, p � .01.The multivariate effect was supported by resultsfor physical challenge (Ms � 4.10 and 3.12),F(1, 240) � 15.64, p � .01; pain (Ms � 2.92

Figure 1. Mean ratings of initiation activities reported by undergraduates belonging toathletic groups (open bars) and Greek-letter organizations (dark bars). Means are adjusted forratings of fun.

113UNPACKING INITIATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

and 2.14), F(1, 240) � 14.08, p � .01; psycho-logical challenge (Ms � 3.69 and 3.23), F(1,240) � 4.10, p � .05; and social deviance(Ms � 3.54 and 2.96), F(1, 240) � 6.95, p �.01. In each instance, mean ratings were higherfor men than for women. Mean differences forembarrassment did not differ for men andwomen (Ms � 3.18 and 3.28), F(1, 240) � 1.0.The overall main effects for both gender andgroup membership were robust in that there wasno multivariate interaction, F(5, 236) � 1.0.

Initiation tactics and perceptions of hierar-chy. To test the hypothesis that initiation tac-tics support perceptions of group hierarchy, weused initiation practices as predictor variables inregression analyses, with power perceptions asthe dependent or predicted variables. Includedin these analyses were participants with groupmembership in any of the three single-gendergroups we originally sampled (athletic, Greekletter, and a cappella). The variables availablefor the predictor set included ratings of thephysical and psychological challenge, pain, em-barrassment, social deviance, and fun involvedin initiations. However, these variables intercor-related. To avoid problems with multicollinear-ity, we generated composite predictor variablesfrom the original measurements. This procedurehad the additional advantage of reducing thesizes of subsequent predictor sets, which in-cluded initiation experiences, gender, and alltwo- and three-way interactions, to a more man-ageable number.

Composite predictor variables were createdby submitting responses to each of the six scaleson which initiation activities were rated to aprincipal-components factor analysis and vari-max rotation. Separate analyses were performedfor men and women. The resulting factor struc-ture for each gender was the same.

For men, two factors with eigenvalues greaterthan 1.0 explained 60% of the variance in rat-ings. On the first factor, ratings for physical andpsychological challenge and for pain loaded.81, .67, and .76, respectively. On the secondfactor, ratings for deviance and embarrassmenthad loadings of .90 and .86, respectively. Rat-ings of fun were not strongly associated witheither factor (loadings � .45).

For women, the same two factors emerged(eigenvalues � 1.0) and explained 62.5% of thevariance in ratings. As for men, Factor 1 com-prised ratings for physical and psychological

challenge and pain (loadings � .88, .80, and.89, respectively). Factor 2 contained ratings fordeviance and embarrassment (loadings of .81and .79, respectively). Once again, ratings forfun were not strongly associated with eitherfactor (loadings � .35).

On the basis of these analyses, three compo-nents of initiation experiences were identifiedand measured for men and women. Compositescores for experiencing harsh treatment encom-passed the averaged ratings of the physical chal-lenge, psychological challenge, and pain in-volved in initiation activities (interitem correla-tions ranged from .50 to .61; � � .78).Composite scores for engaging in activities in-volving social deviance incorporated the aver-aged ratings of initiation experiences perceivedas deviant and embarrassing (interitem correla-tion � .52; � � .70). Ratings of the perceivedfun of initiation activities were maintained asthe third component.

Hierarchy, the predicted variable, was as-sessed in two ways. First, we constructed apower discrepancy score by subtracting partic-ipant ratings of the power new members hadover other group members from the power lead-ers had over group members. Second, we com-puted a role differentiation measure by sum-ming the number of leadership positions partic-ipants identified as part of the structure of thegroup (president/captain/group leader, copresi-dent/cocaptain/coleader, vice president, secre-tary, treasurer). Separate regression analyseswere computed for each measure of hierarchy.

Predictor sets for these analyses initially in-cluded the three categories of initiation experi-ences (fun, harsh, and deviant), gender, and alltwo- and three-way interactions among thesevariables. A hierarchical regression approachwas used to control for the underlying overlapbetween perceived fun and perceptions of harshand deviant activities. Thus, ratings of fun andparticipant gender were added first to the pre-dictor set, followed by ratings of harsh treat-ment and social deviance, then by all two-wayinteractions, and finally by all three-wayinteractions.

When perceptions of the deviance and harsh-ness of initiations were added to the model,predictive power increased significantly, R �.17, R2 change � .025, F(2, 263) � 3.42, p �.04. A summary of these results appears inTable 2. As expected, initiations involving so-

114 KEATING ET AL.

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

cial deviance predicted greater power differen-tials between leaders and members. But unex-pectedly, harsh initiation activities were associ-ated with marginally less of a distinctionbetween leader and member power.

Because prediction of the perceived powerdifferential between leaders and initiates wasadditionally altered by entry of the set of two-way interactions, R � .27, R2 change � .045,F(6, 257) � 2.07, p � .06, separate regressionanalyses were computed for women and men.The results for women reflected the overall re-sults; social deviance predicted perceptions ofenhanced power differentials, B � .33, � � .24,t(124) � 2.75, p � .01, whereas harsh treatmentwas linked to reductions in these perceptions,

B � �.24, � � �.19, t(124) � �2.16, p � .04.The case for men was more complex; onlyperceptions of social deviance and fun com-bined predicted relative power, B � .13, � �.68, t(130) � 2.08, p � .04. When men weretrichotomized into categories based on reporteddeviance and fun, power differentials werehighest for those who perceived initiation activ-ities to be both extremely fun and highly deviant(see Figure 2).

The hierarchical regression analysis for oursecond measure of hierarchy, role differentia-tion, yielded R � .30, R2 change � .08, F(2,264) � 11.92, p � .01, when perceptions ofsocial deviance and harshness were added to themodel. Table 3 summarizes these results. Onceagain, whereas social deviance conformed topredictions by being linked to increased roledifferentiation, perceived harshness producedresults contrary to expectations by being asso-ciated with reductions in this measure of hier-archy. The additional entry of interaction setsdid nothing to improve prediction ( ps � .50).

Individuals belonging to groups character-ized by formal hierarchy were expected to re-port more frequent engagement in initiation ac-tivities overall. Only one of our two indexes of

Table 2Summary of Hierarchical Linear RegressionAnalysis for Variables Predicting Perceptions ofthe Power Differential Between Leaders and NewMembers in Single Gender Groups (n � 269)

Initiation activities B � t p

Socially deviant 0.18 .14 2.15 .032Harsh �0.16 �.13 �1.94 .054Fun 0.01 .03 0.54 .588

Figure 2. Mean power differences between leaders and members of male groups as rated bymen trichotomized into levels of social deviance and levels of fun experienced duringinitiations. Dashed-and-dotted line indicates low levels of fun, dashed line designates mediumlevels of fun, and solid line indicates high levels of fun.

