a

POCUMEMT RESUME

AUTHORTITLE

_,11011 DATENOTE,-

E BS PRICEDESCRIPTORS

IDENTIFIERS

troxton, Jack S.; Miller, Arthur. G.'Impression Formationland the Attfibu ionAttitudes: "Sleeper" Effect?May BO20p.; Paper presented at the Annual Meeting of theMidwestern Psychological Association (52nd, St.Louis, Mitl,sMay 1-3, 1980) Best copy available.

1F61/PC01 Plus Postage.*Attitude change; *Attitude measures; *Attribution ,

Theory; *Behavior Theories; *Influences; Time Factors(Learning); *Time Perspective*Impression Formation

ABSTRACTRecent studies indicate that investigators.arenow-

focusing on the. cognitive determinants-of 't.he-.attribution ltoroCess:..however, few reSear- ers.are looking specifically at the attributionprocess. over time The itpactofattitudinal'and behavioral ..

information .cn mpression formation was studied to determine howimpressions changeover time. T.4 attitudinal information-given. to BO"-subjects consistedof_a'series at statements purportedly made by a.target person; the behavioral information consisted of. an, essaypurportedly written_' by.that person. Subjects were :told either thatthe essay. .position had been freely Chosen, by the target 'person ot4that-tie essay-pdsition had been' assigned. The influence Ilf-theessaywas initially -a function of eoth its contentand-dia*loSticity. Twoweeks later however, the diagnoStiCity of-the behavior exertedlittle influence. This result was similar to,the-"sleeper" effect inthe area of attitude change. Findings suggest that it may bedangerous to assume that the immediate discounting of a givenbehavior means that the impression of the actor will remainunchanged, or that the content ofthe behavior will be ignored..:(Author/KMF)

*** *** * ******** ** *** ** * ***ReproddetionS supplied by EDR5 are the beSt tha

from the original document.************ ******* ******* ************

* *** * *** * * #** * * **

can, be made

***

CQ

0

0

IMPRESSION FORMATION.ANb. HE ATTRiBUTIOH'OF-ATTITU6kS.-.7 -

A "S ES ER" EFFECT?

Jack S. Croxton.Sta_e University of New ydrk = College at FredOnia

A paper presenAssociation.

Arthur. G. MilleritY (Ohio.

at the 1980 meeting of the Midwestern Psychological

U'S DEPARTMENT QF HEALTH.EDUCATION (WELFARENATIONAL INSTITUTE OF

EDUCATION

TEWL DOCUMENT FMTT1 BEET, E PF1O.

DUCED EXA1T1Y Ai pccroiro FOOMpEpwf., IJR opc,ANIzATIoN opirol-

AriNt-i IT PD,NTTL OF vi rod',TA TED DO NOT NFCFS', ARII_Y urpoSENT OFF AL NAT IONAL NT, T I TiiTi TE OFEr_3,r T pos. TION OR POI it v

PERMISSION TO REPRODUCE THISwTERIAL EI EN OnANTiD RY

TO THE EDUCATIONAL RESOURCESINFORMATION CENTER. (ERIC=)

s

approach to the study of psychological 'issues. Nowhere

trend more dvdent than in the recent attempts to inves

,relationship between. Coglii 1-447a processes4

'and social behavior (cf.

OgistfIL

INTRODUCTION

e, currently advocating. an.intetdisciplinarY

Carroll & Payn 1976, Harvey Ickes, & Kidd, 1976, 1978;:Wyer 1974).Inpviduals mmonly find themselves in situations where they must

te ate',diverse types and often large quantities of information

and ubsequently make an interpretation or arriie at some type of

judgment based-upon their dilderstanding of that information.

.Researchers,in the area of attribution theory have focused,

on a common themes How ao pebple interpret their own behaviors

and the behaviors of otilers? Attribution 'theorists implicitlyassume that individuals are ntringically interested in the causes

,of behavior' since this enables them to anticipate and control the

behaVior Of others in given.eitilatione. Pntil recently, attrition

theorist6 .have hbt-d_ireet y investigated cognitive processes such

as encoding and retrieval. Rather, the focus has been on the

resultant attribution itself accompanied by an Implicit acknowledgment

of the 'intervening processes. Studies'using the typical attribution

paradigm have-converged in a number of well-corroborated findings?

The.eletent to which behavior is perceived to reflect an internal

disposition of an actor is positively related to the dygree to which,

that behavior appears tobe,free of any outside influences (Heider,

19581 Jones & Davis, 1965I.Kelley, 1971 This principle has been

demonstrated in- a-mutberAlf

19614 Steiner 19604 ,Thibaut &.klecken:.1955)

even hmmi potential constraints on the actor '.s behavior are, madE

quite evident,to an observer,, the tendency remainE4to assume that

- the behavior_ of an adtor-reflects the actor's true InalinatiOns.

