15
Effects of Friendship and Gender on Peer Group Entry Lynne Zarbatany, Michelle Van Brunschot, Ken Meadows, and Susan Pepper University of Western Qntario ZARBATANY, LYNNE; VAN BRUNSCHOT, MICHELLE; MEADOWS, KEN; and PEPPER, SUSAN. Effects of Friendship and Gender on Peer Group Entry. CHILD DEVELOPMENT, 1996,67,2287-2300. Effects of bosts' conflicting motives (to win a game vs. to be a good friend) on peer group entry processes and outcomes were examined. Subjects were 68 triads (35 female) of 10—12-year-old predomi- nantly Wbite cbildren. Two bost friends played a game for a large prize tbat was forfeited for a smaller prize if tbe guest (a friend or nonfriend of botb bosts) was included. Hosts admitted guest friends more often tban nonfriends (44% vs. 26%), suggesting tbat friendsbip norms prescribe self-sacrifice. Hosts bebaved similarly witb guest friends and nonfriends, but guest friends were more active tban nonfriends, reflecting freedom derived from friendsbip security. Female bosts admitted guests more often tban male bosts (51% vs. 21%), consistent witb communal and agentic gender role prescriptions for girls and boys, respectively. Results suggest tbat bosts' friendsbip obligations and psycbological orientation affect tbeir response to a newcomer in a group entry situation. To date, research on group members' (hosts') decision to include a newcomer (guest) in a group entry situation has focused primarily on the quality of the guests' entry bids (e.g.. Dodge, Pettit, McClaskey, & Brown, 1986; Dodge, Schlundt, Schocken, & Delugah, 1983; Putallaz, 1983; Putallaz & Gottman, 1981; Putallaz & Wasserman, 1989). Motivated by well-founded clinical concerns that children who are unable to en- ter igroups will be deprived of opportunities to socialize with their peers, researchers have learned much about guest entry strate- gies that lead to success or failure (see Putal- laz & Wasserman, 1990, for a review). Con- siderably less is known about the hosts' conitribution to the entry process, even though control of the outcome ultimately is in the hands of the hosts, and hosts' behavior may predict entry outcome better than guests' behavior (Borja-Alvarez, Zarbatany, & Pepper, 1991). We hypothesize that hosts' decisions to include or exclude guests may be affected by motives that are independent of the na- ture of the guest's entry bids. For example, we assume that children participate in peer activities for the sake of achieving a variety of goals, such as socializing with others, playing games, winning contests, sharing in- formation or secrets, or simply having fun. To the extent that a newcomer facilitates (or does not interfere with) the fulfillment of these goals, inclusion is likely. When goal achievement is threatened by inclusion of another member, the newcomer may be ex- cluded. We further hypothesize that friendship may motivate hosts' behavior in group entry situations. Because the majority of peer in- teraction occurs with friends (see Hartup, 1989), we would expect that the majority of entry approaches are made by friends. Based on the obligations implicit in friendship, we would expect that guest friends would be more successful than nonfriends because hosts would want to spend time and share activities with their friends (Hartup, 1989, 1993), to behave prosocially toward them, to exhibit loyalty, faithfulness, and support (e.g., Berndt, 1981a; Berndt & Perry, 1986; Bigelow, 1977), and to be responsible for them (Hartup, 1993). On some occasions, the multiple mo- tives affecting the hosts' behavior likely are Tbis study was supported in part by a grant from tbe Vice-President (Researcb) at tbe University of Western Ontario. Tbe autbors tbank Steven Killip at tbe London Board of Educa- tion, and tbe principals, teacbers, and students at Rick Hansen, Tecumseb, Sberwood Forest, and Emily Carr Public Scbools for tbeir cooperation. Address requests for reprints to Lynne Zarbatany, Department of Psycbology, University of Western Ontario, London, Ontario, Canada, N6A 5C2. [Child Development, 1996,67,2287-2300. © 1996 by the Society for Research in Child Development, Inc. All rights reserved. 0009-3920/96/6705-0012$01.00]

Effects of Friendship and Gender on Peer Group Entry

Embed Size (px)

Citation preview

Effects of Friendship and Gender onPeer Group Entry

Lynne Zarbatany, Michelle Van Brunschot,Ken Meadows, and Susan PepperUniversity of Western Qntario

ZARBATANY, LYNNE; VAN BRUNSCHOT, MICHELLE; MEADOWS, KEN; and PEPPER, SUSAN. Effects ofFriendship and Gender on Peer Group Entry. CHILD DEVELOPMENT, 1996,67,2287-2300. Effectsof bosts' conflicting motives (to win a game vs. to be a good friend) on peer group entry processesand outcomes were examined. Subjects were 68 triads (35 female) of 10—12-year-old predomi-nantly Wbite cbildren. Two bost friends played a game for a large prize tbat was forfeited for asmaller prize if tbe guest (a friend or nonfriend of botb bosts) was included. Hosts admitted guestfriends more often tban nonfriends (44% vs. 26%), suggesting tbat friendsbip norms prescribeself-sacrifice. Hosts bebaved similarly witb guest friends and nonfriends, but guest friends weremore active tban nonfriends, reflecting freedom derived from friendsbip security. Female bostsadmitted guests more often tban male bosts (51% vs. 21%), consistent witb communal and agenticgender role prescriptions for girls and boys, respectively. Results suggest tbat bosts' friendsbipobligations and psycbological orientation affect tbeir response to a newcomer in a group entrysituation.

To date, research on group members'(hosts') decision to include a newcomer(guest) in a group entry situation has focusedprimarily on the quality of the guests' entrybids (e.g.. Dodge, Pettit, McClaskey, &Brown, 1986; Dodge, Schlundt, Schocken, &Delugah, 1983; Putallaz, 1983; Putallaz& Gottman, 1981; Putallaz & Wasserman,1989). Motivated by well-founded clinicalconcerns that children who are unable to en-ter igroups will be deprived of opportunitiesto socialize with their peers, researchershave learned much about guest entry strate-gies that lead to success or failure (see Putal-laz & Wasserman, 1990, for a review). Con-siderably less is known about the hosts'conitribution to the entry process, eventhough control of the outcome ultimately isin the hands of the hosts, and hosts' behaviormay predict entry outcome better thanguests' behavior (Borja-Alvarez, Zarbatany,& Pepper, 1991).

We hypothesize that hosts' decisions toinclude or exclude guests may be affectedby motives that are independent of the na-ture of the guest's entry bids. For example,we assume that children participate in peeractivities for the sake of achieving a variety

of goals, such as socializing with others,playing games, winning contests, sharing in-formation or secrets, or simply having fun.To the extent that a newcomer facilitates (ordoes not interfere with) the fulfillment ofthese goals, inclusion is likely. When goalachievement is threatened by inclusion ofanother member, the newcomer may be ex-cluded.

We further hypothesize that friendshipmay motivate hosts' behavior in group entrysituations. Because the majority of peer in-teraction occurs with friends (see Hartup,1989), we would expect that the majority ofentry approaches are made by friends. Basedon the obligations implicit in friendship, wewould expect that guest friends would bemore successful than nonfriends becausehosts would want to spend time and shareactivities with their friends (Hartup, 1989,1993), to behave prosocially toward them, toexhibit loyalty, faithfulness, and support(e.g., Berndt, 1981a; Berndt & Perry, 1986;Bigelow, 1977), and to be responsible forthem (Hartup, 1993).