115UNPACKING INITIATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

hierarchy yielded the expected result. Consis-tent with predictions, the correlation betweenfrequency of engagement and role differentia-tion was r(136) � .37, p � .01, for men andr(129) � .27, p � .01, for women. Correlationsbetween frequency of engagement and powerdiscrepancy were nonsignificant for men,r(136) � .09, and for women, r(129) � .11,ps � .05.

To summarize, the results linking initiationpractices to social hierarchy were mixed withregard to predictions. Consistent with expec-tations, initiations characterized as involvingsocial deviance were linked to perceptions ofthe relative power of leaders over new mem-bers and to elaborated power structures. Con-trary to predictions, initiations typified byharsh treatment were associated with dimin-ished differentials for each index of hierar-chy. One of two measures of hierarchy, rolediversity, was associated with more frequentengagement in initiation activities, aspredicted.

Predicting identity from initiation experi-ences. The predicted (dependent) variableswere measurements of group and individuatedidentity operationalized by two questions, re-spectively: “How important is the group toyou?” and “How important are you to thegroup?” These measures intercorrelated formen, r � .48, p � .01, and for women, r � .70,p � .01. Because we predicted differences inthe relationships between these aspects of iden-tity and initiation practices, we maintained themas separate dependent variables and computedseparate analyses.

For these tests, hierarchical multiple linearregression analysis was again used to separateoverlapping perceptions of initiation experi-ences within predictor sets. Ratings for fun wereentered first, along with gender. Next, percep-tions of harsh and deviant initiations were

added. The entry of all two- and then three-wayinteractions followed.

Importance of the group to the individual.Participant reports of the group’s importanceto them from rated fun and gender alone pro-duced significant prediction, R � .38, F(2,266) � 22.58, p � .01, but improved signif-icantly when the other perceptions of initia-tions were added to the model, R � .42, R2

change � .03, F(2, 264) � 4.85, p � .01.Interaction sets did little to improve predic-tion ( ps � .10). Table 4 presents a summaryof contributing factors.

As Table 4 reveals, initiation activities per-ceived as fun contributed significantly to per-ceptions of the importance of the group to theindividual. Above and beyond perceptions offun, harsh initiations were associated withelevated perceptions of the importance of thegroup to the individual, confirming predic-tions. Social deviance, however, showed noindependent relationship with this measure ofidentity.

Importance of the individual to the group.Participants’ reports of their individual im-portance to the group were also regressedon perceptions of initiations, gender, andall two- and three-way interactions. As shownin Table 5, fun was the only significant inde-pendent predictor in the set. Perceived funduring initiations was associated with en-hanced individuated importance to the group,at least in the eyes of the individuals whobelonged to it.

In Study 2, a laboratory study, we testedwhether relatively severe initiations fostered so-cial dependency in the form of conformity togroup opinion, and attraction to group membersand its activities.

Table 3Summary of Hierarchical Linear RegressionAnalysis for Variables Predicting Perceptions ofRole Diversity in Single Gender Groups (n � 269)

Initiation activities B � t p

Socially deviant 0.34 .28 4.66 .001Harsh �0.18 �.16 �2.48 .014Fun 0.01 .02 0.39 .696

Table 4Summary of Hierarchical Linear RegressionAnalysis for Variables Predicting Ratings of theImportance of the Group to the Individual (n �269)

Initiation activities B � t p

Socially deviant 0.01 �.05 �0.79 .430Harsh 0.17 .19 3.11 .002Fun 0.40 .42 7.26 .001

116 KEATING ET AL.

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

Study 2

Method

Participants. Eighty 1st- and 2nd-year maleundergraduate students participated in the studyand received credit toward an introductory psy-chology laboratory. Data from 5 men who de-tected the purpose of our conformity manipula-tion were dropped and another man opted not togive consent, which resulted in a final sampleof 74 participants.

Procedure. Two induction protocols wereused to establish generalizability. For the cha-rades protocol, participants performed eitherembarrassing or innocuous charades purport-edly as a test of “creative body movement.” Forthe motor skills protocol, participants per-formed either embarrassing or innocuous motortasks purportedly as a test of fine and gross“motor skills.” Participants were inducted indi-vidually. They were told they would performbehaviors for evaluation by upper-class re-search students and could be selected to helpevaluate the performances of other students.Each participant was escorted to the observationroom to meet two male, upper-class studentresearchers who acted in an outgoing, confidentmanner by shaking hands and introducing them-selves with broad smiles according to a stan-dardized script.

The upper-class students were actually con-federates of the experimenter made to seemattractive and powerful. The observation roomthey appeared to control was appointed to looklike their “hangout.” It featured comfortable,upholstered furniture, a rug, and warm, incan-descent lighting. Each participant was shownhow the upperclassmen would be able to ob-serve and judge their performance by lookingthrough a one-way mirror into the performanceroom. These details were meant to convey that

the benefits of membership included status (be-ing an evaluator of other men’s performancesrather than being the performer evaluated), af-filiation with upperclassmen, and privilege (ac-cess to creature comforts such as food and com-fortable seating).

Once back in the performance room, partici-pants completed self-report measures (measur-ing handedness and personality traits). Eachwas instructed to choose the first of a seriesof 10 index cards from the table and, at thesound of a buzzer, act out the activity describedon it for 15 s. At the sound of the next buzzer hewas to stop the activity, pick up card numbertwo, and portray the activity described on it.This procedure was to continue until the re-searchers determined that the participant hadacquired enough points to stop the first part ofthe experiment and join their group. Havingdelivered these instructions, the experimenterleft the room.

After each participant had completed the firstfour activities, he was interrupted by the exper-imenter and told that he could join the upper-class researchers and judge the videotapedperformances of other students. In fact, the con-federates were prevented from watching perfor-mances. They were kept unaware of which mo-tor activities participants performed in order toprevent any systematic bias in the way confed-erates acted toward participants in differentconditions.

