Sie sind auf Seite 1von 9

Reciprocity of Interpersonal Attraction: A Nonconfirmation of a Plausible Hypothesis

Author(s): Theodore M. Newcomb


Source: Social Psychology Quarterly, Vol. 42, No. 4 (Dec., 1979), pp. 299-306
Published by: American Sociological Association
Stable URL: http://www.jstor.org/stable/3033801 .
Accessed: 22/06/2014 22:09

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to
Social Psychology Quarterly.

http://www.jstor.org

This content downloaded from 62.122.79.56 on Sun, 22 Jun 2014 22:09:25 PM


All use subject to JSTOR Terms and Conditions
Social PsychologyQuarterly
1979, Vol. 42, No. 4, 299-306

ofInterpersonalAttraction:
Reciprocity ofa
A Nonconfilrmation
PlausibleHypothesis
THEODORE M. NEWCOMB
The University
of Michigan

The hypothesiswhose nonconfirmation is here reportedis that withcontinuingacquaintance,


dyads withina group willincreasinglyexchange similacr levels of interpersonalattraction.The
followingcircumstancesacreconsideredas possible sources of the nonconfirmation: the use of
rank-ordering as a measure of each person's attractionto each of the othergroupmembers,as
creatingartifactualeffects;the use of ranks at all levels, instead of restrictingthemto high
levels of attraction:thefact thatestimnatesof others' attractionto oneselfdid notbecome more
accurate over time;and the development,over time,of consensual norms about the attraction
status of all group members. The last of these is shown to be the most probable single
explanation.Conditionsunder whichthe hypothesisprobablycomm be supportedatrediscussed.

It would seem to be a common-sense principlesof positive and negative rein-


propositionthatgroupmemberswho have forcement:attractionhas been viewed in
continuingopportunitiesto become ac- terms of reward value attributedto an-
quaintedwill develop varyinglevels of at- other person (Newcomb, 1960; Byrne,
tractiontowardone another,and thatover 1971). Or one can draw upon consistency
timethe generaltrendwill be forpairs of theories; that of Heider (1958), referring
individualsto accord to each othersimilar only to perceived,not to 'objective" bal-
levels of preferenceat any level of attrac- ance, is perhaps most directlyrelevant,
tion.The presumedrationaleforthe prop- althoughhis criteriaforbalance have been
ositionincludes,first,the assumptionthat clearly validated, empirically,only when
people are sensitive, in some degree at attractionis positive (Newcomb, 1968).
least, to cues revealing others' feelings More important,probably, are the phe-
about them, and are most comfortable nomenaof attribution (e.g., Kelley, 1972).
with requited attraction. The second as- If P's attractionto 0 is influencedby O's
sumption is that members will judge presumedattractionto P, and vice versa,
others' attractionto themselves with in- then attractionlevels that they exchange
creasing accuracy. The propositiondoes presuppose exchanges of attribution.
not rest entirelyon common sense, how- The data by which my presenthypoth-
ever; Jones and Gerard (1967:307), at the esis is tested consist mainly of weekly
end of a chapter citing many empirical rankingsof attractionby each of 17 mem-
studies,conclude that "It is clear thatat- bers of a living group to each other
traction begets attraction,that we like member,over a four-month period. The
those who appear to feel positivelyabout members(all men) were recruitedas total
us." They do not assert thatwe also dis- strangers;a second population,recruited
like otherswho appear to dislike us. one year later and then livingtogetherin
In social psychological terms, each the same ways, provided a replicationof
memberof a dyad both emits behaviors the firstset of data. (See Newcomb, 1961,
thatimplysomethingabout his or herown for a complete descriptionof all steps of
attitudetowardthe otherand attributesto the enterprise.) The data were not ob-
the othercertain attitudestoward him or tained forpurposes of testingthe present
herself.One can account forthe effectsof hypothesis.
interactionbetween dyad members by The 1961 data were designedto test my
Heider-like notions of interpersonalbal-
* Critical comments by my colleague, J. R. P. ance (Newcomb, 1953). These data con-
French,Jr.,on an earlierversionof thispaper have sist mainlyof weekly rankingsof attrac-
led to the exclusion of certainsections of it and the tion by each of the 17 membersof either
additionor revisionof others.Address all communi-
cations to Theodore M. Newcomb, 1045Cedar Bend populationto each othermember,
over a
Drive, Ann Arbor, MI 48105. period of 16 weeks. Findings indicated
299