Thid phenomenen has -been dimonstrated empii7ieally on a number of

differdnt occasions (Jones & Harr is ; 1967: Schneider & Miller; 19754

Snyder & jefitis; 1976: Miller, 1976)0' Stimulus --materials are9

-pieally in the form of an essay that has purportedly been written

by the actor.

Attribution theorists see thit individuals reflect on the

available information when asked `..to make an attribution of some type.

A few investigators are currently studying the relationship between

memory.and the attribution process. For example, Pryor and Kriss

(1977)in two related studies manipulated the salience of information

and found reaction times for recognizing the more salient information

to,beless and attributions of causality to be greater. Taylor andr ,

Fiske (1979) assumed that memory .might mediate perceptions .of

causality but their results were inconclusive. Harvey, Yarkin,

Li ter Breslauer, and-Strange (1978) found a positive relationship,

between the.number of causalstatements made by an observer when

asked to view an interpersonal encounter and the accuracy of recall

r the specifics of the interaction. The authors also discovered

that subjects who were given instructions to remember as much as

Tossible ebOut:the interaction were more apt to later.desorihe

it in terms -of causal units,:. -As these recent 'studies

,investigators are.now focusing on the cognitive detertirnints of

,:attributionprodesses. Relatively few researchers, however, -have

4

-lookedspecifically

time.Therefore the

.e attribution proC ol'er he co

or pu4000 ef the pros z 4. ex +p

determine how impress one and attributiondchangeivith the -pa a

to

of time.

An individual's impression of a given person ,should naturally

reflect the information made available about that person.. If

pattern of information differs from one individual to another,

their corresponding impressions-should reflect this difference.

However, behavioral inforMation may not always be consistent wi

previously held impression. In the present experiment we var

he, consistency between the pr or-information-provided about a

ven,person'and the subsequent behavior performed by that person.

I addition we varied the diagnosticity of that behavior. The

-reasions people form should be influenced by-the accumulation

of new information._ If a behavior of an actor 'ie consistent. with

th impression formed from previous statements-made by that actor,

Ite impression should be strengthened. If the behavior ,is incon-

sis ent however, the impression should Vs Weakened. Furthermere,

the impact of a given, behavior should be related-to its diagnos-. .

tiji Impression change should/be more likely,to result, when

the actor freely cheoSes:toengage in the.behaiiiori However such

chan

cons

should beiess likely when the actor's behavior has been

fined in some manner. ,Our goal west() test the above

pred4tions, both when the resulting impressions and attributions

easured immediately and when they are measured after a passage

-of.time.

rolETHOp'

A total of.80 subjects from the subject: pool of the

Two target pereOns were created by varying the attitude

purportedly ,ode each. Ten `liberal attitude statements

OW neutral statements were choler ..as those to be made by the

`Ten conservative attitude statements and

thes same ,ter} neutral statements were ehosen as those to be made

the Conservative get, Person. Two essays were also constructed,

\one generally favoring the legalita of- marijuana and one

general* op sing legalization.

The attitude stimuli were randomly displayed on a white screen

for apprOximately three seconds by a Kodak Ektographic RA-960

Random AcCess Projector., All subjects were told that. these

statements had been endorsed by a fellow undergraduate. Half.of

the subjects viewed the te'n liberal statements and the ten irrelevan

statements. The remaining subjects Viewed the ten conservative

statements and the same ten irrelevant .statements.,

All Aubjects were then given-an essay to read that had been

purofted1y written by the target person. eHalf of the subject-J.

were given the pro- marihuana essay to read and half were given the

- marijuana essay. Furthermore, half of the subjects were

told that the target person had freely chosen tha position taken

in the essay and half were told that the target person had been

as gned the position, taken.

The design consisted of a 2 (Target Pers nIg-attitudess Liberal

vs,. Conservative) X 2 (Essay Directions Pro-marijuana vs. Anti-

a) X 2 (Writer's Choices Free Choice vs. Assigned) complete

h 10 sub jects 'randomly. assi ned ,to each of the eight

experimental conditlenw.