On some occasions, the multiple mo-tives affecting the hosts' behavior likely are

Tbis study was supported in part by a grant from tbe Vice-President (Researcb) at tbeUniversity of Western Ontario. Tbe autbors tbank Steven Killip at tbe London Board of Educa-tion, and tbe principals, teacbers, and students at Rick Hansen, Tecumseb, Sberwood Forest,and Emily Carr Public Scbools for tbeir cooperation. Address requests for reprints to LynneZarbatany, Department of Psycbology, University of Western Ontario, London, Ontario, Canada,N6A 5C2.

[Child Development, 1996,67,2287-2300. © 1996 by the Society for Research in Child Development, Inc.All rights reserved. 0009-3920/96/6705-0012$01.00]

2288 Child Development

compatible, and entry decisions are madewith relative ease. On other occasions, im-portant motives may confiict with one an-other, creating dilemmas for the hosts. Forexample, in cases where inclusion of a friendwould prevent the group members from ful-filling an important goal such as winning agame, hosts must make difficult choices andcope with the consequences. Ghoices thatfavor their own goal achievement maythreaten the friendship; choices that favorthe friendship may interfere with their goalattainments. The nature of the choices madeand the process by which they are negoti-ated can illuminate the norms of friendship,and, in the present case, the peer group en-try process.

This study was designed to assess theeffects of conflicting motives on peer groupentry processes and outcomes using agroup entry analogue procedure. Specifi-cally, hosts' friendship motives were pittedagainst their desire to win a game and a largeprize. Preadolescents (10—12-year-oId chil-dren) were the focus because friendship isa strong motive among children of this age(Berndt & Perry, 1986; Bigelow, 1977; Sulli-van, 1953) and because investigations ofgroup entry among older children have beenrare (but see Putallaz & Wasserman, 1989).

Sixty-eight friads were formed, eachconsisting of two hosts who were friendswith each other, and one guest who eitherwas or was not a friend of both hosts. Priorto the guest's arrival, the hosts played agame in which they worked as a team for anattractive ("large") prize. To win the largeprize, the hosts needed to work their wayaround the game board in a maximum of 15min. Inclusion of the guest required that thegame (but not the clock) be restarted. Be-cause the game could not be completed ifrestarted, the hosts forfeited the large (butnot a moderately attractive smaller) prize byincluding the newcomer. The choice forhosts was either to include the guest,thereby giving up the large prize but ensur-ing that all three group members receive thesmall prize, or to exclude the guest, keepingthe large prize for themselves and forcingthe guest to accept an unattractive consola-tion prize.

Because the literature on peer group en-try has been silent on the topic of friendship,predictions regarding entry processes andoutcomes were derived from studies com-paring children's behavior with friends andnonfriends in analogous social situations. Of

specific relevance to the prediction of entryoutcome were situations in which childrenhad the opportunity to share scarce re-sources or provide aid to friends or non-friends at a cost to themselves. Findingsfrom this literature have been contradictory.In some cases, no differences have beenobserved in cooperation (Newcomb, Brady,& Hartup, 1979; Philp, 1940), generosity(Berndt, 1981b, girls; Berndt, 1985, Grades4 and 6), or helping (Berndt, 1981b, 1985)between friends and nonfriends. In other in-stances, children have behaved more gener-ously to friends than nonfriends (Berndt,1985, Grade 8; Berndt, Hawkins, & Hoyle,1986, Grade 8; Jones, 1985; Knight &Ghao, 1991; Staub & Sherk, 1970), while instill others, children have been less gener-ous with friends than nonfriends (Berndt,1981b, boys; Berndt et al., 1986, Grade 4;Staub & Noerenberg, 1981, boys).

Berndt's (1986) analysis of the settingconditions that may account for these contra-dictory findings suggests that the presenceor absence of competition in the interper-sonal context may account for the differenteffects. Berndt proposed that when sharingconfiicts with self-interest, children sharemore with friends than nonfriends if equaloutcomes are possible (i.e., noncompetitive),but they will share less with friends thannonfriends if sheiring means losing a contestwith a friend (i.e., competitive). Gompetitivebehavior with friends occurs presumably be-cause friends are an important source of so-cial comparison, and a defeat to a friendwould cause a diminution of self-worth (seeTesser's 1984 self-evaluation maintenancemodel). Because equal outcomes were possi-ble in the present group entry situation (i.e.,hosts could opt to give up the large prize butobtain the small prize for everyone in thegroup by including the guest), we predictedthat hosts would admit guest friends morethan nonfriends, even though they were mo-tivated to win the game and would lose thelarge prize for doing so.

In addition to the differences in entryoutcome associated with friendship, differ-ences in entry processes were expected. Thebehavior of hosts who were faced with aguest friend was expected to refiect a confiictof motives that would be evidenced bygreater deliberation and lengthier entry de-cision times than hosts of nonfriends. Wefurther anticipated that regardless of the en-try decision ultimately reached, hosts wouldconsult with guest friends in the process ofmaking the entry decision more than with

nonfriends because a failure to solicit thefriend's input might be construed as a lack ofrespect and/or caring, both of which wouldviolate expectations of friendship (Sullivan,1953). Finally, hosts were expected to offerjustifications for entry exclusions to friendsmore than nonfriends (Jones, 1985) to as-suage the guest's feelings of resentment andpreserve the friendship.

Differences in guest behavior also wereexpected as a function of friendship. In gen-eral, guest friends were expected to be moreactive in the entry process, initiating behav-ior with the hosts and requesting entry morethan nonfriends. The freedom to engage inthis assertive behavior derives from the se-curity brought by friendship (Hartup, 1983)and expectations for benevolent treatmentfrom friends (e.g., Bemdt, 1981a; Bigelow,1977). Nonfriends were expected to be lessactive, like "newcomers" who cannot antici-pate the nature of their reception or who ex-pect rejection (Dodge et al., 1983; Putallaz& Gottman, 1981).

In the entry situation we employed,guest friends were faced with the same di-lemma that was experienced by hosts; theirchoice was to seek inclusion and the oppor-tunity to gEiin a reasonably good prize forthemselves or to opt out of the game andallow their friends the chance to win thelarge prize. However, the structure of thesituation did not allow for an unambiguousinterpretation of the effects of these con-flicting motives on the guest's behavior. Forexample, in keeping with the notion thatself-sacrifice might be an obligation offriendship, it was possible that guest friendswould deliberately choose not to play thegame more than would nonfriends so thattheir host friends would not lose the largeprize (Bemdt, 1981a; Staub & Noerenberg,1981). Unless the guests made specific state-ments to this effect, though, it was not possi-ble to interpret the meaning of the guests'failure to insist on inclusion (e.g., self-sacrifice vs. resignation). Consequently, weview the unambiguous test of conflictingmotives as being applicable primarily to thehosts' behavior, even though we believe thatbothi hosts and guest faced a similar di-lemma.

Regardless of the friendship status ofthe hosts and guest, gender differences inentry outcomes and processes were ex-pected. Based on past research, male hostswere expected to place a higher priority onwinning (see Lever, 1978; Moely, Skarin, &

Zarbatany et al. 2289

Weil, 1979; Zarbatany & Pepper, in press),as evidenced by (a) higher excitement levels(Zarbatany & Pepper, in press), (b) less con-cern for the needs of the guest, as reflectedin fewer attempts to consult with the guestduring the decision making process, and (c)higher exclusion rates of guests than femalehosts (Borja-Alvarez et al., 1991). These pre-dictions are consistent with gender role the-ory (e.g., Eagly, 1987), which accounts forgender-differentiated behavior in terms ofthe relative strength of two general psycho-logical orientations (Bakan, 1966; Wiggins,1991): agency (i.e., the need to be indepen-dent, masterful, and powerful) and com-munion (i.e., the need to be connected, af-filiative, nurturant). For boys and men,gender role prescriptions for agency are pre-sumed to take precedence over commu-nion, whereas the reverse holds for girlsand women (e.g., Bakan, 1966; Block, 1973;Eagly, 1987; Helgeson, 1994; Huston, 1983).Although we do not assess agentic and com-munal orientations directly in the presentstudy, we view these orientations as usefulheuristics for conceptualizing patterns ofgender-differentiated behavior.