This content downloaded from 62.122.79.56 on Sun, 22 Jun 2014 22:09:25 PM


All use subject to JSTOR Terms and Conditions
300 SOCIAL PSYCHOLOGY QUARTERLY

thatchangingrelationsof attraction,espe- WhytheNonconfirming


Findings?
cially those leading to clique formation,
One factcould contributeheavilyto the
influenceand are influencedby agreement
nonconfirmation of the hypothesis:accu-
(both actual and perceived) about values
racy of judging others' attractionto one-
and attitudes toward issues of common
self did not increase with acquaintance.
interest. Interpersonal attraction and
Correlationsbetweenranksactually given
agreementtended to change, interdepen-
by dyad membersto each otherand esti-
dently. Since that study and the present
mates (perceived attraction of those
one deal with interactionand reciproca-
others to oneself) of ranks given were
tion, the earlirerdata seemed appropriate
identicalat Weeks 1 and 15-+.47 at both
for the studyreportedhere.
times in Year II (estimates were not ob-
The hypothesis whose testing is re-
tained in Year I). Thus accuracy was at a
ported here is that continued dyadic in-
barely significantlevel, and did not in-
teraction involves mutual adaptation of
crease with acquaintance.
dyad members such that discrepancies
My central hypothesishas to do with
between levels of attractionthat are ex-
changes in reportedinterpersonalattrac-
changed (i.e., givenby personA to person
tion; insofaras level of attractiontoward
B, and by B to A) decrease withcontinued
acquaintance. The criticaltest of this hy- other persons is influenced by judg-
ments of others' attractionto oneself,
pothesis consists of comparingranks of
changes in suchjudgmentsmustchange as
attractionexchanged withineach of the
others' attraction to oneself actually
136dyads (172 - 17 2), pooled in each of
?
changes, if the hypothesisis to be sup-
the two populations,on early and on late
ported. It turnsout thatthe two kinds of
acquaintance. A decrease in discrepancies
change are not related. The evidence
duringthisintervalwould supportthe hy-
shows that there is no relationshipbe-
pothesis.
tween changes in ranksactuallygivenand
in estimatesof attractionto oneself (per-
ceived) between Weeks 1 and 15.
The Nonconfirming
Findings
Thus, while discrepancyis indexed by
Basic comparisonswere, in both years, differences between attraction levels
between Week 1 responses, after10 days actually exchanged, the hypothesisof de-
of acquaintance, and Week 15 responses, creasing discrepancyis based in part on
98 days later. (Comparable data had also the assumption that existing degrees of
been obtained on the thirdday, but re- attractionare influencedby theattribution
sponses were incompleteor in some cases to others of some degree of attractionto
of questionable validity.) As shown in oneself(perceivedreciprocationof attrac-
Table 1, there were virtually no dif- tion). The centralhypothesisis tested by
ferences between early and late frequen- discrepancies of actual ranks exchanged,
cies of intradyadic discrepancies. This while perceived attractionto oneself is
crucial test does not supportthe hypoth- used only in the search for reasons why
esis. Neitherearlynorlate distributionsof the hypothesis,tested by comparisonsof
discrepancies, in either year, differsig- actual ranks exchanged, was not sup-
nificantlyfromchance expectations. ported.

