Journal of Personality and Social Psychology 1979, Vol. 37, No. 12, 2221-2239 An Examination of Self-Perception Mediation of the Foot-in-the-Door Effect William Dejong Brandeis University In 1966, Freedman and Fraser demonstrated that an individual is more likely to comply with a large request for help if that person has previously agreed to an initial small request—a phenomenon they called the "foot-in-the-door" ef- fect. In the present survey, studies that have sought to replicate the foot-in- the-door effect are reviewed. The adequacy of a self-perception explanation for the foot-in-the-door effect is assessed by examining (a) the importance of the size of the initial request; (b) the effect of noncompliance with the initial re- quest; (c) the impact of salient external justifications for the initial act of compliance; (d) the impact of social labels on subsequent levels of compliance; and (e) attempts at actually measuring changes in self-perception. Alternative explanations of the foot-in-the-door effect are considered and rejected, and di- rections for future research are outlined. In 1966 Freedman and Fraser tested the notion that once an individual has complied with a small, sometimes trivial request, that person will be more likely to comply with a larger and more substantial demand made in the future—an effect they christened the "foot-in-the-door" phenomenon. During the last 12 years, social psychologists have con- tinued to be fascinated by those results. How is it that the simple act of assenting to a small request for help can dramatically increase, even double, the probability of a person's agreeing to give help in the future? Because the foot-in-the-door paradigm has become an important vehicle by which to study the link between self-concept and be- havior, it is important to take a step back and ask some basic questions about the progress of this research: (a) Is the foot-in-the-door Preparation of this article was supported in part by Public Health Service Postdoctoral Fellowship 1/F32/MH07294-01. My deepest appreciation is extended to Teresa Amabile, Nancy Cantor, David Funder, Albert Has- torf, Mark Lepper, and David Rosenhan for their comments on an earlier draft of this article. Requests for reprints should be sent to William Dejong, Department of Psychology, Brandeis Uni- versity, Waltham, Massachusetts 02154. effect a reliable one? Under how wide a range of situations has it been demonstrated? (b) Self-perception theory (Bern, 1972) is most often used to explain the effect. What hypoth- eses can be derived from this explanation, and what evidence has been gathered for each of them? (c) How else can the phenomenon be explained? What is the evidence for and against the alternative propositions that have been offered? (d) What directions can future research take? It was in the hope of answer- ing these questions that the present article was designed. First, I begin with a review of the Freedman and Fraser (1966) experi- ment. The Freedman and Fraser Study The procedure of their experiment was straightforward. A male experimenter first contacted several suburban housewives in their homes, identifying himself as a member of either the Community Committee for Traf- fic Safety or the Keep California Beautiful Committee. The initial requests he made of these women were small and innocuous, and almost all of the subjects agreed to them. Half were asked to display a small sign in the front window of their homes. The other sub- Copyright 1979 by the American Psychological Association, Inc. 0022-3514/79/3712-2221$00.75 2221
Transcript
Journal of Personality and Social Psychology1979, Vol. 37, No. 12, 2221-2239
An Examination of Self-Perception Mediationof the Foot-in-the-Door Effect
William DejongBrandeis University
In 1966, Freedman and Fraser demonstrated that an individual is more likelyto comply with a large request for help if that person has previously agreed toan initial small request—a phenomenon they called the "foot-in-the-door" ef-fect. In the present survey, studies that have sought to replicate the foot-in-the-door effect are reviewed. The adequacy of a self-perception explanation forthe foot-in-the-door effect is assessed by examining (a) the importance of thesize of the initial request; (b) the effect of noncompliance with the initial re-quest; (c) the impact of salient external justifications for the initial act ofcompliance; (d) the impact of social labels on subsequent levels of compliance;and (e) attempts at actually measuring changes in self-perception. Alternativeexplanations of the foot-in-the-door effect are considered and rejected, and di-rections for future research are outlined.
In 1966 Freedman and Fraser tested thenotion that once an individual has compliedwith a small, sometimes trivial request, thatperson will be more likely to comply with alarger and more substantial demand made inthe future—an effect they christened the"foot-in-the-door" phenomenon. During thelast 12 years, social psychologists have con-tinued to be fascinated by those results. Howis it that the simple act of assenting to a smallrequest for help can dramatically increase,even double, the probability of a person'sagreeing to give help in the future?
Because the foot-in-the-door paradigm hasbecome an important vehicle by which tostudy the link between self-concept and be-havior, it is important to take a step back andask some basic questions about the progressof this research: (a) Is the foot-in-the-door
Preparation of this article was supported in partby Public Health Service Postdoctoral Fellowship1/F32/MH07294-01.
My deepest appreciation is extended to TeresaAmabile, Nancy Cantor, David Funder, Albert Has-torf, Mark Lepper, and David Rosenhan for theircomments on an earlier draft of this article.