Participants were randomly assigned to eithersevere or innocuous induction conditions.Those assigned to the severe induction per-formed embarrassing activities, whereas partic-ipants assigned to the innocuous induction per-formed mundane ones. Embarrassing charadeswere as follows: kneel and polish the boots ofsomeone standing in front of you; act like a dog;act like a slave; and lying on your stomach,place your chin on the floor and hands behindyour back. Benign charades were as follows:row a boat; brush your teeth; read a newspaper;and juggle. Embarrassing motor skills were asfollows: dance to music played on a boom box;trace the outline of your lips on the mirror usingthe lipstick; bowl with a miniature bowling set;and play the game Perfection. For the innocuousinduction, bouncing a ball to music and tracinga pencil maze were substituted for dancing andusing lipstick.

Once a participant joined confederates in the

Table 5Summary of Hierarchical Linear RegressionAnalysis for Variables Predicting Ratings of theImportance of Individuals to the Group (n � 269)

Initiation activities B � t p

Social deviance 0.01 �.08 �1.37 .170Harsh 0.01 .02 0.34 .733Fun 0.34 .34 5.65 .001

117UNPACKING INITIATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

observation room, he watched a series of vid-eotaped clips showing other students (accom-plices) performing novel, innocuous activitieseither poorly, impressively, or in a mediocrefashion. After every clip, the confederates andparticipant each gave a verbal rating of theperformance on a 10-point scale. On five criticalconformity trials, the confederates respondedfirst and gave similar (within 1 point), unrealis-tically positive or negative ratings (according topilot tests of the stimulus tapes), and the partic-ipant’s spoken response was subsequentlyrecorded.

After the judgment task, the confederates leftthe room on a pretext. The experimenter askedthe participant to give the Psychology Depart-ment feedback on how well the experiment wasrun. Each participant was left alone to completeseveral 5-point, unipolar rating scales withpoles labeled not at all to extremely. On thebasis of a priori criteria, two scales assessed thecongeniality of the evaluators (i.e., their friend-liness and whether they were “easy to be with”;� � .95). Three scales assessed perceptions oftheir social power (i.e., competence, knowl-edge, and influence; � � .88). Two scales in-dexed their poise (i.e., poise, confidence; � �.85). These scales were intermingled with othersfor authenticity (e.g., politeness of researchers;clarity of instructions). Participants also ratedwhether the experiment was fun (i.e., interest-ing, creative, fun, and not boring; � � .94) andhow embarrassed it made them feel. Last, par-ticipants were asked to write down what theythought the experiment was about, and thenthey were thanked and debriefed.

Results and Discussion

Conformity. To measure conformity, wecalculated the absolute difference between theaverage ratings of the confederates and that ofthe participants for each of the five critical tri-

als, in which the two confederates gave uni-formly biased ratings (varying by a point orless). These differences were added to formdiscrepancy scores. Thus, conformists gave rat-ings that were least discrepant from the biasedratings of confederates.

To test whether severe treatment inducedconformity, we compared discrepancy scoresfor each induction condition using a one-wayanalysis of variance (ANOVA). As predicted,men who experienced the discomforting initia-tion conformed most by producing discrepancyscores that strayed least from the biased judg-ments of confederates (Ms � 7.26 and 11.10),F(1, 72) � 12.67, p � .01.

Social–emotional bonds. To test predic-tions that discomforting initiations would ele-vate ratings of the group’s attractiveness andincrease the perceived enjoyment of the exper-iment, we computed a multivariate analysis ofvariance (MANOVA) comparing inductionconditions (innocuous vs. discomforting) acrossratings of the confederates’ poise, congeniality,and power and ratings of how much fun andembarrassment participants experienced. Theoverall main effect for induction was signifi-cant, F(5, 68) � 2.26, p � .05. Table 6 depictsthe univariate means differences on which thisresult was based. The univariate tests indicatedthat participants who received the discomfort-ing induction perceived confederates as morepowerful, F(1, 72) � 5.88, p � .02, and asmarginally friendlier, F(1, 72) � 3.34, p � .07,than did those who received the innocuous in-duction. Participants who experienced the dis-comforting induction also tended to report hav-ing had more fun than did those inducted viainnocuous procedures, F(1, 72) � 3.51, p � .06.Induction did not significantly alter perceptionsof the confederates’ poise or of inductees’ per-ceived embarrassment by the end of the exper-iment, Fs(1, 72) � 1.0.

Did perceptions of the group and the experi-

Table 6Mean Ratings of Confederates and Experiment by Men Who Experienced EitherInnocuous or Discomforting Inductions

Induction Poise Power Congeniality Fun Embarrassing

Innocuous 4.15 3.40 4.16 3.76 3.24Discomforting 4.22 3.81 4.42 4.05 3.00

Note. For each induction group, n � 37.

118 KEATING ET AL.

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

ence predict conformity? The correlations pre-sented in Table 7 indicate that discrepancyscores were significantly associated only withthe men’s perceptions of the confederates’power; the greater the degree of power attrib-uted to inductors was, the less men’s opinionsdiffered from them.

Did perceptions mediate the effect of treat-ment condition on conformity? We performedseparate mediation analyses (Baron & Kenny,1986) on ratings of confederates (power, poise,congeniality) and the experiment (fun, embar-rassing). Beta for the condition–discrepancy re-lationship was �.39 ( p � .01). Only powerrevealed a mediating role, and it was weak.Specifically, confederates were perceived to bemore powerful by men in the discomfortingthan in the innocuous condition (� � .27, p �.02), and these perceptions were associated withlowered discrepancy scores (� � �.26, p �.02). But partialing power from the equationreduced the condition–discrepancy relationshiponly slightly (� � �.31, p � .01), suggestingthat perceptions of confederates’ power onlypartly accounted for the effect of initiation con-dition on discrepancy (conformity) scores.

Summary. Consistent with a dependencyinterpretation, severe treatment produced in-ductees who yielded most to conformity pres-sure from the group. In addition, men who ex-perienced a psychologically discomforting in-duction before joining a group tended to rate itsmembers as more powerful and friendly andreported having had more fun than did menexposed to an innocuous induction. In sum, byorchestrating a discomforting induction routinewe elevated men’s dependence on and attractionto the agents who inducted them. Perceptions ofthe group’s power rather than aspects of theirattractiveness predicted compliance.

However, the uniformly attractive demeanorof our charismatic group may have masked theeffects of attractiveness of the group. InStudy 3, we varied the attractiveness of the

group. We also refined our design by inviting“qualified” participants to choose whether theywished to join the group. This modificationmade the process more closely resemble real-world fraternity and sorority rushing.