Table 1. Frequencies of Intra-DyadicDiscrepancies in Ranks

Discrepancy Year I Year II Chance


Level Week 1 Week 15 Week 1 Week 15 Expectations
0-1 42}74 } 71 34 66 36}69 1} 58

4>87 }62
416 4} 65 45 7 47} 67 42 } 78
Total 136 136 136 136 136

This content downloaded from 62.122.79.56 on Sun, 22 Jun 2014 22:09:25 PM


All use subject to JSTOR Terms and Conditions
RECIPROCAL ATTRACTION: NONCONFIRMED HYPOTHESIS 301
The unrelatedness ofthesetwokindsof ingofattraction to all othermembers with
changecan hardlybe attributed to ran- thesumsof ranksassignedto thosesame
domnessof responsessince, both early othersby all members(himself excluded,
and late,estimatesof attractionreceived of course). The higherthe correlation
are highlycorrelated(as noted below) betweenan individual'srankingand the
withown attractions givento the same group'sranking,the greaterthe group's
individuals. influence upontheindividual. Findings for
But whythisfailureto increaseaccu- Year II appearin Table 2; thedegreeof
racyofjudgments? The menlivedclosely normativenessobviously increases: at
together;theyhad dailyopportunities to Week 15, therewas no individualwhose
observeeach other,including others'be- correlation was not significant at .05 or
havior toward themselves.Their judg- beyond.
mentsof others'attraction to themselves A comparabledifference does not ap-
shouldtherefore become moreaccurate pear in Year I; themeancorrelations de-
withincreasing observation ofothers'be- cline,nonsignificantly, from.54 to .46. It
haviortowardthemselves.Two possible seeemedto me likelythatthisdifference
sourcesofsucheffects appear.First,cor- betweenthe two populationsmightbe a
relationsbetween(a) ranksactuallygiven consequenceofthefactthat,inYear I but
by dyad membersto each otherand (b) notin Year II, a sociometric split(involv-
those same persons'estimatesof ranks ing some hostility)developed between
receivedfromthe same otherswere sig- Weeks I and 15. The effectsof division
nificantlyhigherthancorrelations repre- would be to reduce the degree of
sentingaccuracy.Meansoftheindividual population-wide consensus. I therefore
rhocoefficients between(a) and (b) were consideredeach high-attraction cluster
.66 at Week I and .71 at Week 15. Both separately,correlating individuals'rank-
earlyand latejudgmentsof others'liking ingswiththegrouprankings oftheirown
foroneselfseemto be inferred fromone's cliques,at bothWeek 1andWeek15.The
own likingof thoseothers-a subjective mean correlationswere .45 and .41 at
mirror-imaging. Evidently,manyor most Week 1 and .63 and .68 at Week 15-thus
otherpersonsare seen as reciprocating,resemblingthe corresponding levels in
moreor less closely,the same degreeof Year II, fortheentirepopulation.
likingthatone hasforthoseotherpersons. Anotherindication of normativeness is
This constitutesperceivedbut not the shownby subjects'responsesto thefol-
actualreciprocity thatwas hypothesized. lowinginstructions (Weeks 12 and 14,
This subjectiveprocess is as prevalent each year): "'Divide the 17 men in the
afterfourmonthsof acquaintanceas at house intogroups,as you thinktheydi-
thebeginning. One is reminded oftheold vide themselvesinto groups. Make as
adage,"We see thingsnotas theyare but manygroupsas you want;put as many
as we are." menintoany groupas you want.If you
Norms about personal status. Again, thinka particular manbelongsin two or
one mustask whythisis so; whyshould moregroups,put him in both or all of
attributionsto othersbe heavilyinflu- them.Do notincludeyourself."
enced by "internal" processes when Two findings emergefromthesedata.
"external," behavioral evidence First,thosedyadsbothofwhosemembers
presumablyabounds? Anotherkind of were mentionedmostfrequently as be-
evidence suggests an answer: Widely
sharednormsdeveloped(in bothpopula- Table 2. Distribution of Correlationsof Individual
and Group Rankingsof Members' Attrac-
tionsthoughwithinteresting differences) tion Status (Year II)
concerning the relativeattractivenessof
groupmembers;theseled to thedevelop- Correlation
mentofwell-established statushierarchies Level Week 1 Week 15
of individuals'attractiveness. a .60 4 14 x2 = 11.8, p < .001
can be shownsim- ?.59 13 3
Such normativeness Mean .50 .67
plyby correlating each individual'srank-