Requests for reprints should be sent to WilliamDejong, Department of Psychology, Brandeis Uni-versity, Waltham, Massachusetts 02154.
effect a reliable one? Under how wide a rangeof situations has it been demonstrated? (b)Self-perception theory (Bern, 1972) is mostoften used to explain the effect. What hypoth-eses can be derived from this explanation, andwhat evidence has been gathered for each ofthem? (c) How else can the phenomenon beexplained? What is the evidence for andagainst the alternative propositions that havebeen offered? (d) What directions can futureresearch take? It was in the hope of answer-ing these questions that the present articlewas designed. First, I begin with a reviewof the Freedman and Fraser (1966) experi-ment.
The Freedman and Fraser Study
The procedure of their experiment wasstraightforward. A male experimenter firstcontacted several suburban housewives intheir homes, identifying himself as a memberof either the Community Committee for Traf-fic Safety or the Keep California BeautifulCommittee. The initial requests he made ofthese women were small and innocuous, andalmost all of the subjects agreed to them.Half were asked to display a small sign in thefront window of their homes. The other sub-
Copyright 1979 by the American Psychological Association, Inc. 0022-3514/79/3712-2221$00.75
2221
2222 WILLIAM DEjONG
jects were asked to sign a petition advocatingcertain legislation. These requests were con-cerned with one of two issues, driving safetyor making California more beautiful.
Two weeks later, the housewives wereagain contacted. A different experimenter,who claimed to be a representative from theCitizens for Safe Driving, asked the womenif they would be willing to have a large, uglybillboard reading "Drive Carefully" installedin their front yard for a period of one week.Thus emerged a simple two-way factorial de-sign. For half of the subjects this second re-quest was like the first in terms of the actionrequired of them (i.e., displaying a sign). Forthose who had previously signed a petition,the action now required of them was quitedifferent. As a cross-dimension, the issue ofconcern was the same as before for half of thesubjects (i.e., driving safety). Those orig-inally approached by the Keep CaliforniaBeautiful Committee were now dealing witha different issue.
Freedman and Fraser demonstrated a foot-in-the-door effect of remarkable strength andgenerality. Each of the four experimental con-ditions produced significantly greater compli-ance with the final request than did a controlcondition in which subjects were not ap-proached with a first request: Only 20% ofthe subjects in that control condition werewilling to have the billboard installed, com-pared with 55% of the experimental subjects.To be sure, those subjects for whom the sec-ond request was similar to the first along thetwo dimensions of issue of concern and modeof action were somewhat more helpful thansubjects in the other three experimental con-ditions, but that difference did not approachstatistical significance. This increased compli-ance effect did not seem to depend much onthe type of request that had been made in-itially. Even the seemingly trivial action ofagreeing to sign a petition about the need to"keep California beautiful" more thandoubled the probability that a subject wouldagree to display a billboard that exhortedmotorists to drive more safely. The excitingfeature of these results is that the effectcannot be explained merely in terms of eitherthe subject's involvement with a particular
experimenter or her increased commitment toa certain issue or mode of action. As Freed-man and Fraser recognized, a more compli-cated (and psychologically interesting) ex-planation is required.
Attempts at Replication
Since Freedman and Eraser's (1966) orig-inal report, several replications of the foot-in-the-door effect have been attempted. Beforethis research is reviewed, the criteria that anystudy must meet before it can be considereda valid replication attempt should be listed:
1. Obviously, a proper control group, con-sisting of subjects not receiving an initial re-quest, must be included as part of the experi-mental design. Three studies must be excludedfor this reason: Harris, Liguori, and Joniak(1973); Harris and Samerotte (1975); andSchmidt (1973). The first two, in all fairness,were not explicitly designed to test for a foot-in-the-door effect.
2. The data analysis must include all sub-jects assigned to the experimental conditionand not just those who agree to the first re-quest. Any analysis that excludes those notcomplying with the first request can be crit-icized on grounds of differential subject self-selection. Studies by Arbuthnot et al. (1976-1977) and Beaman, Svanum, Manlove, andHampton (1974) must be excluded from con-sideration because of this problem.
3. High compliance with the initial requestmust be obtained. Harris, Liguori, and Stack(1973, Study 3), for example, failed to obtaina significant foot-in-the-door effect, but inter-pretation of their findings is made problematicby the fact that only 42% of their experi-mental subjects agreed to the first request(see also, Dutton & Lennox, 1974). Ideally,of course, 100% of the subjects would con-sent to the first request, but that goal israrely obtained. Some studies have reporteda significant foot-in-the-door effect with arate of initial compliance as low as 67%(e.g., Freedman & Fraser, 1966, Study 1).