Research on gender differences in theexclusivity of children's friendships (Eder &Hallinan, 1978) raised the additional possi-bility that female hosts would be more con-cerned about safeguarding their friendshiprelationships than would boys. This wouldmean that female hosts would be even morediscriminating in their entry decisions (i.e.,to include guest friends and exclude non-friends) than male hosts. Prediction of an in-teraction of this type, however, required theassumption that for girls, motives to main-tain fidendship boundaries would take pre-cedence over the desire to be kind to anypeer who was dependent on them for accessto resources. Because extant theory and re-search were not sufficiently precise to war-rant such an assumption, we did not pre-dict an interaction. Rather, we looked to ourdata to possibly shed light on the relativestrength of these relationship motives forgirls and boys.

In summary, the primary purpose of thisstudy was to assess the effects of competingmotives (i.e., winning a game vs. being goodto a friend) on hosts' willingness to includea guest in a group entry situation and on theprocess by which this decision was made.Based on the requirements of friendship,hosts were expected to include guest friendsmore than nonfriends and to show greatercaring and consideration for guest friends

2290 Child Development

than nonfriends regardless of the entry de-cision that was made. Guest friends wereexpected to be more active in the entry pro-cess than nonfriends based on the securitybrought by friendship. Further, female hostswere expected to include guests more thanmale hosts and to evidence more concern forthe needs of the guest; male hosts were ex-pected to be more invested in winning thegame, and hence, less likely to allow theguest's needs to interfere with their activity.

MethodParticipants

Participants were 132 fifth-grade (73boys and 59 girls; M age = 10.71 years, SD= 0.33, range = 10.18-12.01) and 111 sixth-grade (49 boys and 62 girls; M age = 11.77years, SD = 0.32, range = 11.19-12.62)children from 15 classes in four publicschools in a medium size Ganadian city. Themajority (97%) of children were White; theremaining children were Asian. Socioeco-nomic status ranged from lower-middle toupper-middle class. The mean participationrate for eligible boys and girls in each gradeacross schools was 64.5% (range = 33%-85%). Parental consent was received for allsubjects.

ProcedureTriad formation.—To form triads meet-

ing the specifications of the experimen-tal design, sociometric assessments weregroup-administered to all participants. Ghil-dren were given lists of same-gender partici-pating grade mates and asked to (a) nomi-nate the three peers they most enjoyedbeing with at school, (b) rate how much theyenjoyed being with each peer at school us-

ing a scale ranging from 1 (mostly don't en-joy being with) to 5 (mostly enjoy beingwith),'̂ and (c) identify an unlimited numberof peers who were their friends. All possiblefriend and nonfriend triads for each groupwere derived based on the friendship nomi-nation data. Friend triads consisted of threechildren who reciprocally nominated eachother as friends,^ with one child arbitrarilydesignated as the guest. Nonfriend triadsconsisted of two reciprocally nominatedfriends, and a third child (the guest) whoneither identified nor was identified by theother two children as a friend. Popular andrejected children, identified using the Asherand Dodge (1986) procedure (standard socialpreference scores greater than or less than+ 1 or - 1, respectively, based on grade materatings and nominations), were disqualifiedfrom the guest position to ensure that hosts'admission decisions were based on friend-ship rather thain sociometric status.

Using these criteria, 68 triads wereformed: 22 male-friend triads, 11 male-nonfriend triads,^ 19 female-friend triadsand 16 female-nonfriend triads. As a checkon the validity of these groupings, separate2 (gender) X 2 (friendship status: friend,nonfriend) analyses of variance (ANOVAs)were conducted on the sum of the two hosts'ratings of the guest, the sum of the guests'ratings of the two hosts, and the sum of thetwo hosts' ratings of each other. As expected,hosts in the friend triads rated their guestssignificantly more positively than hosts inthe nonfriend condition (M = 8.29, SD =1.81, and M = 3.00, SD = 1.57, respec-tively), F(l, 64) = 159.71, p < .001, andguests in the friend condition rated their

' Occasionally, children indicated that they did not know the ratee. In these cases, a scoreof 0 was assigned, following Berndt and Perry (1986). As regards triad composition, there wereeight guests who indicated that they did not know one (n = 4) or both (n = 4) hosts and fourtriads in which one (n = 3) or both (n = 1) hosts indicated they did not know the guest. All ofthese cases occurred in the nonfriend condition.

^ The more stringent criterion (reciprocal nominations and ratings of 4 or 5 on the ratingscale) recommended for friendship identification (Bukowski & Hoza, 1989) could not be usedbecause the sample was not large enough to ensure formation of an adequate number of triads.However, we examined entry outcome as a function of degree of liking between the hosts andguests in the friendship triads on a post hoc basis; these findings are available on request fromthe first author.

•* When the list of all possible triads was generated for each group, the number of friendtriads exceeded the number of nonfriend triads by an average of 7 to 1 (range = .75:1 to 24:1).Thus, the nonfriend cells generally were more difficult to fill than the friend cells. However,the lower number of male than female triads in the nonfriend condition is not reflective ofbase-rate occurrences of these triad types; in fact, male nonfriend triads outnumbered femalenonfriend triads by almost 2 to 1. In the absence of further information regarding idiosyncraticcharacteristics of this cell, we attribute the small size of the male nonfriend condition to chancefactors (i.e., unexpectedly low participation rates of sixth-grade boys at one school and a largenumber of fifth-grade boys at another school who did not meet the inclusion criteria).

hosts significantly more positively thanguests in the nonfriend condition (M = 8.34,SE) = 1.53 and M = 3.92, SD = 2.06, respec-tively), F(l, 64) = 101.76, p < .001. The ma-jority of hosts in the friend (78%) and non-friend (70%) conditions were classmates.The majority of guests in the friend condi-tion (70%) were classmates of both hosts,whereas the majority of guests in the non-friend condition (81%) were not classmatesof either host. The low mean ratings of non-friends and classroom assignment of friendsand nonfriends suggests that nonfriends arelikely best conceived of as strangers or dis-liked peers rather than as neutral acquain-tances. Hosts in the friend and nonfriendconditions rated each other equivalently (M= 8.29, SD = 1.49 and M = 8.33, SD =2.00, respectively), F(l, 64) = 0.06, N.S., al-though female hosts rated each other some-what more positively than male hosts (M =8.69, SD = 1.32 and M = 7.91, SD = 1.96,respectively), F(l, 64) = 4.43, p < .04.

Thirty-nine children (23 boys, 16 girls)were not included in triads that met the de-fining criteria for four reasons: (a) the num-ber of children in their group was not divisi-ble by 3 and thus they were "left over," {b)their friends already had been assigned toother triads, (c) they had no reciprocalfriends, or {d) they were too high or low insocial status (i.e., popular or rejected) toserve as guests. In the interest of fairness,these 39 children were assigned to dyads ortriads and given the opportunity to play thegarae. Their data were used for coder train-ing and were not included in the analyses.

To ascertain whether guests and hostsin the friend and acquaintance conditionswere equivalent in social status, social statuswas estimated based on average ratings ob-tained from same-gender classmates.^ Sepa-rate 2 (gender) X 2 (friendship status: friend,nonfriend) ANOVAs were conducted onguests' social status scores and on the sumof ;the two hosts' social status scores. Forguests, there were no significant effects forfriendship status, F(l, 64) = 2.58, p > .10;thus the social status of guests in the friendand nonfriend conditions was roughlyequivalent For hosts, however, there was amain effect for friendship status, F(l, 64) =6.52, p < .02. Hosts of friends were higherin status than hosts of nonfriends (M = 7.85SD = 0.91 and M = 7.27, SD = 0.99, respec-tively). In view of this status difference.