This content downloaded from 62.122.79.56 on Sun, 22 Jun 2014 22:09:25 PM


All use subject to JSTOR Terms and Conditions
302 SOCIAL PSYCHOLOGY QUARTERLY
longingto the same group(or groups) are, Such findingssupporta probable basis
by and large the same dyads thatactually for the nonconfirmation of the initialhy-
exchanged highattractionratings.The x2 pothesis. One's own attractionto others
values, for Years I and II, respec- becomes increasinglyinfluencedby group
tively, are 22.2 and 34. 1, df = 1; thus norms,and one's judgmentsof others' at-
rankingscorrespond closely to observed tractionto oneselfare assumed to be fairly
affiliation.Second, the structuring of at- close reciprocations of one's own,
tractionis increasinglydifferentiatedwith group-influenced attractionto them.Such
acquaintance: as cliques form, the pro- judgments of attraction received from
portions of all dyads not mentioned as othersdo not closely correspondto actual
belonging in any group larger than two reciprocationfrom others, and thus the
(i.e., eitheror neithermentioned,but not hypothesizeddecrease in actual intradyad
both) increases at highsignificancelevels discrepancydoes not occur.
in bothpopulations.Thus thepopularmen
and the isolates become clearly distin- OtherPossible Sources of
guished, and each set of individuals re- Nonconfirmation
ceives many extremeranks.
These findings provide an important Artifactualconsequences of theranking
validationof the rankings,and also of the procedure? It can hardly be argued that
norms concerning attraction status. If these considerations,alone, account for
rankingsbecome "habitual," duringthe my "negative" findings. One possible
last monthor two, theirstabilitystemmed source of nonconfirmationis an appar-
notjust fromindividualperseveration,but entlyartifactualeffectof the use of ranks
also fromsocial reality. of attraction.Every subject must distrib-
There is another interestingway in ute all of the 16 given ranksamong all the
which normativenessmay have contrib- 16 othersubjects but thereis no such re-
uted to the nonconfirmationof the hy- quirementforranksreceived. In fact, ex-
pothesis. In Year II only, each subject treme ranks received tend, increasingly,
rated all the othersubjects on a 100-point to be concentratedamonga comparatively
scale of attraction,and also estimated few individuals. An individual who re-
ratingsgiven to himselfby each of the ceives only ranks at 13-16, for example,
others. Estimates of ratingsreceived by cannot assign closely reciprocatingranks
the four most popular men at Week 14 to all 16 subjects; such "artifactual" dis-
were significantly higherthanestimatesof crepancies occur both early and late, of
those received by the four least popular course, but they increase with acquain-
men; this was true for 15 of the 17 sub- tance.
jects, each considered separately. The It is obviously the case thatconcentra-
probability that 15 of 17 respondents tionsof similarranksreceivedby a person
would show a bias in the same directionis who must assign 16 differentranks will
< .001 (cumulativebinomialprobability). resultin a wide variationof discrepancies
Almost everyone, regardless of his own between ranks given and ranks received.
attractionstatus, assumed that he would And it follows that, since increases be-
be liked betterby the popular thanby the tween early and late responses in concen-
unpopular individuals. This findingmay trationof ranks received by individuals
reflectthe followingkindof "reasoning:" are the general rule, discrepancies be-
"I agree withthegeneralopinionthathe's
a greatand thoroughlydecent (or a gener- suggestedto me by Sonya R. Kennedy: It is possible
ally contemptible)guy. As such, he would that "popular' individualsare people whose gener-
probablynot dislike(or like) me." If such ally friendlyand pleasant behaviormakes othersfeel
processes were indeed occurring, they liked. If so, and if "we like those who appear to feel
positivelyabout us' (Jones and Gerard, 1967:307;
would suggest a stereotypingtendency- emphasisadded), an individual's"popularity" might
all good or all bad characteristics-often reflecta large numberof "likeability" votes based
predominatingover awareness of unique, on his abilityto make a largenumberof subjectsfeel
person-to-personrelationships.I liked. This mightaccount forthe finding,and might
suggest some furtheravenues for research on reci-
I Another explanation of this finding was procityof interpersonalattraction.