It should be noted that in testing for sig-nificant foot-in-the-door effects, each experi-mental group should be compared individuallywith the control group. Some studies (e.g.,
FOOT-IN-THE-DOOR EFFECT 2223
Cann, Sherman, & Elkes, 197S; Harris, 1972,Study 1) combined experimental groups intheir analyses, and the data from such studieshave been reanalyzed when possible.
Table 1 summarizes known replications ofthe foot-in-the-door phenomenon that hadbeen attempted by the time of this review(November 1978). The table lists the tworequests made of the subjects, the rate ofcompliance with the second request for boththe experimental groups and the no-initial-request control group, and the conclusionsdrawn from appropriate statistical compari-sons.
A glance at the table shows that a signif-icant foot-in-the-door effect was not alwaysfound. It is unwarranted, however, to declarethat the effect cannot be reliably obtained.It must be noted that only a few studies(Cann, 1976; Cialdini & Ascani, 1976; Cial-dini, Cacioppo, Basset, & Miller, 1978; De-Jong & Funder, 1977, Study 1; Harris &Samerotte, 1976; Miller & Suls, 1977; Tipton& Browning, 1972; Dejong, Note 2) showedthe percentage of compliance with the secondrequest for one or more experimental groupsto be lower than or equal to that for the con-trol group. Thus, almost all of the failures toreplicate were in the predicted direction, butdid not reach traditional levels of statisticalsignificance. And as noted previously, twofailures to replicate (Button & Lennox, 1974;Harris, Liguori, & Stack, 1973, Study 3) aredifficult to interpret because of the very lowlevels of compliance with the first request inthe experimental group.
The subsequent discussion of many of thesestudies reveals other potential explanationsfor failures to replicate. They can be men-tioned here briefly. First, the initial requestmust be large enough to cause people to thinkabout the implications of their own behavior;if the request is too small, the effect will prob-ably not be obtained (Seligman, Bush, &Kirsch, 1976). Thus, for example, it may bethat a request to take a small card is toosmall to increase compliance with a high-costrequest like donating blood (Cialdini & As-cani, 1976). Moreover, some evidence sug-gests that if the initial request is too large,the probability of obtaining the effect will be
reduced (Baron, 1973; Miller & Suls, 1977).Second, people must feel that their initialcompliance resulted from the exercise of freechoice, not because of pressure to comply(Uranowitz, 197S; Zuckerman, Lazzaro, &Waldgeir, in press). Fish and Kaplan's(1974) findings are difficult to interpret forthis reason. Subjects in their experiment werefirst asked to write an essay on various waysof fighting poverty, but they were asked to dothis as part of a class about the poverty pro-gram. It is possible, then, that the subjectsmay have felt pressured into writing theessay because they saw it as a class assign-ment.
A key issue of concern is the generality ofthe foot-in-the-door phenomenon. Under howwide a range of situations has it been demon-strated? As Table 1 reveals, most of the re-quests that have been made of subjects in-volved little time or effort. But some did in-volve major commitments on the part of thesubjects. For example, Lowman (1973) usedthe foot-in-the-door technique to increasehousehold participation in a complicated trashrecycling program. In one study, Freedmanand Fraser (1966, Study 1) induced house-wives to allow a survey team of five or sixmen to come into their homes to classify thehousehold products they used.
The length of the delay between the tworequests has also been varied. The foot-in-the-door effect has been demonstrated with delaysbetween the two requests of up to two weeks(e.g., Freedman & Fraser, 1966, Study 2;Harris, 1972, Study 2), but most successfulreplications have involved much smaller de-lays of a day or two. The effect of delay hasnever been systematically explored, save Cannet al.'s (197S, Study 1) comparison betweendelay (7-10 days) and no-delay procedures(see Table 1). The length of the delay itselfmay be less important than whether the occa-sion of the second request somehow remindspeople of their earlier behavior.