Zarbatany et al. 2291

hosts' social status was used as a covariatein all of the analyses.

Group entry task.—On the test day, thetwo hosts were brought to a separate roomin their school by a female or male experi-menter who was naive to the children's con-dition. The female experimenter thenclipped on the lapel microphones and in-structed the hosts how to play the game.Hosts were told that the game was played inteams and that the team would play againstthe clock for prizes. Teammates were to taketurns playing, helping each other if neces-sary. Each turn involved first spinning a nee-dle that could land on one of three catego-ries: jobs, names, or animals (see Putallaz& Gottman, 1981). The player then rolled aletter block and said a word fitting the cate-gory and beginning with the letter rolled. Ifthe answer was correct, the player rolled adie and moved the team marker the corre-sponding number of spaces on the board.Three rules were stipulated and written ona large piece of bristol board that was lefl inthe room with the children: (a) if you can'tthink of a word, pass the turn and go on, (b)you can't use the same word twice, and (c)if another person comes in and you want tolet the person play, you have to start overagain.

After it was clear that the hosts under-stood the game rules, the hierarchy of prizeswas explained. Hosts first were shown thesixty-fifth board space, on which the words"small prize" were printed, and told that ifthey reached this space they would earn asmall prize, a brightly colored black leadpencil. Then, the array of "large" prizes(value $1.50 or less) was brought forwardand the hosts asked to select the large prizefor which they wished to play. Large prizesincluded hockey and baseball cards, mar-bles, and a variety of brightly colored or dec-orated pens and pencils. Hosts were toldthat if they reached the last (174th) boardspace before time was up (15 min) they eachwould receive the large prize of their choice.The children were shown the large clockthat would be counting down as they played.

Before tbe experimenter lefl tbe room, sbe pro-vided tbe following additional information: Be-fore you begin, I want to tell you tbat tbere willonly be time to play one game. Also, anyone wbocomes out bere only gets to come bere once. Tbeonly way to get one of tbese prizes is to play tbe

^ Tbe reliability of tbese status scores varies as a function of tbe percentage of eligibleclassmates wbo contributed to tbem (range of number of raters is 2-14).

2292 Child Development

game. People who come in here late and don'tplay will get this (an ordinary black lead pencil).It's OK for someone who comes in late to play,but if anyone new joins the game, you all have tostart over at the beginning. That's just one of therules.

Pilot testing revealed that despite thepresence of the clock, some hosts were un-able to accurately estimate the amount oftime necessary to complete the game and ev-idenced confidence that they could still winthe large prize if they started over. There-fore, the experimenter added that "we'vewatched a bunch of groups play, and I cantell you that if you decide to start over, thechances are pretty good that you won't makeit to the end." This instruction made clearthat the hosts' choice was between the self/team and the newcomer.

Finally, the experimenter reminded thehosts that their game would be videotapedand that access to the videos would be re-stricted to the research team. The experi-menter added that prizes would not be dis-tributed until all of the participants in theschool had a chance to play the game andthe videos could be checked to insure thatplayers had followed the rules. This instruc-tion was included to circumvent the cheat-ing that was observed in pilot episodes andto allow us to give all participants the largeprize of their choice regardless of the out-come of the entry episode. The hosts weretold that the experimenters would be out-side the room while they played and thatthey could come out to ask any questions.The experimenters then set the clock for 15min and left the room.

One experimenter remained outside theroom, watching the children via a televisionmonitor while the other experimenterfetched the guest and brought him or her tothe test room. Prior to sending the guest intothe room, the female experimenter de-scribed the game and the rules and informedthe guest that two children were alreadyplaying. The system of prizes also was ex-plained (including the consolation prize)and the guest was asked to select a largeprize for which to play. The guest was re-minded of the video camera and assured thataccess to the videos would be restricted. Oc-casionally guests asked what they should doif they did not play the game or what theycould do to get the hosts to let them play.In the former instance, guests were asked toremain in the room until the session ended.

In the latter case, guests were told that theyhad to try to "work it out" with the hosts.

The guest entered the room 6 min afterthe hosts had begun playing and remainedthere until the end of the episode. After thegame was over, the experimenters returnedto the room and the female experimenter at-tempted to solicit the hosts' perceptions ofthe entry process by questioning them abouttheir impressions of the game and its rules.Of particular interest were rationales pro-vided by hosts for including or excluding theguest that reflected motives related to win-ning and friendship; these rationales wereexpected to supply additional evidence re-garding the validity of our manipulation.Following this brief interaction with the ex-perimenter, the children were sent back totheir classrooms.

Behavioral and Affective CodingGuests' and hosts' behavior.—Defini-

tions and examples of the guest and host be-haviors coded from the videotapes are pro-vided in Table 1. The behaviors were codedas having occurred or not occurred betweenthe time the guest opened the door of theroom and the time the decision regardinginclusion or exclusion of the guest was made(i.e., the playing piece was returned to thestart position or the hosts resumed play).Guest behavior reflected both entry-relatedand friendship motives and included open-ers (e.g., greetings), requests for entry, vol-untary withdrawals, and deferrals of the de-cision to the hosts. Host behaviors reflectedthe same motives and were coded as havingoccurred if performed by one or both hosts.Host behaviors included openers, consulta-tions with the guest regarding the guest's de-sired outcome, and deliberations regardingthe current situation. Hosts' justifications forexclusion of the guest were coded only afteran exclusion decision was made. In additionto these variables, the time taken (number ofseconds) by hosts to reach an entry decisionfollowing the guest's arrival was noted.

Hosts' excitement.—In addition to thebehavioral variables reported above, hosts'excitement intensity was coded as an indexof the hosts' investment in the game prior tothe guest's arrival. Global assessments of thehosts' excitement intensity were obtainedusing a five-point scale ranging from 1,which signified a slow pace, lack of interestin the game, and distractability, to 5, whichsignified a fast-paced interaction with manyreferences to time, admonishments to part-

Zarbatany et al. 2293

TABLE 1

DEFINITIONS AND EXAMPLES OF BEHAVIOR CODES AND RELIABILITIES (Kappas)

Code Definition

Cuest or bost codes:Openers

Cuest codes:Entry requestsWithdrawal

Decision deferral

Host codes:Consultations

Deliberations

Justifications

Verbal or nonverbal remarks made by tbe bosts or guest upon tbe guest'sarrival including greetings ("Hi"), comments ("You guys are far"; "Obno"), or questions ("Wbat are you guys doing?" "What are you doingbere?"). (.78)

Cuests' requests for participation in tbe game ("Can I play?"). (1.00)Decisions made by guests not to play tbe game ("It's OK. Don't restart").

(1.00)Cuests' invitations to permit bosts to decide tbe outcome ("It's up to you

guys"). (1.00)

Hosts' direct or indirect appeal to tbe guests to permit exclusion of tbeguest, or bosts' solicitation of guest's preferred outcome (e.g., "Wbat doyou want to do?"). (1.00)

Hosts' consideration of factors tbat migbt affect an entry decision, sucb asrules ("We bave to start at tbe beginning") time or progress ("We baveonly 8 min left"), or fairness ("It's only fair tbat sbe gets to play"). (1.00)

Unsolicited apologies or explanations provided by bosts following entry ex-clusions ("We were so far"). (.91)

ners to hurry, and a sense of urgency. Sepa-rate excitement ratings were obtained foreach host in the 2 min preceding the guest'sarrival; prior to analysis, the hosts' excite-m€!nt scores were averaged across the twomembers of the dyad.