This content downloaded from 62.122.79.56 on Sun, 22 Jun 2014 22:09:25 PM


All use subject to JSTOR Terms and Conditions
RECIPROCAL ATTRACTION: NONCONFIRMED HYPOTHESIS 303
Table 3. Frequencies of Discrepancies in 100-point data were obtained account for the non-
RatingsExchanged by all Dyads, at Levels confirmation,or that the commonly ac-
and at Times Noted
cepted hypothesisis a dubious one.
Discrepancy
Level Week 1 Week 15
Dyadic Discrepancies as Related to Level
0-9 59 55 ofAttraction
10-19 45 56
> 20 32 25 x2 = 1.73 p < .50 At the outset I cited a statementof
Total 136 136 (2 df) findingsrelated to my hypothesisin the
formof liking,with no referenceto dis-
tweendyad membersone or bothof whom liking,the implicationbeing that the hy-
have increasinglyhigh concentrationsof pothesisdoes not necessarilyapply to dis-
similarranks,will inevitablyincrease with liking.This is typicalof researchersin this
time. Thus, given the fact of increasing field-e.g., Berscheid and Walster
concentrationsof similar ranks received (1978:40ff). One of the puzzles that
by many individuals, forces toward de- emergesfrommy data is that,in termsof
creasing dyadic discrepancies,over time, close reciprocation by dyad members,
would be counteredby such "artifactual" there is almost no differencebetween
countertendencies.If these tendenciesare dyads in which one or both membersre-
pronouncedtheycould, alone, account for ceive highlevels of attraction(ranks 1-4)
the nonconfirmation. and those receiving low levels (13-16).
The 100-pointratingprocedure, previ- The findingsappear in Table 4, in which
ously noted, was not constrained,as the frequenciesfor Weeks 1 and 15 are com-
rankingprocedure was, by the require- bined, since they show only trivial dif-
mentthatadjacent levels of attractionas- ferences. Not only is there no significant
signedto the other 16 men mustbe sepa- differencebetween reciprocationat high
rated by a fixed amount-one rank. Re- and low levels of attraction,but thereare
spondents could, and did, report large also no real differencesbetween the two
gaps betweenfirstand second choices, for populationsnor betweenearlyand late re-
example, or thirdand fourth,or 14thand sponses. Such a degree of consistency
15th,or (more rarely)between ninthand across populations,and especially across
tenth.Nevertheless, early and late rated earlyand late acquaintance, suggeststhat,
discrepancies did not differfromranked at both extremes of attraction,influence
ones, for purposes of testingthe hypoth- otherthan feltlikingor dislikingaffected
esis. For the totalpopulationof 136 dyads rankingsassigned. In view of the factthat
(Year II only) levels of discrepancy be- individuals' assignments of ranks were
tween ratingsexchanged did not differat much more influencedat Week 15 than at
Weeks 1 and 15, as appears in Table 3. Week 1 by norms of personal status,
Since these findingsagree closely with while degrees of discrepancy were not,
those providedby the rankingprocedure, othersources of distortionmusthave been
it cannot be concluded that the noncon- involved. I shall returnto a consideration
firmationof the hypothesisis attributable of what these may have been.
to the limitationsof the rankingproce- It seems to me, however,thatthe con-
dure. It appears either that certain fea- centrationof extremeranks among rela-
tures of the conditions under which the tively few individuals, especially after

Table 4. Numbersof Dyads in Which Ranks are Exchanged at Two Levels of Discrepancy,at Highestand
Lowest Levels of Attraction
AttractionRanks Received
High (1-4) Low (13-16)
Ranks of
Discrepancy Year I Year II Year I Year II

0-3 39 (57%) 43 (63%) 34 (50%) 36 (53%)