After everyone had finished reading the e- ay subjec were-

asked to estimate the arget person's true attitude toward the

legalization of marijuana on a 0 Strongly Opposed) to (Strongly

Le.. Subjects were

which containedlthirty new attitude statements 'and were instructed

-in Favor) c given a deck of computer cards

.to:Tank the-statements:appearing on the cards from those: most likely

to be made by the target perfson to those least likely to be made

by that. person. These statementswere also rated by subjects on

a censervative-liberaldimension. A subject's rating of each

statement was -ffiultiplied-by its ranking. These resulting products

were summed in ,order to provide a measure o4 each sub ect's current:

impression of the tget person. Lower numbers indicated a more

liberal 4mpression and higher numbers indicated A sore conservative

impression. Subjects also simply rated the target person on a

0 (Conservative) to 10 (Liberal)

All participants returned approximately two weeks later. for

the second session. Afthis.tile ,subjects respOnded to the dependent

measures once again. At theconcluttion of the second session,

each p cipant was thanked for his or her parti.ci.pation in the

research projdct.

Results. The impact of the essay on subjects impressions was

assessed analyzing their ranking of the'attitude statements.

A marginally siinificant two-way interaction between essay direction

and constraint (F = 3.49, df = 1/79,-p( .10) reveals that:during-

the first session the effect of essay direction on subjects'

.,impressions was greater when the essay position was freely chosen

:trier than assigned. The nature-of the in

in Figure However analysis of the ranking

-weeks-lIter rev gIs nO-such interaction (

_njs displayed

collected two

=1/

Instead there is only amain effect for essay direction

14, df = 1/79, e(.05) Furthermore an analysis of the

change scores between the firitt and second sessions reveals that

the effect of essay direction is more pronounced at Time 2 than

at Time 1 (F = 4 03 df = 1/79. p<.05) This difference

displayed in Figure 2)..

An analysis of the ratings Of the target' person on

ConseryatiVe-Liberal dimension reveals a sign

ConStraint interaction -(F 7 8.99'0 df =1 79i,I):(41)fduring:the

first session. The effect of essay direction Al ratings was greater-

he

ficant Essay Direction

when the position taken in the essay had been freely chosen rather

than assigned. This interaction is- displayed in Figure 3. An

analysis of the ratings made two weeks later reveals no such

interaction (F = 2,50, df = 1/79, p-= ns).

Thus, the behavior engaged in by the target person did affect

subjects' impressions. HoWever0 the amount of constraint placed-,

on the target person when writing the essay only mediated this

effect when the impressions were measured immediately after subjec

had read the essay. Two weeks later, only the position taken in ,

the essay was significantly related to subjects' resultant impression;

As erected, attributions ofattitude were influenced by

the attitude information provided about the target person, the

position taken in the essay, and the constra nt..placed on the target

person when writing the essay.. The effect of essay direction

s much eatexe

hosen rather

gardiess of

resulting.int

when thS,' position-ion .t an -freely

assigned to the target p.erson., This Was true

th?spravious sttitude- inform**ion provided. The

raction between essay direction and constraint

tithin.each.Attitudeaonditiois,diSpianaphicallYinFigure 4.

This inters.

immediately

made two wee 'later me 1i

n was obtained b6th when the attribution was made

er reading the ess'ay and when the attributiOn was

df 1/79. p< 011 Time

sr pOit2 F:.= 40.550 df 1/79 p( .01). when the e.freely, chosen by the target ,person, the, effect of essay direction

on .attributions of attitude was SubStantial..(TiMe'it PH=489.94

p<.01;..Time .30.00 df = 1/39, p( .01) .,

However, when the essay position was purportedly assigned to the

targqpersenl,the effect of essay direction on attributions of

attitude was. minimal (Time it 7 = .170 df = 1/39, p = As: Time 2:

=.040 df

On the

inferma.ion

'.diagnostidi

was greater when. the essay position was assigned to the essay wr

rather than freely. chosen.. The interaction_ between attitude

information, and constraint was statistically significant both when

the attribution was made immediately after reading the essay and

when the 'attributionwaS made o weeks later (Time 1:.P

df = 091 134.01; Time 2: F = 1/79, p..05 When the

essay position was freely chosen by the target person,. the effect

on was

= 1/39, p = ns .

other hand, as expected, the effect of the attitude

on ,attribut ens was inversely related to the dss

That i the effect of-the attitude, nformat on

of atitude information on attributions was netattributions

F = 1.47, df = 1/39, P = ns; Time 2: F = 1.39, d

er

ign c t (Time is

= I/391 P no

However, when the essay position was purportedly assigned to the

target person, the effect of Attitude Direction on attributions

was substantial (Time 11 F = 16. df f 1/39; p<.01: Time 2,

= 10.64 df = 1/39. P4 .01 ) . In BUM, the utilization of the

behavioral information appears to have been directly related to

its diagnoSticity, while the utilization of the attitude information

was inversely related to the subsequent behavior's diagnosticity.