Reliability assessments.—The video-tapes were coded in two passes by twotrained coders who were naive to the friend-ship status of the triad. Behavioral codingwas undertaken in the first pass and excite-ment ratings in the second pass. Reliabilityof the behavioral coding was ascertainedbased on the two coders' independent cod-ing of 26% of the tapes. Overall, kappa was.95 (range = .78—1.00); individual kappas foreach behavior code are presented in Table1. Excitement ratings were made by bothcoders for all of the videotapes. Reliabilityof the ratings, averaged across hosts withina dyad and calculated as a correlation was.99.

ResultsOverview

To present the results, we first assessentry outcome as a function of gender andfriendship status of the triads.^ Next, we ex-amine the group entry process by assessinggender and friendship status differences inthe guests' behavior, the hosts' excitementprior to the guest's arrival, the hosts' behav-ior after the guest's arrival, the time takenby hosts to make the entry decision, and thehosts' justifications for exclusion of theguest. Low frequency behavioral variables(i.e., those for which all four cell frequencieswere less than 5) were identified and omit-ted from these analyses.® In the last section,to ascertain hosts' perceptions of the entryprocess, hosts' rationales for their entry deci-sions are examined.

In the following analyses, several de-pendent variables (i.e., outcome, behavioral

' All analyses were performed witb grade included as a factor. Because only one significanteffect involving grade was obtained (tbe gender x grade interaction for bost openers, F(l, 59)= 5.94, p < .02, witb flftb-grade boys evidencing tbis bebavior more tban other cbildren), andno grade differences were expected, analyses are presented excluding grade assessments.

' Two bebavioral variables met tbis criterion. Only four guests (two male friends and twofemale nonfriends) voluntarily witbdrew from tbe game and only seven guests (tbree femalefriends, tbree female nonfriends, one male nonfriend) explicitly deferred tbe entry decision totbe bosts.

2294 Child Development

variables) were dichotomous rather thancontinuous. Means pertaining to these vari-ables refer to the proportion of the samplein which the variable was observed. Ac-cording to Tabachnick and Fidell (1989),ANOVA can be performed on dichotomousdata as long as the two categories of the de-pendent variable occur fairly evenly (i.e.,with no greater than a 25%-75% split). Thiscriterion was met in every instance in whichfactorial analyses were performed.

In each factorial analysis reported be-low, gender of hosts and friendship status ofguest to host (friend, nonfriend) served asthe independent variables, and the hosts' so-cial status (i.e., average liking rating) was en-tered as a covariate.^ Gomplete least squaresanalyses were performed due to the unequalnumber of subjects in the four cells (Overall& Spiegel, 1969, Method 1).

Entry Outcome as a Function ofFriendship Status and Gender

The proportion of male and femalefriends and nonfriends who experienced en-try success is presented in Table 2. The anal-ysis of covariance (ANGOVA) produced amain effect for gender, F(l, 63) = 9.74, p <.003, and a main effect for friendship status,F(l, 63) = 4.87, p < .04. As expected, femaleguests were more likely to be admitted tothe game than male guests, and friends weremore likely to gain entry than nonfriends.

Entry Process as a Function of FriendshipStatus and Gender

Guests' behavior.—Means and standarddeviations for guests' and hosts' pre-er^frydecision behavior and hosts' excitement arepresented in Table 3. Guests' openers andguests' requests for entry were included ina 2 (gender) x 2 (friendship status) multivar-iate analysis of covariance (MANGOVA).Only the multivariate main effect for friend-ship status was significant, Wilks's lambdaF(2, 62) = 6.50, p < .003. The univariatemain effect for friendship status was signifi-cant only for guests' entry requests, F(l, 63)= 11.05, p < .002. Guests who were friendsof the hosts requested entry more often thanguests who were not friends of the hosts.

Hosts' excitement.—The ANGOVA per-formed on the averaged ratings of the twohosts' excitement levels prior to the guest'sarrival produced a significant main effect forgender, F(l, 63) = 5.41, p < .03. Male hosts

TABLE 2

PROPORTION (and Standard Deviations) OF GUESTFRIENDS AND NONFKIENDS ACHIEVING ENTRY

GONDITIONGENDER OF

HOSTS Friend Nonfriend TOTAL

Male 32 .00 .21(.48) (.00) (.42)

Female 58 .44 .51(.51) (.51) (.51)

All 44 .26 .37(.50) (.45) (.49)

NOTE.—Proportions are calculated based on columnfrequencies within gender groupings.

were more excited than female hosts priorto the guest's arrival (see Table 3).

Hosts' behavior.—Hosts' openers, hosts'consultations with the guest, and hosts' de-liberations were entered into a 2 (gender) x2 (friendship status) MANGOVA. There wasa multivariate main effect for gender, F(3,61) = 4.48, p < .007, and a multivariate in-teraction between gender and friendshipstatus, F(3, 61) = 3.83, p < .02. Only theunivariate gender main effect for hosts' con-sultations with the guest was significant, F(l,63) = 8.11, p < .006. Female hosts weremore likely to consult with the guest aboutthe entry decision than were male hosts. Theunivariate interaction between gender andfriendship status was significant for hosts'openers, F(l, 63) = 7.97, p < .007. Malehosts were more likely to offer openings tononfriends than friends, F(l, 63) = 7.06, p< .02, whereas female hosts did not dis-criminate between friends and nonfriends intheir openings (see Table 3). The majorityof host openers took the form of greetings.

Entry decision time.—The ANGOVA onamount of time (number of seconds) takenby hosts to reach a decision regarding inclu-sion of the guest produced a marginally sig-nificant main effect for friendship status, F(l,63) = 3.89, p < .06. Decisions were reachedsomewhat more rapidly when the guest wasa nonfriend (M = 11.37, SD = 11.75) thanwhen the guest was a friend (M — 20.78, SD= 19.92).

Hosts' justifications for exclusion of theguest.—Because the hosts' postdecision jus-tifications could occur only when the guest

' All analyses were repeated using guest status as the covariate, and the findings werevirtually identical in every case.

Zarbatany et al. 2295

TABLE 3

MEANS (and Standard Deviations) FOR GUESTS' AND HOSTS' BEHAVIORAL VARIABLES

MALE

VARIABLE BOYS GIRLS NF NF

FEMALE

F NF

Guest openers 39 .23 .39 .19 .50 .18 .26 .19(.50) (.43) (.49) (.40) (.51) (.40) (.45) (.40)

Cuest entry requests 45 .49 .61 .26" .55 .27 .68 .25(.51) (.51) (.49) (.45) (.51) (.47) (.48) (.45)

Host excitement 3.06 2.64'> 2.76 2.96 2.98 3.22 2.53 2.78(.85) (.59) (.75) (.76) (.84) (.90) (.56) (.60)

Host openers 52 .37 .41 .48 .36 .82 .47 .25"=(.51) (.49) (.50) (.51) (.49) (.40) (.51) (.45)

Host consultations 06 .34'' .17 .26 .05 .09 .32 .38(.24) (.48) (.38) (.45) (.21) (.30) (.48) (.50)

Host deliberations 27 .34 .37 .22 .36 .09 .37 .31(.45) (.48) (.49) (.42) (.49) (.30) (.50) (.48)

n 33 35 41 27 22 11 19 16

NOTE.—F = friend, NF = nonfriend. Host excitement is rated on a five-point scale; all other variables are codeddichotomously.