4-15 29 25 34 32
Total 68 68 68 68

This content downloaded from 62.122.79.56 on Sun, 22 Jun 2014 22:09:25 PM


All use subject to JSTOR Terms and Conditions
304 SOCIAL PSYCHOLOGY QUARTERLY

longacquaintance, is notmerelya regret-And,as we learnedlater,mostofthemen,


tableconsequenceof a procedureforob- neartheendoftheirfourmonthsofliving
taining data.Rather,itis a consequenceof together, had fewor no close friends out-
increasingconsensus; it is normalfor side the house, and nearlyall of them
norms to develop. It was this to-be- came to considerat least some of their
expected social influence,not a pro- fellow-members thereamongtheirclosest
cedural happenstance,that contributedfriends,anywhere,of the same age and
mainlyto thenonconfirmation of thehy- sex. Smallwonderthattheyall, including
pothesis.Since, not surprisingly, the de- the least popular members,developed
greeof normativeness increasedwithac- normsaboutattraction status.
quaintance,the inevitableconsequence Thereis anotheraspectofobtaining the
was to increaseintradyadic discrepancy, rankingsthatmayhave hadunanticipated
in oppositionto the hypothesizedde- effects.In particular,attractionratings
crease. Whether or notthehypothesizedwere, uniquely,obtainedin exactlythe
changes would, apart fromincreasing sameformat weekafterweekafterweek.
normativeness, have been supportedby It would hardlybe suprisingif, aftera
thedataremainsconjectural, butsurelyit few weeks, these responses became
is possible. routinized.In bothpopulations, therewas
Features of the researchsetting.It also relativelylittlechangeinthe"popularity"
seems likelythatcertainaspects of the ranks(the inverseof sums of all ranks
social settinghad unintendedeffects. received)aftertheearliestweeks.Corre-
While subjectshad close daily contact lationsbetweenranksgivenin adjacent
withone another,as I have noted,it was weeks were neverbelow .80 and were
in some ways enforcedcontact;theydid oftenin the .90's afterWeek 2. It seems
notchooseeach otheras livingmates,nor probablethatto somedegreesuchfreez-
weretheyfreeto leave the setting;they ing of attraction status,so early,repre-
were forcedto live, ratherintimately,sented habit formation-indeed, shared
even withpersonstheyhad come to dis- habitsthatby definition represent norms.
like strongly.Their lives togetherwere Ifattraction rankings hadbeenmadeonly
both close and closed. Could their veryearlyand verylate,theymighthave
"sharedhabits"of assigningrankshave been less "standardized"on final ac-
becomea wayofdoingwhatwas expected quaintance.In thatcase, the hypothesis
(afterall,theywerepaidforit),inspiteof mighthave been bettersupported.
thefactthattheydidnotreallymakeclear
distinctions amongthe attractiveness of Overview
othermembers, exceptperhapsat theex-
tremes?A bitof boredom,together with The principalsourcesof thenonconfir-
awarenessof group norms,mightwell mationof the centralhypothesisthatI
have contributed to responsesthatwere have consideredare the following:The
not onlyindividually routinized but also developmentof normsconcerningper-
group-shared. sonal status;the stability (ratherthanan
Certainaspectsof the settingvirtually increase)inaccuracyofestimating others'
insured that norms about individuals' attraction to oneself;and theuse ofrank-
popularitystatus would develop. The ingsas an indexof attraction. If each of
subjectshad muchin common:age, sex, theseelementsdoes in factcontribute to
membership in the same university, suc- the unexpected findings, they are
cessful previousexperiencein another presumably interrelated.The natureofthe
college or university, voluntary enroll- interrelationships may be somewhatas
mentin and commitment to a four-monthfollows.
residence together.Furthermore, they 1. The rankingprocedure required
enteredtheirnewuniversity on the same each personto viewhispeersintermsofa
day and, with a few trivialexceptions, hierarchy.
knowing no one inAnnArbor.Theywere 2. The weeklyassignment of ranksre-
thus,especiallyin theearlyweeks,thrust sultedin habituation, or routinizing, of
intoeachothers' arms forcompanionship. ranks assigned (fairlywell established by