Overall, these results provide a, strong validation of previous

findings in the area of attitude attribution.

DISCUSSION

Based on our results can conelude'that individuals form

predictable impressions about others on the basis of specific

attitudinal information when combined with subsequent behavioral

information. It should beThoted that the attitude information

provided for subjects in this research contained no glaring in-

consistencies. In real life people's attitudes are probably not

as systematic as those presented here. In fact, at times people

may seem to be a bundle of contradictions. On the other hand,

certain behaviors are more informative about an individual'stl

true nature than others. In this research the, directionality ofo

the behavior (i.e., essay) was varied but each essay was only of

moderate strength. If the essay writer had appeared to take a

strong stand on the issue of the legalization of marijuana, the

impact of this information may.have been greater. A more detailed

investigation would involve* varying the strength as well as the

direction of the stimulus information. In fatt there may be ways

of solidifying an impression in the mind.of-a pereeiver other

1i)

than vary ng

-9-

ength. Te er- (1978 ) has demOnstrated that

attitudes toward a stimulus object become more polarized when

individuals are instructed to spend time thinking about that

object. Perhaps the Ampression of a target person can be

strengthened in an analogous manner by asking individuals to

think about the persoft,for a few minutes based on the information

provided. Another possibility is to have perceivers explicitly

describe the target person in writing. Either type of prObedure

might make the impression more resistant to change.A:

As in other attribution studies, we have found th

of behavioral direction on attributions of attitude is less when

e effect

the behavior has been constrained in some manner. Moreover, thisH

is true regardleas of whether the attribution is made immediately

or later in time. -Howeverd the lack of Any essay direction effect

in the assigned condition appears to be inconsistent with earlier

results which indicate that observers tend to continue to make

attributionafof attitude consistent with the essay position even whe3'

the position has been assigned. The majority of these previous

studies,- however did not specifically create a behavioral expectanc

Jones, Worchel, Goethals, and Grumet (1971) did manipulate behaviora

expectancies and discovered that when weak essays were written

Ander constraint conditions and were inconsistent with the behaviore

expectancy, judges attributed attitudes in reverse of the essay's

directionality. Since a behavioral expectancy was created in the

present research, and since the essays used were only moderate in',

strength, the lack of any direction effect in the assigned conditioy

is not surprising.-

The finding that attributions made two weeks later paralleled

use made immediately after reading the e

its of a- re'iont study conducted by Jon0

-(1979)R d'Auattrone

vheitti authobs, origirially thought that the

information t be fgrFotten later. Therefore, the

that the attributions of perceivers who were initiall

that-lhe essay position had been assigned.would later

assigned

p edicted

ormed

shift in a

direction more correspondent ith.the essay position. However,

they discovered thatthis was not the'ease. The results of our

research are consistent with this finding.

Perhaps the most interesting outcome of our research is the

finding that initially the influence of a behavior on people's

impressions is a function of both its direction and its diagnosticity.

As time passes however, the diagnosticity of that behavior becomes

less important. The dims nishing effect of the'constraint information

is analogous to a sleeper effect in the area of attitude Change.

Previous results indicate that the information in a pei-suasive

communication tends to be remembered while related information.

Concerning the source's credibility tends to be forgotten (Hovland &

Weiss, 1951: Kel an '& Hovland, 1953). In the present case; the

position t_aen iii the essay was apparently remembered two weeks

later:while the accompanying constraint information no longer

exerted, effect. Of course, if the essays had themselves been.

extremely strong.the constraint information may have wielded little

influence either initially or later. In any case, it does appear

that With the passage,'of time certain types of information may

assume more weight than others and that later impressions tend to

reflect this diffetence.

Apparently the constrain4

_a on was impo

perceivers when it had direct licgtions judgment to be

made, such as an attribution. However, the directionality of

the essay took precedence over the accompanying constraint infol

mation when perceivers were asked to indicate- their general impression

of the target person. This result may have certain practical

implications. Suppose, for example, that a prisoner or war or a

hostage makes a series .of, unpatriotic statements,. Observers May.

readily acknowledge that the behavior was constrained and they may

also assume that these statements are not an accurate reflection

of the person's true attitudes. However, if the results of this

research are ey -.cation, impressions of the person may still

change as a result of the person's behayior. Thus it may be

dangerous to assume that the i )ediate discounting of a given

behavior means that the impression of the actor will remain

undhanged or that the content of the behavior will be ignored.

Whatever direction future research in this area should take,

it promises to be both provocative and stimulating. more importantly,

the outcome of such re -carob should contribute substantially to

better understanding of human behavior.