In Study 3, female undergraduate studentswere “initiated” by feeling and smelling eitherunpleasant or innocuous objects while blind-folded. Following induction, they joined upper-class students (confederates) who pressuredthem to conform. Confederates either behavedin a warm, friendly, attractive fashion or mani-fested a cool, formal, less appealing demeanor.Ratings of the confederates and the experiencewere collected from inductees as in Study 2. Inaddition, self-reports of negative and positivefeelings were gathered, and we surreptitiouslymeasured the proximity inductees maintained toconfederates as indicative of social bonds(Bowlby, 1969/1982; Rajecki et al., 1978).

Consistent with maltreatment effects, a rela-tively severe induction was predicted to pro-mote dependency in the form of greater confor-mity and closer proximity to confederates,whether or not their demeanor was attractive.Social bonds are characterized by anxiety inresponse to separation (Bowlby, 1969/1982;Rajecki et al., 1978). Thus, negative arousal wasexpected to intensify when confederates leftseverely initiated inductees alone.

Study 3

Method

Participants. Forty female undergraduatestudents participated in the study and earnedpsychology laboratory credit. Data from 2women who detected the purpose of our con-formity paradigm were subsequently omitted,and a third opted to withdraw. A total of 37participants remained.

Procedure. Individual participants wereseated at a table facing a video camera. The

Table 7Correlations Between Men’s Discrepancy Scores and Their Ratings of Confederates and the Experiment(n � 74)

Variable Poise Power Congeniality Fun Embarrassing

Discrepancy .14 �.34*** �.16 �.19 .02

*** p � .005 (one tailed).

119UNPACKING INITIATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

study was described as an investigation of cre-ativity, personality, and the senses. The partic-ipant’s task was ostensibly to explore severalitems using only her sense of touch and smelland then to describe the items as creatively asshe could. To block the visual sense, each par-ticipant was told she would be blindfolded. Par-ticipants were informed that some of the items“might be unpleasant” to touch or smell, thattwo upper-class student researchers would ob-serve her through the one-way mirror, and thather responses would be videotaped. Each par-ticipant was also told that she might have theopportunity to join the upper-class researchersand rate the creativity of other students.

Next, participants were escorted to the ob-servation room (appointed to look like thewomen’s “hangout”) to meet the upper-classresearchers. There, on the basis of randomassignment, two female upper-class studentresearchers (confederates) introduced them-selves in either a warm, friendly, outgoingmanner (attractive group) or a cool, formal,less outgoing way (unattractive group) in ac-cordance with standardized scripts.

After participants returned to the perfor-mance room, the confederates donned ear plugsand faced away from the one-way mirror sothey would neither see nor hear the participantmanipulating and describing objects. In thisway, confederates remained unaware of partic-ipants’ induction condition.

Meanwhile, each participant completed amood checklist that included descriptors withpositive tone (i.e., happy, excited, enthusiastic,secure, calm, confident) and negative tone (i.e.,sad, lonely, frightened, angry, nervous, anxious,tense). Participants also completed three briefpersonality inventories (for purposes of futureresearch). The experimenter returned with oneof two randomly selected boxes. The boxeswere kept closed so that the experimenter wasunaware of their contents and thus of experi-mental condition. Each box contained stimulusitems housed in plastic containers. On the basisof pilot tests, stimuli for the discomforting (se-vere) hazing condition were a “water snake”toy, cooked shrimp, a dog snack food, and, as acontrol item, feathers. For the innocuous treat-ment, stimulus items included a rubber ball,aluminum foil, a scented candle, and feathers.

As instructed, each blindfolded participantsequentially removed items from boxes and ma-

nipulated and creatively described them. Duringthis time, the experimenter waited in an adja-cent room where she could neither hear nor seeparticipants. When finished, participants sig-naled the experimenter by pressing a button thatsounded a buzzer in the adjacent room andalerted the experimenter by switching on a light.The participant’s blindfold was removed, andshe was given a choice; she could either join theupper-class women in the next room and judgethe creativity of other students or put the blind-fold back on and describe new objects. If aparticipant chose to continue with new objects(7 did), the experimenter left the room ostensi-bly to get the objects, only to return a fewmoments later to explain that the keys to thecabinet that housed them could not be found.The participant was then asked to join the re-searchers (all agreed). Data from participantswho initially chose to continue to judge itemswere analyzed to avoid a self-selection bias.

Once the participant entered the observationroom, the confederates offered her a chair theyhad been using as a footstool. The edge of thechair was placed 10 in. from the confederates,and participants typically pulled it back beforesitting down. The experimenter surreptitiouslyrecorded how far back participants placed thechair, estimating the distance by using tile gridson the floor. Interrater reliability for the distancemeasurements was calculated across the 25% ofobservations for which a confederate as well asthe experimenter estimated and recorded dis-tances. Reliability was acceptable (r � .75).Distance estimates were converted to a scalescore ranging from 0 (did not move chair/movedchair back from confederates by less than oneinch) to 4 (moved chair back 18 inches ormore).

After the participant was seated, the experi-menter explained that the second part of thestudy involved judging creative responsesrather than inventing them. On four critical tri-als, participants were exposed to the similarlybiased judgments of confederates before render-ing their own opinion. Biased judgmentswere 1.5 standard deviations higher or lowerfrom mean ratings determined by pilot studies.Oral and written judgments were involved. Forcritical oral trials, descriptions were read outloud by the experimenter, and the participantreported her judgment out loud after the con-federates had reported theirs, using a 10-point

120 KEATING ET AL.

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

rating scale from not at all creative to excep-tionally creative. On critical written trials, par-ticipants read the creative statements and re-corded their judgments in pen along creativityrating scales adjacent to scales already com-pleted by confederates. To disguise the confor-mity manipulation, critical trials were intermin-gled with 10 other trials in which confederateseither disagreed with one another or gave real-istic (unbiased) ratings, or participants re-sponded first. Differences between confeder-ates’ scripted ratings and those produced byparticipants formed the basis of discrepancyscores: the smaller the discrepancy, the greaterthe conformity.

With ratings completed, the confederates leftthe lab. The experimenter reentered the obser-vation room, and the participant completed asecond mood checklist. Next, the participantwas escorted back to the performance room,where the experimenter left her alone to fill outpostexperimental questionnaires ostensibly forPsychology Department feedback. In actuality,these assessments of the confederates and theexperiment were identical to those used inStudy 2 and served largely as a check for theattractiveness manipulation.