This content downloaded from 62.122.79.56 on Sun, 22 Jun 2014 22:09:25 PM


All use subject to JSTOR Terms and Conditions
RECIPROCAL ATTRACTION: NONCONFIRMED HYPOTHESIS 305
the thirdweek, and changingverylittle group,and whichfacilitate thepursuitof
afterthe sixthweek). commoninterestsotherthan members'
3. The commonknowledge thatall sub- personalrelationswithone another.
jects weresimultaneously assigning ranks
of attraction, combinedwiththeirclose More Suitable Conditionsfor Testingthe
and continuing livingtogether,led to the Hypothesis
development of group norms about
everyone'splace in thehierarchy (again, What,finally, wouldbe theconditions
beginning early and stabilizingbetween underwhichthehypothesis ofdecreasing
theearlyand late responses). intradyad discrepancywouldmostprob-
4. Individuals'judgments aboutothers' ably be supported?My findings suggest
attraction to themselves came to be influ- twoofthem.Thoughtheyare quitediffer-
enced by the assumption thatone's own ent,procedurally, theyare interrelated.
attraction to otherswas moreor less re- First,groupmembership shouldbe less
quited.To an unknownextent,this as- arbitrarily determined;membersshould
sumptionconflictedwithor outweighed be freeto leave,andothersshouldbe free
considerations of interpersonal "fit." to join. Such arrangements would both
5. Finally,discrepancies betweenranks minimizea fixed status hierarchyand
exchangedby all dyads remainedcon- facilitate thereshuffling of smallcliques.
stant,insteadof decreasing,in spite of My data suggestthat persistingnorms
changes in individuals'status,because about individuals'attraction, commonly
normativeness itselfpersisted,unaffectedreinforced by close-knitcliques,tendto
by shiftsamongindividuals'positionsin oppose changesbothin expressedattrac-
thehierarchy. tionand in estimatedreciprocation.
It seems clear that my hypothesisis Another probable consequence of
"true" only (if at all) underconditions changingmembership is a decrease in
thatwerenotpresentin thesetting ofmy numbers ofhighlyunpopular members.If
study.The principalsourceof nonconfir-someof themshoulddropout and be re-
mation,I now believe, lies not in pro- placedbynewmembers, thegroupnorms
ceduralartifacts,but in the constraints concerning personalstatuswouldbe more
imposedupongroupmembers. In a closed elastic,so thatjudgmentsof others'at-
societynormsare apt to become more tractiveness wouldbe moredirectly based
rigidthaninan openone,whosemembers on observedinterpersonal behavior,and
changefromtimetotime.Whenentryinto estimatesof others'attraction to oneself
a societyis "arbitrarily" determined by moreaccurate.
"outsiders,"andwhenthosewhoarethus Second,themeasurement of attraction
chosen cannot leave, certain devel- shouldalso be less rigid.For example,
opmentscommonto all groupsbecome subjectscouldbe asked simplyto listthe
intensified. In the absence of common namesof personswithwhomtheywould
interestsas a criterionfor admission, like to spend time, or do thingsto-
members'person-to-person relationshipsgether.The centralhypothesis wouldthen
became a dominantcommoninterest.It be tested not by comparingearly and
followsthatnormsaboutpersonalattrac- late responsesin termsof discrepancy of
tivenesswouldstabilize,especiallyat the ranks, but in terms of the numbers of
extremes. Increasing concentration choicesreciprocated, bothearlyand late.
amongrelativelyfew individualsof re- Some such procedure would auto-
ceivinghighlikingor disliking is necessar- maticallyeliminatethe routinizednature
ilyopposedto a decreaseinintra-dyad re- of the rankingprocedure.Estimatesof
ciprocation. others'attractionto oneselfcouldbe ob-
Thus the everydayassumptionem- tainedin the same way: "Please listthe
bodied in my hypothesis-whichsurely names(perhapsincluding yourown) that
can be confirmed under certain you believe each of the othermembers
conditions-needsto be testedundercon- would list as preferred people to spend
ditionswhich,for example,permitvol- timewith,or do thingswith." Such re-
untaryjoiningand withdrawing fromthe sponses,I suspect,wouldbe morespon-