* Significant friendship status difference.^ Significant gender difference.^ Significant gender X friendship status interaction.

was excluded, only the latter triads were in-cluded in the ANCOVA performed on thisvariable. There were no significant effects,indicating that justifications were no morelikely offered to friends (M = .52, SD = .51)than to nonfriends (M = .30, SD = .47) whowere excluded, F(l, 38) = 1.01, N.S.

Hosts' Rationales for Their Entry DecisionReasons provided by hosts to the experi-

menter for tlieir entry decisions were as-sessed informally. In all 43 cases in whichthe guest was excluded, hosts referred totheir progress (e.g., "We were already sofar") or the prizes (e.g., "We wanted theprize"). For guest friends who were in-cluded, rationedes referred to friendship (n= 10), fairness (n = 2), or to the notion thatthe team's performance could be improvedby the addition of the guest (n = 3). Thelatter rationale was cited as the reason forincluding guest nonfriends (n = 3) alongwiith fairness (n = 3) and friendship (n = 1).Three sets of hosts failed to produce codableresponses.

DiscussionThe present findings affirm that friend-

ship motivates group entry decisions. In asituation where hosts were required to

choose between winning a game and a largeprize for themselves and providing a peerwith the opportunity to win a game and aprize at a cost to themselves, self-sacrificialbehavior was more likely to occur if the peerwas a friend than a nonfriend as Berndt(1986) would have predicted. Although wehad anticipated that hosts' decisions regard-ing the inclusion of friends would be moreconflict-ridden than decisions regardingnonfriends, there was no evidence that thiswas the case based on the occurrence ofhosts' deliberations^ or justifications for ex-clusion. Entry decisions involving friendstook somewhat longer than those involvingnonfriends, as we expected, but presumablythis was due to the time taken to respond tothe more active behavior (i.e., entry re-quests) of guest friends. In short, the behav-ior engaged in by hosts when faced with afriend or nonfriend did not differ markedly,even though the outcomes were notably dif-ferent.

It is conceivable that friendship-relateddifferences in the hosts' behavior did notemerge partly as a function of the group en-try situation employed. In many instances,hosts committed themselves to an entry de-cision (i.e., exclusion) prior to the guest's ar-rival when the rule involving inclusion of

* In an additional cbeck of tbis point, we performed a gender x friendsbip status ANCOVAon tbe duration (number of seconds) of deliberations beld by bosts wbo deliberated. Tbere wereno significant effects.

2296 Child Development

newcomers first was stipulated.® For thesedyads, further interaction was not required,even when the guest was a friend. Moreover,the time constraints imposed on the gamemay have suppressed discussion amonghosts. Further research is necessary to ascer-tain whether the expected effects of friend-ship on the group entry process are evidentunder circumstances where hosts do not an-ticipate the possibility of a new arrival, orwhen time is unlimited.

The greater generosity of hosts towardfriends than nonfriends suggests that normsof preadolescent friendship may prescribeself-sacrifice for friends (e.g., Berndt &Perry, 1986; Bigelow, 1977). This interpreta-tion is bolstered by two forms of evidence.First, and most obviously, hosts often citedfriendship as their reason for including theguest. Second, guest friends were morelikely to request entry than nonfriends. Be-cause entry requests are direct, refusalsplace children at risk for experiencing lossof face (Putallaz & Wasserman, 1990). Thisrisk may be minimized when hosts arefriends, due to the security offered by friend-ship (Hartup, 1983) and implicit obligationsfor benevolent behavior toward friends.Friends who requested entry likely ex-pected that their friends would includethem. In contrast, nonfriends (i.e., strangers,disliked peers) took fewer risks, likely be-cause the outcomes were less predictable ormore certain to be negative.

The notion that self-sacrifice is man-dated by friendship might appear to be miti-gated by our finding that many host dyads(56%) excluded their guest friends from thegame. We believe that such a conclusionwould be premature, however, in the ab-sence of further information about theguest's motives in the group entry situation.Although guest friends rarely withdrew fromthe game voluntarily, it is possible that theyviewed their exclusion by hosts as under-standable due to the hosts' proprietary rightsover the game (Morgan & Sawyer, 1967;Staub & Noerenberg, 1981) and were con-tent to witness their friends' efforts to winthe prize. This might reflect the precedenceof ownership over friendship norms, but italso could reflect the generosity mandatedby friendship. That is, friendship may re-quire a willingness to accept and support theself-strivings of friends even when to do so

exacts a cost to oneself. Based on the princi-ple of reciprocity that underpins friendship(e.g., Hartup, 1993), these sacrifices likelyare reciprocated by friends across time. Inthe future, more extensive interviews withhosts and guests in which rationales for andreactions to entry decisions are solicitedmight help to ascertain the merits of theseinterpretations and determine more defini-tively the role of self-sacrifice (on the part ofguests, hosts, or both) in friendship.

The conception of friendship obligationemployed in this study has assumed that thehosts' friendship motive would be expressedthrough the act of including the guest in thegame and sacrificing the large prize. It mightbe argued, however, that a decision to ex-clude the guest also may have reflectedfriendship. In the latter case, the friendshipact would derive from the hosts' friendshipwith each other rather than from their friend-ship with the guest. Technically, it is impos-sible to discern whether an exclusion deci-sion refiected a desire to be a good friend toone's cohost, a desire to win the game, orboth, using the methodology we employed.From a theoretical standpoint, however, itwould be difficult to explain our findingsbased on the interpretation that hosts' exclu-sion decisions were motivated by friend-ship. Such an explanation would have to ac-count for gender differences in exclusion(why would female hosts be less committedto each other than male hosts?) as well asdifferences in exclusion that were due to thehosts' relationship with the guest (whywould hosts of guest friends be less commit-ted to each other than hosts of nonfriends?).In view of these difficulties, we are satisfiedwith our original conceptualization of thehosts' motives. We do believe, however, thatadditional systematic research on the rela-tive effects of varying host motives (exclud-ing motives pertaining to the guest) wouldbe useful.

Gender influenced entry outcomes asexpected based on past research (e.g., Borja-Alvarez et al., 1991). Female hosts weremore likely to include the guest than malehosts. These findings are consistent withgender role prescriptions for agency andcommunion for boys and girls, respectively(e.g., Eagly, 1987). Both male and femalehosts' preferential treatment of guest friendscan be viewed as expressing communal mo-

^ Apparently, pacts among hosts regarding treatment of potential newcomers are not uncom-mon; Gorsaro (1979) repeatedly witnessed formation of these agreements among preschool-agegroup members.

Zarbatany et al. 2297

tivation. However, male hosts' higher exclu-sion rates coupled with their greater excite-ment levels relative to female hosts prior tothe guest's arrival suggest that for boys, the(agentic) motive to win the game out-weighed the (communal) motive to attend tothe guest's needs.^° There is no indicationthat the boys' greater investment in thegame was attributable to the greater induce-ment value of the prizes selected by boysthan girls because all prizes were of approxi-msitely equal monetary value. Indeed, prizevalue likely had little to do with boys'achievement strivings because similar gen-der differences in entry outcome have beenobserved when no prizes have been offered(Borja-Alvarez et al., 1991). If we are correctthat winning the game was the primary fac-tor motivating male hosts, we would predictthat gender differences in entry outcomeswould be less pronounced or absent whenthe hosts' activity is not mastery oriented.This prediction is supported by Putallaz andGottman (1981), who found no gender differ-ences in entry outcome when hosts were en-gaged in a rionmastery-oriented task (i.e.,hosts were not told to play to win).

In contrast to male hosts, female hostswere more likely to give up their quest forthe large prize when faced with a child whowas dependent on them for access to theprizes. We believe that this reflected com-munal motives to fulfill interpersonal obliga-tions (Eagly, 1987; Leaper, 1994). Femalehosts' interpersonal concerns were evidentin their higher guest admission rates andtheir greater efforts to consult with the guestregarding the entry decision. For girls, gen-der role prescriptions may compel con-siderate behavior to peers who are in adependency relationship with them, andtransgressions may result in feelings of guilt(Williams & Bybee, 1994).