The degree of similarity between the tworequests is an important dimension as well.Unfortunately, most investigators have em-ployed requests that share many common fea-tures, so that the robustness of the foot-in-the-door phenomenon has rarely been put to
2224 WILLIAM DEjONG
Table 1Replications of the Foot-in-the-Door Effect
Study First request Second request Results"Con-
clusion
Baer, Goldman, &Juhnke (1977)
Baron (1973)
Cann (1976)
(1) Give the time toexperimenter
(2) Same as (1); latermisinformation given todifferent experimenter
(1) Accept leaflet on thedangers of pollution
(2) Sign antipollution peti-tion, get two friends tosign, and mail in
(1) Agree to receive aquestionnaire on recycl-ing in the mail and fill itout
Correct misinformation given to (1) 70experimenter by another (2) 35elevator passenger (C) 33
Agree to put 3 foot X 5 footantipollution sign in frontyard
Volunteer time for a neighbor-hood cleanup project
(1) 50>>(2) ?(Q20
(1) 26(Q26
FITDNo FITD
FITDNo FITD
No FITD
Cann, Sherman,& Elkes (1975,Study 1)
Cann, Sherman,& Elkes (1975,Study 2)
Cialdini & Ascani(1976)
(1) Answer three questionson driving habits; nodelay between the tworequests
(2) Same as (1); 7-10-daydelay between requests
(1) Answer three questionson driving habits; nodelay between the tworequests
(1) Take and display smallcard advertising blooddrive
Agree to accept 15 pamphlets (1) 78 FITDon traffic safety and dis- (2) 70 No FITDtribute to neighbors (C) 50
Agree to accept 15 pamphlets (1) 72 No FITDon traffic safety and distri- (C) 45bute to neighbors
Agree to donate blood the next (1) 32 No FITDday (C) 32
Cialdini, Cacioppo,Basset, & Miller(1978, Study 2)
Crano & Sivacek(Note 1, Study1)
Crano & Sivacek(Note 1, Study2)
Dejong (Note 2)
(1) Agree to display aUnited Way windowposter
(1) Answer 10 questionsabout beverages
(1) Answer 10 questions onhousehold products
Agree to pick up a United Wayposter packet at dormitorylobby
Agree to answer 30 questionson driving habits
Agree to answer 45 questionson the mass media
Notify a second experimenterthat he dropped a quarter
Dejong & Funder(1977, Study 1)
(1) Agree to sign a petitionfor pro-disabled legisla-tion ; learn they were oneof many to sign
(2) Same as (1); learn theywere the first to sign
(1) Answer 15 questions on Agree to answer 50 questionsthe quality of life in the on highway safetylocal community
(1) 70°(Q70
(1) 66(C)31
(1) 56(C)32
(1) 64(2) 28(C)32
(1) 46(C) 56
No FITD
FITD
No FITD
FITDNo FITD
No FITD
FOOT-IN-THE-DOOR EFFECT 2225
Table 1 (continued)
Study First request Second requestCon-
Results" elusion
Dejong & Funder(1977, Study 2)
(1) Answer 15 questions on Agree to answer 50 questionsquality of life in the on highway safetylocal community; receiveletter acknowledgingparticipation
(1) 66(C)56
No FITD
Dejong & Musilli(Note 3)
(1) Agree to participate ina 5-minute survey onparking facilities forcompact cars; experi-menter appearedphysically normal
(2) Agree to participate ina 5-minute survey onparking facilities for dis-abled drivers; experi-menter appeared phys-ically normal
Agree to participate in a 30- (1) 55minute telephone survey on (2) 53highway laws and driving (C) 40hazards
FITDNo FITD
Dutton & Lennox(1974)
Fish & Kaplan(1974)
(1) Give money to a whitepanhandler
(1) Write a short essay onways of fighting poverty
Agree to donate time to various (1) 54.8d No FITDactivities as part of an inter- (C) 46.2racial Brotherhood Week
Volunteer time and services to (1) 36 No FITDa welfare agency (C) 33
Freedman & Fraser(1966, Study 1)
(1) Answer eight questionson household soaps
(2) Agree to be in surveyon households soaps
Agree to allow six-man survey (1) 53team to enter home and (2) 33spend 2 hours classifying all (C) 22household products
FITDNo FITD
Harris (1972,Study 1)
(1) Give directions(2) Give the time
Give the experimenter a dime (1) 39(2) 44
No FITDFITD
Harris (1972,Study 2)
(1) Write a letter to aminority high school stu-dent, indicating willing-ness to answer questionsabout the university andstudent life
Sign class list to volunteer (1) 18time to a university publicity (C) 9campaign
FITD
Harris, Liguori, &Stack (1973,Study 3)
Harris & Samerotte(1976, Study 1)
(1) Allow name to be sentto local congressman assupporter of organiza-tion's programs
(1) Watch experimenter'spossessions; a theft at-tempt is later thwartedby the subject
(2) Same as (1); the secondrequest is made by adifferent experimenter
Agree to donate money orcookies to a fund-raisingbaked-cookie sale
Give money to experimenter topermit the purchase of food
(1) Listen to a 2-minute,pro-McGovern speech;agree to display a smallcampaign sign
(2) Agree to display a smallcampaign sign only
(3) Listen to a 2-minutespeech on fire prevention,agree to display a smallfire prevention sign
(4) Agree to display asmall fire preventionsign only
Second request
Give money to experimenter topermit photocopying of anarticle
Agree to participate in a glassand metal trash recyclingprogram
Help a male experimenter pickup dropped groceries
Contribute money to the fund
Agree to participate in a 20-question survey on householdproducts
Agree to answer 55 more ques-tions for the same survey
Agree to display a McGovernposter in front window
Results"