This content downloaded from 62.122.79.56 on Sun, 22 Jun 2014 22:09:25 PM


All use subject to JSTOR Terms and Conditions
306 SOCIAL PSYCHOLOGY QUARTERLY
taneous and perhaps more valid than REFERENCES
thosethatI did obtain.2 Berscheid, E., and Walster, E. H.
In general,ofcourse,a scaledmeasure 1978 InterpersonalAttraction(2nd ed.). Read-
of degreeof a variableis preferableto a ing, MA: Addison-Wesley.
dichotomous,yes-or-no"measure." In Byrne, D.
1971 The Attraction Paradigm. New York:
principal,a Thurstonescale of attraction Academic Press.
is feasible.But if the same scale were Heider, F.
used,repeatedly, responsesto itcouldbe 1958 The Psychologyof InterpersonalRelations.
frozen,just as the rankingswere. My New York: Wiley.
Jones, E. E., and Gerard, H. B.
guess is thatthe dichotomousmeasure 1967 Foundations of Social Psychology. New
would be less subject to routinere- York: Wiley.
sponses:toaddand/or subtractoneortwo Kelley, H. H.
names at various intervalsis probably 1972 "Attribution in social interaction." In
Jones,E. E., D. E. Kanouse, H. H. Kelley,
morelikeeveryday behaviorthanto rear- R. E. Nisbett, S. Valins, and B. Weines
rangea set of ranks. (eds.), Attribution:Perceivingthe Causes
Such a testof thehypothesis mightor of Behavior. Morristown, NJ: General
mightnotresultinconfirming it,butsuch Learning Press.
procedures would correspond more Newcomb, T. M.
"An approach to the studyof communica-
closelythandidthedatafrommyancient 1953 tive acts." Psychological Review 60:393-
studyto the conditionsof everydaylife, 404.
from which the hypothesisprobably 1960 "Varieties of interpersonalattraction." In
evolvedin thefirstplace. D. Cartwrightand A. Zander (eds.), Group
Dynamics (2nd ed.). New York: Harper &
Row.
2 A studyby Norfleet(1948), using such a proce- 1961 The Acquaintance Process. New York:
dure, reportsthataftera period of threeweeks dur- Holt, Rinehart& Winston.
ing which five groups of 12-14 persons held daily 1968 "Interpersonal balance." Pp. 28-51 in
worksessions therewere twiceas manyreciprocated Abelson, R. P., E. Aronson, W. J.
choices as desirable associates as afterthe firstfew McGuire, T. M. Newcomb, M. J. Rosen-
days. I have fairlyclosely duplicatedher procedure berg, and P. H. Tannenbaum (eds),
fromthe data cited above, by comparingearly and Theories of Cognitive Consistency.
late reciprocationsat highlevels of attractiononly. Chicago: Rand McNally.
Again, therewere no differencesbetweenearly and Norfleet,B.
late reciprocations.And, again, it seems likelythat 1948 "Interpersonal relations and group pro-
group norms, becoming increasinglyconsensual, ductivity." Journal of Social Issues
overrodeotherdeterminantsof close reciprocation. IV(2):662-669.

MANUSCRIPTS FOR THE


ASA ROSE SOCIOLOGY SERIES
Manuscripts( 100 to 300 typedpages) are solicitedforpublicationin
the ASA Arnold and Caroline Rose MonographSeries. The Series
welcomes a variety of types of sociological work-qualitative or
quantitativeempirical studies, and theoretical or methodological
treatises.An authorshould submitthree copies of a manuscript for
considerationto the Series Editor,ProfessorRobin M. Williams,Jr.,
Department Ithaca,New York 14853.
of Sociology,Cornell University,

This content downloaded from 62.122.79.56 on Sun, 22 Jun 2014 22:09:25 PM


All use subject to JSTOR Terms and Conditions

Das könnte Ihnen auch gefallen