Though we interpret our findings as re-flective of gender role prescriptions foragency and communion, the viability of thisexplanation awaits direct assessment ofthese psychological orientations. The tech-nology for such assessment in child popula-tions has yet to be developed (although seeMcAdams & Losoff, 1984, for work on re-lated constructs); however, a promising ap-proach, paralleling that employed withadults (see Helgeson, 1994; Spence, 1984),

might be to administer selected items of apublished gender-role inventory for chil-dren (e.g., Boldizar, 1991). We believe thatsuch an effort is warranted in view of thepotential for constructs of agency and com-munion to account for pervasive patterns ofgender-differentiated interpersonal behav-ior observed among children (e.g., see Beal,1994; Belle, 1989; Leaper, 1994; Thome,1994 for reviews), including those observedin the present study. Moreover, assessmentof the strength of agentic and communal mo-tives may provide a mechanism for under-standing individual differences in behaviorwithin gender. For example, in this study,many (49%) female hosts excluded theirguests, despite evidence that they were sen-sitive to the guests' needs (e.g., consulta-tions with the guest). One possible explana-tion for this finding is that exclusionaryfemale hosts had stronger agentic motivesthan those who included their guests, a hy-pothesis that would be simple to test if ade-quate assessments of agentic and communalorientations were available. Further, the rel-ative strength of agentic and communal mo-tives may help to differentiate individualswho attempt to mitigate the negative effectsof their intrapersonal strivings by employingstrategies such as consultations with theguest that maintain connection (high agency,high communion) from those who do not(high agency, low communion; see Leaper,1994). In sum, we believe that knowledgeof children's motivational orientations willhelp to further our understanding of chil-dren's interpersonal behavior, especially insocial contexts that evoke high levels ofagentic and/or communal behavior. We fur-ther believe that psychological orientationwill better predict children's behavior thantheir gender (Zarbatany & Pepper, in press).

When interpreting the friendship- andgender-related findings reported herein, it isimportant to recall that the number of maletriads in the nonfriend condition was rela-tively small. As a result, we may not haveestimated the variability in this cell suffi-ciently well to draw strong conclusions re-garding the behavioral tendencies of malenonfriends. Although we believe that thelow number of male nonfriend triads re-sulted from chance factors, replicationclearly would be helpful to confirm findings

1" We are mindful of tbe possibility tbat for cbildren wbo are primarily agentically oriented,mastery motives may routinely take precedence over friends' needs, if cbildren accept tbisbierarcbical ordering of values as normative, cboices favoring self-interest over friendsbip maybe accepted witb minimum fuss by friends and not pose a tbreSt to tbeir relationsbips.

2298 Child Development

pertaining to this cell, including our majorfindings regarding friendship and gender.Moreover, additional research is required toadequately test the interaction that was an-ticipated based on research on gender differ-ences in friendship exclusivity (Eder & Hal-linan, 1978). Though we neither predictednor found this interaction, power was low inthis study, and the question of how girls andboys resolve conflicting communal motives(e.g., to maintain friendship boundaries vs.to be kind to a needy child) remains interest-ing and unresolved.

For those who choose to undertake re-search of the type we report here, we drawattention to two sampling-related problems.First, relatively large numbers of childrenare needed to provide an adequate sampleof nonfriend triads in view of their relativescarcity. Second, substantial reductions insample size can be expected in the processof unconfounding status and friendship. Weminimized the latter problem by controllingonly the status of the guest, based on thereasoning that it was particularly importantto show that guests gained entry due to theirspecific affective ties with the hosts ratherthan to their overall popularity among class-mates. Had we also controlled status of hosts(i.e., excluded popular and rejected childrenfrom the host position), we would have losta number of additional triads. As a conse-quence of not controlling hosts' status, weobserved a significant difference betweenhosts of friends and hosts of nonfriends inlevels of peer acceptance, which fortunatelydid not affect the major findings. Neverthe-less, because friendship and status are inevi-tably related through their mutual affectiveunderpinnings (i.e., liking), threats to the va-lidity of conclusions pertaining to the effectsof friendship remain a constant concern, andefforts to control status should be madewhenever possible.

The fate of nonfriends was notable inthis study;̂ ^ only 26% of these childrenachieved entry success as compared to 44%of friends. Nonfriends were no more unpop-ular with their classmates overall thanfriends, and had friends among their grademates; however, they were strangers to ordisliked by their hosts. Their outcomes giveus additional perspective on the challengesfaced in everyday life by friendless children

and perhaps also by children who are dis-liked by their peers (see Dodge et al., 1983,1986; Putallaz & Gottman, 1981). If friendshave difficulty penetrating groups when thehosts' self-interest motivates exclusion, gain-ing access to such groups may represent aneven more daunting task for children whohave no friends, regardless of the skill-fulness of their entry bids. The solution forthese children may be to attempt to joingroups in which it is in the self-interest ofthe group members to include additionalmembers.

The findings of this study again makesalient the power of the hosts to affect groupentry outcomes (see Boija-Alvarez et al.,1991). Admittedly, the hosts' control of thesituation was intensified by the prize-winning incentives; these incentives werenecessary to draw out the decision-makingprocess that was our primary interest. Thus,our findings might not be comparable in allrespects to those obtained in previous re-search in which the guest's strategic behav-ior was the main focus. Indeed, the entrybehavior of guests in this study was suffi-ciently constrained so that conventional mo-lecular coding schemes for guests' behaviorwere not applicable (e.g., Boija-Alvarez etal., 1991; Dodge et al., 1983, 1986; Putallaz& Gottman, 1981). Even if the present entrytask invested hosts with a high level of con-trol, however, the nature of group entry dic-tates that hosts always control the outcometo some degree. We have demonstrated thathosts' outcome decisions are affected by nu-merous factors, including the nature of theirrelationships with the guests, the behavioraldispositions prescribed by their genderroles, and the psychological conditions (i.e.,excitement) attending their interaction priorto the guest's arrival.

RefereneesAsher, S. R., & Dodge, K. A. (1986). The identifi-

cation of socially rejected children. Develop-mental Psychology, 22, 444-449.

Bakan, D. (1966). The duality of human existence:An essay on psychology and religion. Ghi-cago: Rand McNally.

Beal, G. R. (1994). Boys and girls: The develop-ment of gender roles. New York: McGraw-Hill.

Belle, D. (1989). Gender differences in children's

^^ We are unsure how to explain the high rate of greetings (openers) offered by male hoststo guest nonfriends relative to guest friends. Although this could reflect a combination of socialobligation to be polite to an unknown/disliked visitor and freedom from social constraint offeredby friendship, we cannot confirm this interpretation in the absence of further research.

Zarbatany et al. 2299

social networks and supports. In D. Belle(Ed.), Children's social networks and socialsupports (pp. 173-188). New York: Wiley.

Berndt, T. J. (1981a). Age cbanges and cbangesover time in prosocial intentions and bebaviorbetween friends. Developmental Psychology,17, 408-416.

Bemdt, T. J. (1981b). Effects of friendsbip on pro-social intentions and bebavior. Child Devel-opment, 52, 636-643.

Beimdt, T. J. (1985). Prosocial bebavior betweenfriends in middle cbildbood and early adoles-cence. Journal of Early Adolescence, 5,307-317.

Bemdt, T. J. (1986). Sbaring between friends:Contexts and consequences. In E. C. Mueller& C. R. Cooper (Eds.), Process and outcomein peer relationships (pp. 105-127). Orlando,FL: Academic Press.