Results and Discussion

Manipulation check. Participant ratings ofthe congeniality of confederates confirmed thatthe manipulation of attractive versus unattrac-tive demeanor worked overall. An attractive-ness main effect, F(1, 33) � 39.73, p � .01,reflected higher congeniality ratings for attrac-tive than for unattractive confederates(Ms � 4.3 vs. 2.9). However, an interactionbetween induction condition and attractiveness,F(1, 33) � 9.32, p � .01, revealed that per-ceived differences in congeniality were moredramatic under severe (Ms � 4.5 vs. 2.4), F(1,17) � 42.03, p � .01, than under mild(Ms � 4.0 vs. 3.4) induction conditions, F(1,16) � 5.54, p � .04.

Conformity. A 2 (induction: innocuous vs.discomforting) � 2 (group attractiveness: lowvs. high) ANOVA was performed on discrep-ancy scores to test predictions that severe initi-ations would generate conformity. Only the ef-fect of induction was significant (Ms � 3.10and 4.25), F(1, 33) � 6.03, p � .02, indicatingthat discrepancy scores were lower for women

initiated under the relatively severe treatment,regardless of the attractiveness of the demeanorof the group. As predicted, experiencing dis-comfort during induction resulted in compara-tively greater yielding to conformity pressure(i.e., less discrepancy) from confederates.

Affective bonds: Proximity and mood. Totest whether the discomforting induction gener-ated closer proximity, we computed a 2 (induc-tion: innocuous vs. severe) � 2 (group attrac-tiveness: low vs. high) ANOVA on the scaled (0to 4) estimates of the distance inductees choseto sit from confederates. Means for this analysisare displayed in Table 8. The analysis revealeda main effect for induction, F(1, 33) � 3.98,p � .05, and indicated that participants whoexperienced the discomforting induction main-tained closer proximity to confederates thanthose treated less severely. No other effectswere significant (Fs � 1.0).

We measured changes in mood by subtract-ing the number of positive and negative moodstates endorsed by women prior to inductionfrom the same measure taken after the confor-mity task when confederates left the room.Means for the positive (felt better) and negative(felt worse) mood change scores appear in Ta-ble 8. The 2 (induction) � 2 (group attractive-ness) MANOVA yielded a marginally signifi-cant main effect for induction, F(2, 32) � 2.57,p � .09. Only the univariate test for negativemood supported the multivariate result, indicat-ing that severely hazed women experiencedheightened negativity as they sat alone in theobservation room near the end of the experi-ment, F(1, 33) � 4.38, p � .04. There were noother effects for negative mood ( ps � .30).

Did women’s affective reactions or percep-tions predict their tendency to conform? Thecorrelations that appear in Table 9 indicate thatchoosing to sit closely to confederates was as-sociated with yielding to group opinion. To a

Table 8Mean Distance Rating and Mood Scores forWomen Who Experienced Either Innocuous orDiscomforting Inductions

InductionDistance

ratingFelt

betterFelt

worse

Innocuous 2.78 �.08 �.12Discomforting 1.63 �.07 .05

121UNPACKING INITIATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

lesser degree, attributing greater power to groupmembers was also associated with conformity.

Similar to Study 2, we tested whether affector perceptions mediated the effect of conditionon conformity. Separate mediation analysespredicting conformity were performed for rat-ings of confederates (power, poise, congenial-ity) and the experiment (fun, embarrassing) andalso on mood and distance scores. Beta for thecondition–discrepancy relationship was �.38( p � .02). Only distance came close to meetingcriteria for mediation. Severely hazed womensat closer to confederates (� � �.31, p � .06),and beta for distance was positively but notsignificantly related to discrepancy scores (� �.22, p � .20).

Summary. Like men, women who experi-enced a severe induction yielded to the opinionsof the confederates responsible for their treat-ment. Whether confederates were portrayed asattractive or unattractive made no difference:Either way, a severe induction increased depen-dency on the group as measured by yielding tothe biased opinions of its members. Womenwho received a discomforting induction showedadditional signs of what could be construed assigns of maltreatment effects: They maintainedclose proximity to confederates and experi-enced negative affect when confederates leftthem alone. These reactions are reminiscent ofthose described by attachment theorists whohave studied the effects of maltreatment on hu-man and nonhuman social bonds (Rajecki et al.,1978). Moreover, women’s proximity to con-federates and, marginally, the power they attrib-uted to them predicted conformity to groupopinion.

General Discussion

We set out to “unpack” initiations and theirconsequences for individual relationships withgroups on college campuses from a functional

perspective. In the field, we assessed whetherinitiation practices related to the developmentof group-relevant attitudes and skills, to thereinforcement of group hierarchy, and to groupidentity. In the laboratory, we tested whetherrelatively severe inductions spawned confor-mity and attraction to group members as mani-festations of social dependency. The resultsacross studies suggested that hazing functionsto preserve group features and promote cogni-tive, social, and emotional manifestations ofsocial dependency.

The premise that initiation activities functionto preserve group-relevant skills and attitudeswas largely confirmed. For example, the initia-tion practices of groups with different missionsvaried as predicted. The success of athleticteams depends on the physical endurance of itsmembers. The success of Greek-letter organiza-tions requires the establishment of an exclusivesocial network, a goal well served by empha-sizing the uniqueness of group membership andits social distinctiveness from out-groups. Con-sistent with this thinking, athletes characterizedtheir initiation activities as physically challeng-ing and painful, whereas members of Greek-letter organizations highlighted the role of so-cial deviance in initiation activities. Taken to-gether, the implication is that initiationsfunction as a training ground for group-relevantskills and attitudes. However, studies that in-clude a wider array of organizations, missions,and hierarchies are needed for confirmation.

Projections about the functional fit betweenhierarchy and initiation severity met with mixedsuccess. The psychological techniques of socialcontrol manifested in initiations were predictedto be especially severe and frequent in groupscharacterized by clear power differences be-tween leaders and members and elaborated rolediversity. To preserve these features of hierar-chy, groups were expected to sponsor initiationsthat effectively tuned initiates’ responses to

Table 9Correlations Between Women’s Discrepancy Scores and Their Ratings of Confederates, the Experiment,and Mood (n � 37)

Variable Poise Power Congeniality Fun DistanceFelt

betterFelt

worse Embarrassing

Discrepancy �.22 �.24* .17 .12 .33** �.18 �.09 �.13

* p � .05 (one-tailed). ** p � .025 (one tailed).