(1) 14(2) 29(3) 50(4) 21(C) 29
(1) 85(C)68
(1) 12(2) 64(C)32
(1) 74(2) 81(C)46
(1) 75(C)S8
(1) 38(2) 35(3) 74(4) 74(C)31
(1) 38(2) 23(3) 30(4) 31(C) 16
Con-clusion
No FITDNo FITDNo FITDNo FITD
FITD
No FITDFITD
FITDFITD
No FITD
No FITDNo FITDFITDFITD
FITDNo FITDNo FITDNo FITD
Snyder & Cun-ningham (1975)
(1) Agree to answer 8questions on householdpaper products or ontraffic safety
Agree to answer 30 questions (1) 52for the other organization (C) 33
No FITD
FOOT-IN-THE-DOOR EFFECT 2227
Table 1 (continued)
Study
Tipton & Brown-ing (1972)
Uranowitz (1975)
Zuckerman, Laz-zaro, & Waldgeir(in press)
First request
(1) Help an elderly womanpick up droppedgroceries
(1) Watch experimenter'sshopping bags while heretrieves a dollar bill
(2) Watch experimenter'sshopping bags while heretrieves his wallet
(1) Agree to participatein a S-minute survey ontraffic safety
Second request
Help a young woman in awheelchair up over a curb
Notify a second experimenterthat she dropped her package
Agree to participate in a 20-minute survey on householdproducts
Results"
(1) 0'(C)36
(1) 80(2) 45(C)3S
(1) 64(C)45
Con-clusion
No FITD
FITDNo FITD
No FITD
Note. The first condition in a study is indicated by (1), the second condition by (2), and so on; results aredescribed accordingly; (C) = the no-initial-request control group; FITD = the foot-in-the-door effect.0 Results are reported as the percentage of subjects complying with the second request. A chi-square test(corrected for continuity) was executed for comparisons between individual experimental groups and thecontrol group. If p < .10, it is concluded that the foot-in-the-door effect was successfully replicated. If achi-square could not be executed based on the information provided by the authors, the conclusions basedon the authors' own statistical comparisons are reported.b Data reported are for subjects run by the male experimenters. Percentage compliance for subjects run bythe female experimenters is not reported, but no FITD effect was found. For Condition 2 (male experi-menters), p > .10 by chi-square, according to Baron (1973).0 Data reported are for verbal compliance with the request. Actual compliance was as follows: (1) 10%;(C) 20%.d Data are reported in terms of average summed difficulty level of volunteered activities; data on percentagecompliance are not reported. In a previous experimental session, all subjects run in this study were givenfalse physiological feedback, which led them to believe they might harbor racial prejudice.e Data have been collapsed across the variable of type of food to be purchased by the experimenter.' Only those who helped the elderly woman were included as subjects in the experimental condition. Despitethe bias toward confirmation this procedure created, no foot-in-the-door effect was found.
a severe test. For example, most studies thatask the subjects to participate in a survey asthe initial request also seek their participa-tion in a second survey. As noted previously,Freedman and Fraser (1966, Study 2) actu-ally manipulated the degree of similarity be-tween the two requests in terms of the issueinvolved and the mode of action required ofsubjects, but they found those factors to havelittle impact on compliance with the secondrequest. Seligman, Miller, Goldberg, Gelberd,Clark, and Bush (1976) also failed to findgenerally higher compliance when the tworequests were similar.
Self-Perception Explanation
Freedman and Fraser (1966) were sur-prised to learn that the size of the foot-in-the-door effect they obtained did not depend on
the nature of the first request. As noted be-fore, possible explanations for the effect cen-tering around subjects' commitment to a par-ticular experimenter, organization, issue, ormode of action were effectively ruled out. Toaccount for their unexpected results, theyoffered the following explanation:
What may occur is a change in the person's feelingsabout getting involved or about taking action. Oncehe has agreed to a request, his attitude may change.He may become, in his own eyes, the kind of personwho does this sort of thing, who agrees to requestsmade by strangers, who takes action on things hebelieves in, who cooperates with good causes. (Freed-man & Fraser, 1966, p. 201)
Thus, the self-perception mediation of thefoot-in-the-door phenomenon involves a two-stage process. First, people must observe theirown behavior and the situational context in
2228 WILLIAM DEjONG
which it occurs and, from those data, makean inference about their own dispositions andattitudes. It has been argued that subjectscomply with the initial request in the ap-parent absence of any external inducementsand then decide on the basis of that evidencethat they are the kind of people who cooper-ate with good causes or help out otherpeople.1 Second, this changed self-perceptionis thought to increase the probability of theirperformance of similar kinds of action in thefuture. As presently articulated, self-percep-tion theory does not adequately explain theprocesses that mediate this last step (cf. Bern,1972).
The self-perception explanation of the foot-in-the-door phenomenon has led to the formu-lation of several testable hypotheses. Theevidence for each of these is now considered.