'Conserva ve 2500

L AL. ...,

2450

2400

235C

2300

2

200

_)()

tido

zuf)t)

U

Pr,o-mari Juanaessay

o Antl-mari anaessay

yy 24 '4 A A

1 .1 .and At[5 costrain, (¼L 1)

Conservative 2500

2400

2350

2300

2250

2200

2ItA

2100

zW,0

zouo

A

Anfl-marijo8naessay

o Pro-mail JUdlicie6bay

1 A

_- joana essayPro arl Juana- essay

ti,,i,e gnedA5, a t A

1 . .ti e ii. Ion a f.htiotL ut LAN Llirtk1un WI I, dill( 1

Pro 10

Anti.

9

2 -

1

Choice

Atticuc1e 5t.

Pro

9 -

8 --o

7

6-0

5 -

0

cjUr LAttr;buLion of 3t..tit 4. :1,ALI-LIpn dt_and writer's constraint (Tir,e 1)

aL_ Anti-marijuanaessay

Pro-Tarijuanaessay

Coice

,=Jn6.e_rva

_ -.,itude_hera, essa :21rec.:ionlt,

,

HEFERENCES

Carroll, J.S.: Payne, J.W. (Eds.). Cognition and social behavior.Hillsdale, N.J.: Lawrence Erlbaum Associates, 1976.

Harvey, J.H., Ickes, & Kidd, R.F. (Eds.). New direction inattribution research (Vol. 1). Hillsdale, NI s Lawrence ErlbaumAssociates, 197 .

Harvey, J. I., Ickes, W.J. & Kidd, R.F. (Eds.attribution research (V.01. 2). HillsdaleAssociates, 1978.

'Harvey, J.H., Yarkin, K.L. Lightner, J.M., Breslauer, H.S., &Strange, A.L. Unsolicited interpretation and recall of interper-sonal events. Paper presented at the annual meeting of the Mid-western Psychological Association, 1978.

Heider, F. Th&ml.asholo of inter ersonal relations. New York:John Wiley & Sons, 195 .

Hovland, C.I., & Weiss, W. The influence of source credibility oncommunication effectivenesp. Public Opinion quarterly, 1951,11, 635-650.

Jones, E.E., & Davis, K From acts to dispositions' The attributionprocess in person perception. In L. Berkowitz (Ed.), Advances nex erimental social gyshology (Vol. 2). New York: AcademicPress, 19 .

New directions inawrence ErlbaumJ.1

Jones, EE., Davis, K.E., & Gergen, K.J. Role playing variationsand their information value for persozl perception. Journal ofAbnormal and Social Ps cholo 1961, 302-310.

Jones, E.E., & Harris, V.A. The attribution of attitudes. Journalof erlmental 1967, 2, 1-24.

Jones, E.E., Riggs,-J.M., & Quattrone, G. Observer biasattitude attributimm paradigm' Effect of time and infoorder. Journal of PersonalitY and Social Psycholo

Jones, g.E., Worchel, S., Goethals, G.R.;& Grumet, Priorexpectancy and behavior extremity as determinants of attitudeattribution. Journal of Ex:erimcntal Social Ps cholcZ. 59-80.

Kelley, H.H. Attribution in social interracti.on. Morristown. N.J.GenerpkLearning Press, 1971.

Kelman, H.C., & Hovland, C.I. "Reinsta emen of the communicatuLin delayed measurement of opinion ch nge. ournal of Abnormaland Social Psychology, 1953, 48, 327*335.

Miller, A.G. Constraint and target affects in the attribution ofattitudes. Journal of Ex erimental Social Ps cholo 1976,la. 325-339

Pryor, J.Bi, & Kriss, M. The cognitive dynamics of salience in theattribution process. .1211171g12fp2E222211tySocial Psychology,1977. 35, 49-55

theation

1971,

Schneider, D.J., & Miller, R.S. The` effects of enthusiasm and(auali of arguments on attitude attribution. Journal of Per on1975. 693-708.

nyder, M., & Jones, F.E. Attitude attribution when behavior isconstrained. talasipyg2.1PscholocJournalofE), 1974,10_, 585-600.

Steiner, I.D. Field, W.L. Role assignment and interparsonalinfluence. Journal of Abno m and Social Ps cholo 1960, 61,

Taylor, S.F., & Fiske, S.T. Point of view and perceptions ofcausality. 1975,la, 439-445

Thibaut, J.W., & Riecken,of the perception of s1955, 2 113-133.

H.R. Some determinants and consequencescial causality.r Journal of Personality,