122 KEATING ET AL.

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

them. As predicted, initiations requiring newmembers to engage in social deviance werelinked to enhanced perceptions of the relativepower of leaders over new members and to rolediversity. In addition, more frequent engage-ment in initiation activities was associated withgreater role diversity, as predicted. Yet our sec-ond measure of hierarchy, power differentials,failed to predict more frequent engagement ininitiation activities. Moreover, initiations typi-fied by harsh treatment (as distinct from socialdeviance) were associated with diminished (in-stead of increased) differentials for each indexof hierarchy. Rather than undermining groupmembers’ sense of power, harsh treatment re-quiring initiates to prevail under physical andpsychological duress actually seemed to em-power them by diminishing perceptions of in-group power differentials.

We hypothesized that the overarching func-tion of all types of severe treatment duringinitiations was to enhance dependency on thegroup, expressed cognitively, socially, andemotionally. Consistent with this idea, howmuch personal importance individuals ascribedto the group they identified with most was pre-dicted not only by how much fun participantshad during initiations but also by how harshinitiation activities were perceived to be. Thecombined, independent effects of these dispar-ate experiences, fun and harsh treatment, per-haps provide a human parallel to the social–emotional impact of the nurturing and punish-ing treatments Fisher dispensed to puppies andHarlow delivered to monkeys in order to primesocial dependency (see Cairns, 1979; Rajecki etal., 1978). If social identity is a social–cognitiveconsequence of social dependency, then theseresults are compatible with the dependency ex-planation of maltreatment effects.

The results from two laboratory studies alsosupported the social dependency interpretation.The male and female undergraduate participantswho experienced a discomforting inductionshowed heightened dependence on the opinionsof group members. Whether or not group mem-bers were viewed as particularly attractive, rel-atively severe treatment increased vulnerabilityto group influence relative to mild induction.We construed conformity as a form of depen-dence and a manifestation of maltreatment ef-fects. Measurement of more traditional attach-ment behaviors augmented these results:

Women who experienced harsh treatment main-tained close proximity to confederates and ex-perienced negative affect after confederates left.In a sense, our laboratory tests simulated the“strange situation” for adults, and inductionprocedures elicited social bonds with its agents.

To us, the pattern of results across studiesseems tailored to a social dependency interpre-tation. The fact that harsh treatment was asso-ciated with a heightened personal stake in groupmembership could be construed as a social–cognitive manifestation of maltreatment effectsand a form of social dependency. Similarly,conformity and group attraction were used toindex behavioral, social, and emotional compo-nents of dependency in the wake of maltreat-ment effects. Such effects are normally appliedto individual attachments but are used here toexplain social bonds between individuals andgroups. Some may be bothered by this leap inlevels of conceptual analysis and feel unready tobridge theoretical approaches to interpersonalaffectional bonds, group attachments, and socialidentity. Others, however, have welcomed at-tempts at a synthesis. In particular, Hazan andZeifman (1999), Shaver et al. (1988), Nelsonand Quick (1991), Baumeister and Leary(1995), Aberbach (1995), and Smith et al.(1999, 2001) have all made cogent argumentsfor a distillation of the basic brew of motivesand proclivities that fuel human social bondsacross contexts and lifetimes.

The integrative story of maltreatment effectsand social bonds might go like this: Maltreat-ment generates uncertainty, emotional ex-tremes, and stress, which elicit affectional andaffiliative behavior toward individuals orgroups that is normally adaptive (Baumeister &Leary, 1995; Bowlby, 1969/1982; DeVries,Glasper, & Detillion, 2003; Festinger, 1954;Florian, Mikulincer, & Hirschberger, 2002;Gump & Kulik, 1997; Rajecki et al., 1978;Schachter, 1959; Smith et al., 1999; Uchino,Cacioppo, & Kiecolt-Glaser, 1996). The social–emotional bonds that result increase vulnerabil-ity to social influence and indoctrination(Baron, 2000; Dolinski et al., 2002; Van Duuren& Di Giacomo, 1996, 1997) and foster a deper-sonalized, social identity (Baron, 2000; Hogg,2001; Rosenblatt, 1964). At the same time, mal-treatment signals differential power: The deper-sonalized individual’s dependence on the groupis not matched by any dependence the group has

123UNPACKING INITIATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

on individuals. This integration suggests thatharsh initiations would enhance perceptions ofthe importance of the group to the individual butnot necessarily heighten members’ feelings oftheir individuated importance to the group. Ourstudies are largely consistent with this tellingand support a social dependency interpretationof initiations.

A few laboratory findings are compatiblewith a dependency interpretation of maltreat-ment effects. Van Duuren and Di Giacomo(1996, 1997) also uncovered a link betweenharsh treatment and dependency across severalexperimental studies. These researchers re-ported that undergraduates who experiencedpublic failure were not only more eager to af-filiate, they were also more vulnerable to socialinfluence (Van Duuren & Di Giacomo, 1996,1997). For example, after it was publicly an-nounced that they had performed poorly on atest, participants were significantly more likelyto join with a confederate and follow that per-son’s suggestions to participate in a theft (VanDuuren & Di Giacomo, 1996). Van Duuren andDi Giacomo (1997) proposed a model in whichsocial degradation increases both affiliative ten-dencies and vulnerability to social influence asways of gaining social approval. Applied to thecontext of hazing, achieving social approvalfrom the group would likely be a first step alongthe road to group identity and attachments.

Behavioral science offers surprisingly littleinsight into the tenacious nature of initiationtraditions. Although the costs of initiation toindividuals are well documented (e.g., Finkel,2002; Galanter, 1999; Hoover & Pollard, 2000),its real or perceived benefits are not, and so it isno wonder that formal studies have yet to iden-tify the roots of hazing’s impressive resistanceto being curbed. Cross-cultural studies couldhelp. For instance, initiation traditions may beespecially persistent in societies where earlyidentities must be attenuated or altered (Segall,Berry, Dasen, & Poortinga, 1999). In the UnitedStates, for example, it may take a mighty per-suader to suppress the individualism inherent inAmerican identity and supplant it with a collec-tive, group identity. Severe hazing rituals mayoperate in just this way. As Confucius notedmany centuries ago, rituals are powerful sourcesof social influence and control (Confucius, 551–479 BCE, translated by Slingerland, 2003, p.171).

As we compiled this research, hazing at anearby college campus ended in the death of ayoung man who expired after engaging in atradition of ingesting large quantities of waterthrough a funnel (Foderaro, 2003). On the evebefore his death, pledges would have undoubt-edly described their initiation activities as devi-ant but fun, their leader as powerful, and thefeeling of belonging to the group as an impor-tant source of their identity. What would it havetaken for any one of them to change his mind?