Size of the Initial Request
A straightforward prediction derived fromself-perception theory is that the larger thesize of the first request to which people agree,the greater the probability of their complianceto the second larger request. In other words,the greater the costs of compliance with theinitial request, the more likely it should befor people to find dispositional meaning intheir behavior. To test this hypothesis, Selig-man, Bush, and Kirsch (1976) first askedsubjects to answer either 5, 20, 30, or 45short questions for a survey on reactions tothe energy crisis and inflation. There were nodifferences among those four conditions withregard to compliance with the first request forhelp. Later, subjects were called back by adifferent experimenter representing the samesurvey group and were asked to answer 55more questions. Unexpectedly, subjects in the5- and 20-question groups were not morelikely to answer the 55 questions than werecontrol subjects contacted for the first time.However, the two larger initial requests wereeffective in inducing significantly higher ratesof compliance with the second request.
Seligman, Bush, and Kirsch recognized thepossibility that sufficiently large initial re-quests might discourage subsequent compli-ance, but evidence for that was not found in
their study. It should be noted that subjectsin all cases were told the survey would requireonly "a couple of minutes," thus making eventhe larger requests seem relatively small. Astudy by Miller and Suls (1977) suggests thepossibility of a curvilinear relationship be-tween size of the initial request and subse-quent compliance. Subjects in their experi-ment were asked to give street directions to afirst experimenter, the directions being eitherdifficult or simple to explain. Compared withcontrol group subjects, those who gave thesimple directions were more likely to help asecond experimenter pick up his droppedgroceries a few moments later. Those whogave the difficult directions were less likelyto help. Existence of a curvilinear relation-ship, of course, could not be explained by self-perception theory alone.
Agreement to the Request VersusActual Performance
It could also be predicted from self-percep-tion theory that compliance with the secondrequest would be more likely if subjects actu-ally performed what was asked of them in theinitial request than if they agreed to complybut were never called upon to carry out theirpromise. The only study that directly ad-dressed this issue is an experiment reportedby Freedman and Fraser (1966, Study 1).Housewives in the so-called agree-only condi-tion were asked to participate in a consumersurvey about household products; those will-ing to help were told the survey would be con-ducted at a later time. Subjects in the per-formance condition actually participated inthe survey. In a later call, the same experi-menter asked subjects if they would permit a
1 Self-perception theory has not articulated wellthe exact nature of the self-perception changes thatare said to occur. It is not clear, in the case of mostfoot-in-the-door studies, whether subjects classifytheir behavior as compliance or as helping. This un-certainty is due to the fact that most experimentersuse compliance with a request for help as their de-pendent measure (see Table 1). Experiments thatinstead require subjects to initiate help giving (e.g.,Miller & Suls, 1977; Uranowitz, 1975) suggest thatmost subjects may view their assent to the firstrequest as helpfulness.
FOOT-IN-THE-DOOR EFFECT 2229
survey team to come into their homes to listthe household products they used. Whereas22% of the control group subjects agreed tothat request, over half the subjects in theperformance condition did so. In contrast, amere 33% of the agree-only condition sub-jects complied with that request. A cleartheoretical interpretation of these results ismade difficult by the fact that subjects in theagree-only condition did not learn the exactsize of the request to which they had agreed.Their compliance to the second request mayhave been lower only because they guessedthe initial request to be quite large (cf. Miller& Suls, 1977). Parenthetically, it must benoted that other experimenters have shownthat actual performance of the initial requestis not requisite in demonstrating the foot-in-the-door effect (e.g., Snyder & Cunningham,1975; Zuckerman et al., in press).
Effect oj Noncompliance With theInitial Request
Self-perception theory led Snyder and Cun-ningham (1975) to predict that subjects in-duced not to comply with an initial requestwould be less likely to comply with a secondrequest. In their experiment, one group ofsubjects was first asked to participate in a50-question telephone survey, a request suf-ficiently large to guarantee almost universalnoncompliance. Two days later, a different-sexed experimenter representing a secondorganization asked subjects to answer 30 ques-tions. Consistent with 'their prediction, sub-jects initially approached with the large re-quest were less compliant with the secondrequest than were subjects in the controlgroup. Similar results were reported by Cannet al. (1975) and Reingen and Kernan(1977). These studies support the idea thatinduction of initial noncompliance leads peo-ple to perceive themselves as the kind ofpeople who do not comply with such requests.One might predict that the smaller the re-quest that people refuse, the lower the prob-ability of their compliance to a second re-quest; this proposition remains to be tested.
An added complication arises when the tim-ing of the second request is considered. In
two studies, Cann et al. (1975) reported thatnoncompliance with a large initial requestactually led to greater compliance when thesecond request was made immediately afterthe first, rather than after a delay period ofseveral days. For example, one study showedthat 90% of the subjects agreed to an im-mediate second request after having refusedthe experimenter's initial large request, com-pared with 50% in the control group. Only29% compliance was obtained when that sec-ond request was made 7-10 days later.