Berndt, T. J., Hawkins, I. A., & Hoyle, S. C.(1986). Cbanges in friendsbip during a scboolyear: Effects on cbildren's and adolescents'impressions of friendsbip and sbaring witbfriends. Child Development, 57, 1284-1297.

Bemdt, T. J., & Perry, T. B. (1986). Cbildren'sperceptions of friendsbips as supportive rela-tionsbips. Developmental Psychology, 22,640-648.

Bigelow, B. J. (1977). Cbildren's friendsbip expec-tations: A cognitive developmental study.Child Development, 48, 246-253.

Block, I. H. (1973). Conceptions of sex role: Somecross-cultural and longitudinal perspectives.American Psychologist, 28, 512-526.

Boldizar, I. P. (1991). Assessing sex typing andandrogyny in cbildren: Tbe Cbildren's SexRole Inventory. Developmental Psychology,27, 505-515.

Borja-Alvarez, T., Zarbatany, L., & Pepper, S.(1991). Contributions of male and femaleguests and bosts to peer group entry. ChildDevelopment, 62, 1079-1090.

Bukowski, W. M., & Hoza, B. (1989). Popularityand friendsbip: Issues in tbeory, measure-ment, and outcome. In T. J. Berndt & G. W.Ladd (Eds.), Peer relationships in child devel-opment (pp. 15-45). New York: Wiley.

Corsaro, W. A. (1979). "We're friends, rigbt?":Cbildren's use of access rituals in a nurseryscbool. Language in Society, 8, 315-336.

Dodge, K. A., Pettit, G. S., McClaskey, C. L., &Brown, M. M. (1986). Social competence incbildren. Monographs of the Society for Re-search in Child Development, 51(2, Serial No.213).

Dodge, K. A., Scblundt, D. C , Scbocken, I., &Delugab, J. D. (1983). Social competence andcbildren's sociometric status: Tbe role of peergroup strategies. Merrill-Palmer Quarterly,29, 309-336.

Eagly, A. H. (1987). Sex differences in social be-havior: A social role interpretation. Hills-dale, NJ: Erlbaum.

Eder, D., & Hallinan, M. T. (1978). Sex differ-ences in cbildren's friendsbips. American So-ciological Review, 43, 237-250.

Hartup, W. W. (1983). Peer relations. In E. M.Hetberington (Ed.), P. H. Mussen (Series Ed.),Handbook of child psychology: Vol. 4. Social-ization, personality, and social development(4tb ed., pp. 103-196). New York: Wiley.

Hartup, W. W. (1989). Bebavioral manifestationsof cbildren's friendsbips. In T. J. Bemdt &G. W. Ladd (Eds.), Peer relationships in childdevelopment (pp. 46-70). New York: Wiley.

Hartup, W. W. (1993). Adolescents and tbeirfriends. In B. Laursen (Ed.), Close friend-ships during adolescence: New directions forchild development (pp. 3—22). San Francisco:Jossey-Bass.

Helgeson, V. S. (1994). Relation of agency andcommunion to well-being: Evidence and po-tential explanations. Psychological Bulletin,116, 412-428.

Huston, A. (1983). Sex typing. In E. M. Hetbering-ton (Ed.), P. H. Mussen (Series Ed.), Hand-book of child psychology: Vol. 4. Socializa-tion, personality, and social development(4tb ed., pp. 387-467). New York: Wiley.

Jones, D. C. (1985). Persuasive appeals and re-sponses to appeals among friends and ac-quaintances. Child Development, 56, 757-763.

Knigbt, C. P., & Cbao, C.-C. (1991). Cooperative,competitive, and individualistic social valuesamong 8- to 12-year-oId siblings, friends, andacquaintances. Personality and Social Psy-chology Bulletin, 17, 201-211.

Leaper, C. (1994). Exploring tbe consequences ofgender segregation on social development. InC. Leaper (Ed.), New directions for child de-velopment: The development of gender andrelationships (pp. 67—86). San Francisco:Jossey-Bass.

Lever, J. (1978). Sex differences in tbe complexityof cbildren's play and games. American So-ciological Review, 43, 471-483.

McAdams, D. P., & LosofF, M. (1984). Friendsbipmotivation in fourtb and sixth graders: A tbe-matic analysis. JoMmo! of Personal and SocialRelationships, 1, 11-27.

Moely, B. E., Skarin, K., & Weil, S. (1979). Sexdifferences in competition-cooperation be-bavior of cbildren at 2 age levels. Sex Roles,5, 329-342.

Morgan, W. R., & Sawyer, J. (1967). Bargaining,expectations, and tbe preference for equalityover equity. JoumaZ of Personality and SocialPsychology, 6, 139-149.

Newcomb, A. F., Brady, J. E., & Hartup, W. W.

2300 Child Development

(1979). Friendship and incentive condition asdeterminants of children's task-oriented so-cial behavior. Child Development, 50, 878-881.

Overall, J. E., & Spiegel, D. K. (1969). Goncemingleast squares analysis of experimental data.Psychological Bulletin, 72, 311-322.

Philp, A. J. (1940). Strangers and friends as com-petitors and co-operators. Journal of GeneticPsychology, 57, 249-258.

Putallaz, M. (1983). Predicting children's socio-metric status from their behavior. Child De-velopment, 54, 1417-1426.

Putallaz, M., & Gottman, J. M. (1981). An interac-tional model of children's entry into peergroups. Child Development, 52, 986-994.

Putallaz, M., & Wasserman, A. (1989). Ghildren'snaturalistic entry behavior and sociometricstatus: A developmental perspective. Devel-opmental Psychology, 25, 297-305.

Putallaz, M., & Wasserman, A. (1990). Ghildren'sentry behavior. In S. R. Asher & J. D. Goie(Eds.), Peer rejection in childhood (pp. 60—89). New York: Gambridge University Press.

Spence, J. T. (1984). Masculinity, femininity, andgender-related traits: A conceptual analysisand critique of current research. In B. A.Maher & W. B. Maher (Eds.), Progress in ex-perimental personality research (Vol. 13, pp.1-97). San Diego, GA: Academic Press.

Staub, E., & Noerenberg, H. (1981). Propertyrights, deservingness, reciprocity, friendship:The transactional character of children's shar-

ing behavior. Journal of Personality and So-cial Psychology, 40, 271-289.

Staub, E., & Sherk, L. (1970). Need for approval,children's sharing behavior, and reciprocityin sharing. Child Development, 41, 243—252.

Sullivan, H. S. (1953). The interpersonal theoryof psychiatry. New York: Norton.

Tabachnick, B. G., & Fidell, L. S. (1989). Usingmultivariate statistics (2d ed.). New York:Harper & Row.

Tesser, A. (1984). Self-evaluation maintenanceprocesses: Implications for relationships andfor development. In J. G. Masters & K. Yarkin-Levin (Eds.), Boundary areas in social anddevelopmental psychology (pp. 271-299).New York: Academic Press.

Thome, B. (1994). Gender play: Girls and boys inschool. New Brunswick, NJ: Rutgers Univer-sity Press.

Wiggins, J. S. (1991). Agency and communion asconceptual coordinates for the understandingand measurement of interpersonal behavior.In W. M. Grove & D. Gicchetti (Eds.), Think-ing clearly about psychology: Essays inhonor of Paul Meehl (Vol. 2, pp. 89-113).Minneapolis: University of Minnesota Press.

Williams, G., & Bybee, J. (1994). What do childrenfeel guilty about? Developmental and genderdifferences. Developmental Psychology, 30,617-623.

Zarbatany, L., & Pepper, S. (in press). The role ofthe group in peer group entry. Social Devel-opment.