Misplaced human social attachments fit thecross-species pattern of behavior spawned bymaltreatment effects. Treatment that incorpo-rates elements of punishment can enhance out-ward signs of attachment toward abusive figures(Rajecki et al., 1978). Supporting the motiva-tion to bond are proximate physiological bene-fits reaped by organisms drawn to those whotypically support and protect them (Uchino etal., 1996) as well as underlying evolutionaryforces responsible for human sociability(Baumeister & Leary, 1995). Maltreatment ef-fects may explain why abused children remainattached to abusive parents, why battered wivesendure relationships with brutal spouses, andwhy hostages develop emotional attachments totheir captors (Boulette & Andersen, 1986;Bowlby, 1969/1982; Dutton & Painter, 1993;Rajecki et al., 1978). Charismatic groups and,perhaps at some level, more ordinary groupscultivate this same kind of devotion by servingboth as agents of abuse and as anchors ofattachment.

References

Aberbach, D. (1995). Charisma and attachment the-ory: A crossdisciplinary interpretation. Interna-tional Journal of Psycho-Analysis, 76, 854–855.

Ainsworth, M. D. S. (1972). Attachment and depen-dency: A comparison. In J. L. Gewirtz (Ed.), At-tachment and dependency (pp. 97–137). Washing-ton, DC: V. H. Vincent.

Aronson, E. (1995). The social animal (6th ed.). NewYork: Freeman.

Aronson, E., & Mills, J. (1959). The effect of severityof initiation on liking for a group. Journal ofAbnormal and Social Psychology, 59, 177–181.

Auerbach, S. M., Kiesler, D. J., Strentz, T., Schmidt,J. A., & Serio, C. D. (1994). Interpersonal impactsand adjustment to the stress of simulated captivity:An empirical test of the Stockholm syndrome.

124 KEATING ET AL.

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

Journal of Social and Clinical Psychology, 13,207–221.

Baron, R. M., & Kenny, D. A. (1986). The modera-tor–mediator variable distinction in social psycho-logical research: Conceptual, strategic, and statis-tical considerations. Journal of Personality andSocial Psychology, 51, 1173–1182.

Baron, R. S. (2000). Arousal, capacity, and intenseindoctrination. Personality and Social PsychologyReview, 4, 238–254.

Baumeister, R. F., & Leary, M. R. (1995). The needto belong: Desire for interpersonal attachments asa fundamental human motivation. PsychologicalBulletin, 117, 497–529.

Bjorkqvist, K. (1994). Sex differences in physical,verbal, and indirect aggression: A review of recentresearch. Sex Roles, 30, 177–187.

Boulette, T. R., & Andersen, S. M. (1986). “Mindcontrol” and the battering of women. Cultic Stud-ies Journal, 3, 25–35.

Bowlby, J. (1982). Attachment and loss: Vol. I. At-tachment. New York: Basic Books. (Original workpublished 1969)

Brehm, J. W. (1960). Attitudinal consequences ofcommitment to unpleasant behavior. Journal ofAbnormal and Social Psychology, 60, 379–383.

Broadbent, L. (2003, January). Dangerous liaisons:The hidden perils of America’s secret societies.Cosmopolitan (UK), 89–92.

Buss, D. M. (2004). Evolutionary psychology: Thenew science of the mind. Boston: Pearson.

Cairns, R. B. (1979). Social development: The ori-gins and plasticity of interchanges. San Francisco:Freeman.

Cassidy, J. (1999). The nature of the child’s ties. In J.Cassidy & P. R. Shaver (Eds.), Handbook of at-tachment: Theory, research, and clinical applica-tions (pp. 3–20). New York: Guilford Press.

Cialdini, R. B. (2001). Influence: Science and prac-tice (4th ed.). Boston: Allyn & Bacon.

Crick, N. R., & Grotpeter, J. K. (1995). Relationalaggression, gender, and social psychological ad-justment. Child Development, 66, 710–722.

DeVries, A. C., Glasper, E. R., & Detillion, C. E.(2003). Social modulation of stress responses.Physiology & Behavior, 79, 399–407.

Dolinski, D., Ciszek, M., Godlewski, K., &Zawadzki, M. (2002). Fear-then-relief, mindless-ness, and cognitive deficits. European Journal ofSocial Psychology, 32, 435–447.

Dutton, D. G., & Painter, S. (1993). The batteredwoman syndrome: Effects of severity and intermit-tency of abuse. American Journal of Orthopsychi-atry, 63, 614–622.

Elliot, A. J., & Devine, P. G. (1994). On the moti-vational level of cognitive dissonance: Dissonanceas psychological discomfort. Journal of Personal-ity and Social Psychology, 67, 382–394.

Ethier, K., & Deaux, K. (1994). Negotiating socialidentity when contexts change: Maintaining iden-tification and responding to threat. Journal of Per-sonality and Social Psychology, 67, 243–251.

Festinger, L. A. (1954). Theory of social comparisonprocesses. Human Relations, 7, 117–140.

Finkel, M. A. (2002). Traumatic injuries caused byhazing practices. American Journal of EmergencyMedicine, 20, 228–233.

Florian, V., Mikulincer, M., & Hirschberger, G.(2002). The anxiety-buffering function of closerelationships: Evidence that relationship commit-ment acts as a terror management mechanism.Journal of Personality and Social Psychology, 82,527–542.

Foderaro, L. W. (2003, September 15). Death inunderground frat’s hazing ritual shakes a SUNYcampus. New York Times, pp. B1–B4.

Galanter, M. (1999). Cults: Faith, healing, and co-ercion (2nd ed). New York: Oxford UniversityPress.

Gerard, H. B., & Mathewson, G. C. (1966). Theeffects of severity of initiation on liking for agroup: A replication. Journal of Experimental So-cial Psychology, 2, 278–287.

Gleick, E. (1997, February 10). Marine blood sports.Time, 30.

Gump, B. B., & Kulik, J. A. (1997). Stress, affilia-tion, and emotional contagion. Journal of Person-ality and Social Psychology, 72, 305–319.

Hautaluoma, J. E., Enge, R. S., Mitchell, T. M., &Rittwager, F. J. (1991). Early socialization into awork group: Severity of initiations revisited. Jour-nal of Social Behavior and Personality, 6, 725–748.