These findings replicate an effect first dem-onstrated by Cialdini et al. (1975), whichthey dubbed the "door-in-the-face" technique.Quite simply, these authors claimed that ifan experimenter first approaches subjectswith an unreasonable request that is sure tobe refused, but then immediately asks asmaller favor, subjects will view the experi-menter's action as a concession. The subjects,in turn, will feel normative pressure to recip-rocate that concession, and they will respondto that pressure with compliance with thesecond request. Replications of this effectabound (Cialdini & Ascani, 1976; Miller,1974; Miller, Seligman, Clark, & Bush, 197,6;Reingen, 1977).
In one experiment, Cialdini et al. (1975)demonstrated that the subjects' perceptionof the encounter as a kind of bargaining situa-tion probably underlay the effect. In the re-jection-moderation condition, both requestsfor help were made by the same experimenter.In the so-called two-requester condition, sub-jects' noncompliance with the first requestwas followed immediately by another requestput to them by a second experimenter. Thissecond experimenter was seemingly unrelatedto the first, but subjects knew this person hadoverheard their conversation with the firstexperimenter. Whereas a control procedureproduced a 31% compliance rate, 55% of therejection-moderation condition subjects con-sented to the second request. But when thatsame second request was made by a differentexperimenter, only 10% of the subjects inthe two-requester condition complied, a re-sult consistent with those reported by Snyderand Cunningham (1975). The authors arguedthat subjects in the latter condition could
2230 WILLIAM DsJONG
not interpret the second request as a conces-sion, simply because it was made by a differ-ent experimenter.2
Why do subjects in the door-in-the-faceprocedure who do not comply with the initiallarge request not come to see themselves asthe kind of people who refuse such requests,as Snyder and Cunningham's (197S) resultsmight lead us to expect? Even-Chen, Yinon,and Bizman (1978) argued that the initialrequests used by Cialdini et al. (1975) andothers were too large for subjects to drawdispositional inferences from their failure tocomply. Implicit in their argument is the sug-gestion that subjects must believe the requestis one to which someone (though not they)might consent in order for a self-perceptionchange to occur. In their study, when subjectsrefused a large, but not reasonable, first re-quest, they were less likely to agree to a sec-ond smaller request. Only those who refusedan extremely large request showed a subse-quent door-in-the-face effect. These resultsare suggestive, but Even-Chen et al.'s argu-ment cannot account for the results obtainedby Cialdini et al. In the earlier experiment,whether a door-in-the-face effect or a resultconsistent with self-perception theory wasfound depended solely on who made the sec-ond request.
In general, research on the effect of initialnoncompliance has been supportive of theself-perception hypothesis. Only under a spe-cial set of circumstances, that is, when thefirst request is extremely large and the secondrequest can be viewed as a concession on thepart of the experimenter, does an initial re-fusal not lead to subsequent refusals.
Effect of Consensus Information
The self-perception analysis of the foot-in-the-door effect leads to the prediction that ifsubjects were informed they were not uniquein their compliance with the initial request,they would be induced to see their behavioras determined by situational pressures andnot to view it as having any implications fortheir own traits or attitudes (cf. Cook, Pallak,& Sogin, 1976; Cooper, Jones, & Tuller,1972). In one recent experiment designed to
test this prediction (Dejong, Note 2), shop-pers were approached individually by a phys-ically able male experimenter and shown acard on which was written the resolution of apetition formulated by a group called theUpper Valley Foundation to Help the Handi-capped. After reading the resolution, 87% ofthe subjects agreed to sign the petition. Atthat point, some subjects were shown a peti-tion with many signatures and one blankspace at the bottom (consensus condition).As they signed, the experimenter said, "As youcan see, almost everyone we've asked hassigned our petition." The other subjects wereshown a completely blank petition, and asthey signed, the experimenter noted that theywere the only ones so far to do so (noncon-sensus condition). When the subjects walkedon, a second male experimenter dropped aquarter as he walked some IS to 20 feet aheadof them. Thirty-two percent of the subjects inthe no-initial-request control group notifiedthe second experimenter of his loss, whereas64% of those told that many people hadsigned the petition did so. Unexpectedly, only2&% of those told they had been unique intheir agreement to sign helped the secondexperimenter.
Consensus information is more psycholog-ically complex than attribution theorists haveadmitted (cf. McArthur, 1973). In additionto pointing to the power of situational con-straints on behavior (Kelley, 1967), consen-sus information can help define the behavioralnorms that are operative in a specific context.Subjects in the consensus condition may havehelped the second experimenter more oftenbecause they were reminded that helpingothers was appropriate behavior. In addition,
2Pendleton and Batson (1979) offered a self-pre-sentation explanation of the door-in-the-face tech-nique, suggesting that subjects are motivated to com-ply with the second request to avoid being perceivedas unhelpful. Although the results for Cialdini et al.'s(1975) two-requester control group seem to counterthis alternative, Pendleton and Batson argued thatthe phrasing of the second experimenter's request("But maybe you could help me") may have in-duced reactance, thus producing a low rate of com-pliance in that condition. Further research is neededto resolve this controversy.