Hazan, C., & Zeifman, D. (1999). Pair bonds asattachments: Evaluating the evidence. In J.Cassidy & P. R. Shaver (Eds.), Handbook of at-tachment: Theory, research, and clinical applica-tions (pp. 336–354). New York: Guilford Press.

Hearst, P. C. (1988). Patty Hearst: Her own story.New York: Mass Market Paperback.

Hermanowicz, J. C., & Morgan, H. P. (1999). Ritu-alizing the routine: Collective identity affirmation.Sociological Forum, 14, 197–214.

Hogg, M. A. (2001). A social identity theory ofleadership. Personality and Social Psychology Re-view, 5, 184–200.

Hoover, N. C., & Pollard, N. J. (2000). Initiation ritesin American high schools: A national survey. Al-fred, NY: Alfred University Press. Retrieved Feb-ruary 6, 2005, from http://www.alfred.edu/news/html/hazing_study.html

Landau, M. J., Solomon, S., Greenberg, J., Cohen,R., Pyszczynski, T., Arndt, J., et al. (2004). Deliverus from evil: The effects of mortality salience andreminders of 9/11 on support for President George

125UNPACKING INITIATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

W. Bush. Personality and Social Psychology Bul-letin, 30, 1136–1150.

Lodewijkx, H. F. M., & Syroit, J. E. M. M. (1997).Severity of initiation revisited: Does severity ofinitiation increase attractiveness in real groups?European Journal of Social Psychology, 27, 275–300.

Lodewijkx, H. F. M., & Syroit, J. E. M. M. (2001).Affiliation during naturalistic severe and mild ini-tiations: Some further evidence against the severi-ty–attraction hypothesis. Current Research in So-cial Psychology, 6, 90–107.

Macfie, N. (2000, September 22). Philippine hostagecrisis takes bizarre twist. Reuters. Retrieved Sep-tember 22, 2000, from http://dailynews.yahoo.com/h/nm/2000922/ts/phillippines_hostages_dc.html

Moreland, R. L., & Levine, J. M. (1989). Newcomersand oldtimers in small groups. In P. Paulus (Ed.),Psychology of group influence (pp. 143–231). Hills-dale, NJ: Erlbaum.

Nelson, D. L., & Quick, J. C. (1991). Social supportand newcomer adjustment in organizations: At-tachment theory at work? Journal of Organiza-tional Behavior, 12, 543–554.

Ostrov, J. M., & Keating, C. F. (2004). Gender dif-ferences in preschool aggression during free playand structured interactions: An observationalstudy. Social Development, 13, 255–277.

Pomerantz, J. J. (1995). All for one and one for all:Do severe initiation procedures increase the influ-ence of a group? Colgate Journal of Sciences, 26,126–135.

Pratto, F., Sidanius, J., Stallworth, L., & Malle, B.(1994). Social dominance orientation: A personal-ity variable predicting social and political atti-tudes. Journal of Personality and Social Psychol-ogy, 67, 741–763.

Rajecki, D. W., Lamb, M. E., & Obmascher, P.(1978). Toward a general theory of infantile at-tachment: A comparative review of aspects of thesocial bond. Behavioral and Brain Sciences, 3,417–464.

Robbins, A. (2004). Pledged: The secret life of so-rorities. New York: Hyperion.

Rosenblatt, P. C. (1964). Origins and effects of groupethnocentrism and nationalism. Conflict Resolu-tion, 8, 131–146.

Schachter, S. (1959). The psychology of affiliation.Stanford, CA: Stanford University.

Schopler, J., & Bateson, N. (1962). A dependenceinterpretation of the effects of a severe initiation.Journal of Personality, 30, 633–649.

Segall, M. H., Berry, J., Dasen, P., & Poortinga, Y.(1999). Human behavior in global perspective.New York: Allyn & Bacon.

Shaver, P. R., Hazan, C., & Bradshaw, D. (1988).Love as attachment: The integration of three be-havioral systems. In R. J. Sternberg & M. L.Barnes (Eds.), The psychology of love (pp. 68–99).New Haven, CT: Yale University Press.

Singer, M. T., & Ofshe, R. (1990). Thought reformprograms and the production of psychiatric casu-alties. Psychiatric Annals, 20, 188–193.

Slingerland, E. (Trans.). (2003). Confucius analects.Indianapolis, IN: Hackett.

Smith, E. R., Coats, S., & Murphy, J. (2001). The selfand attachment to relationship partners and groups:Theoretical parallels and new insights. In C.Sedikides & M. B. Brewer (Eds.), Individual self,relational self, collective self (pp. 109–122). NewYork: Psychology Press.

Smith, E. R., Murphy, J., & Coats, S. (1999). Attach-ment to groups: Theory and measurement. Journalof Personality and Social Psychology, 77, 94–110.

Tajfel, H., & Turner, J. (1979). An integrative theoryof intergroup conflict. In W. G. Austin & S.Worchel (Eds.), The social psychology of inter-group relations (pp. 33–47). Monterey, CA:Brooks/Cole.

Turner, J. C. (1985). Social categorization and theself-concept: A social–cognitive theory of groupbehavior. In E. J. Lawler (Ed.), Advances in groupprocesses: Theory and research (Vol. 2, pp. 77–121). Greenwich, CT: JAI Press.

Uchino, B. N., Cacioppo, J. T., & Kiecolt-Glaser,J. K. (1996). The relationship between social sup-port and physiological processes: A review withemphasis on underlying mechanisms and implica-tions for health. Psychological Bulletin, 119, 488–531.

Van Duuren, F. R., & Di Giacomo, J. P. (1996).Degrading situations and antisocial behavior: Anexperimental approach to delinquency. EuropeanJournal of Social Psychology, 26, 763–776.

Van Duuren, F. R., & Di Giacomo, J. P. (1997).Degrading situations, affiliation, and social depen-dency. European Journal of Social Psychol-ogy, 27, 495–510.

Wegner, D. M., Lane, J. D., & Dimitri, S. (1994). Theallure of secret relationships. Journal of Personal-ity and Social Psychology, 66, 287–300.

West, L. J. (1993). A psychiatric overview of cult-related phenomena. Journal of the American Acad-emy of Psychoanalysis, 21, 1–19.

Winslow, D. (1999). Rights of passage and groupbonding in the Canadian Airborne. Armed Forcesand Society, 25, 429–457.

Received July 29, 2004Revision received February 8, 2005

Accepted February 8, 2005 �

126 KEATING ET AL.

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.