FOOT-IN-THE-DOOR EFFECT 2231
consensus information can reaffirm the rea-sonableness of one's own behavior. Subjectsin the nonconsensus condition may have beenshocked to learn of their unique status andmay literally have been "lost in thought"when the second experimenter dropped hisquarter. To date, no foot-in-the-door studyinvolving the introduction of consensus infor-mation has yielded results consistent withself-perception theory.
Presence of External Justification
Theoretical analyses of attributional pro-cesses (Kelley, 1967) propose that peopleassign dispositional meaning to behavior aftera careful assessment of the possible explana-tory power of controlling influences in theenvironment. Thus, a self-perception accountof the foot-in-the-door phenomenon stressesthat the amount of external pressure used toinduce compliance with the initial request isparamount. If those external pressures pro-vide people with an adequate explanation oftheir behavior, they will not be led to inferanything about their own traits or attitudes.Under those circumstances, an increased prob-ability of compliance with subsequent re-quests for help would not be expected. In-deed, Lepper (1973) even suggested the pos-sibility of an "overjustifkation" effect; thatis, if people comply with an initial requestunder conditions of strong external justifica-tion, they may come to infer that they areextrinsically motivated and that they are thekind of people who agree to such requestsonly when external pressures are present.As a result of this kind of changed self-per-ception, people may actually come to be lesslikely to comply with future requests whenthose kinds of pressures are absent.
The impact of external justifications for theinitial act of compliance on subsequent help-ing was demonstrated by Uranowitz (1975).Female shoppers were asked by an experi-menter to watch his grocery bags while he ranback to retrieve a lost article. In the high-justification condition, he claimed he had losta wallet containing a lot of money; in thelow-justification condition he said that he haddropped only a dollar bill. The assumption,
of course, was that subjects in the high-justi-fication condition would believe that circum-stances demanded their compliance and thatanyone faced with such a request would agreeto it. When the first experimenter returnedwith the lost article, the subject proceeded onher way. A female experimenter later droppeda package in the subject's path, and the sub-ject's response was noted. The results werestriking. Eighty percent of the subjects in thelow-justification condition helped the secondexperimenter, whereas only 45% of the high-justification and 35% of the control subjectsdid so.
Zuckerman et al. (in press) investigatedthis same problem using an experimental pro-cedure modeled after that used by Snyder andCunningham (1975). Housewives were firstasked to take part in a 5-minute telephonesurvey; some were promised a monetary pay-ment in exchange for their cooperation, andothers were not, If a subject agreed to par-ticipate, she was told that the interview wouldbe conducted at a later time. Subjects prom-ised the monetary payment were told theywould receive a check after the interview.Two or three days later, the subjects werecalled by a second experimenter representinga different service organization and wereasked to consent to a 20-minute interview.Forty-five percent of the control group sub-jects agreed to that request, whereas 64%of the subjects promised no money consentedto the interview. In contrast, only 33% ofthose promised a monetary payment in returnfor their initial compliance agreed to the sec-ond request. Similar results were reported byReingen and Kernan (1977).
Dejong and Funder (1977), however,found the monetary payment to have the op-posite effect. The day after their participationin a 15-question survey, subjects in the pay-ment condition received the $2 that had beenpromised them in exchange for their help.Subjects were later asked to take part in a50-question survey being conducted by a sec-ond organization. Fifty-six percent of controlsubjects never before contacted agreed to thesecond request, whereas 78% of the subjectsin the payment condition did so. These find-ings were unexpected. It was thought that the
2232 WILLIAM DEjONG
actual receipt of money by subjects in thepayment condition would increase the salienceof the external justification for their compli-ance with the first request. A follow-up studyshowed that this finding was not due to sub-jects' expectation that they would be paid forhelping with the second survey. When thesecond caller informed subjects they could notbe paid for their help, the pattern of resultswas virtually unchanged. One possible ex-planation for this unexpected set of results isthat a letter that accompanied the paymentmay have labeled subjects as "doers" (cf.Kraut, 1973). That certification may havehad greater implications for subjects' self-perceptions than did the small monetary pay-ment.
Finally, in a study conducted by Dejongand Musilli (Note 3), subjects were ap-proached at home with a small initial requestmade by a female experimenter who appearedto be either handicapped or physically normal.It was hypothesized that subjects approachedby the disabled experimenter would feel psy-chological pressure to comply with her requestand would, in turn, attribute their complianceto the fact of her disability. The experimenterasked half the subjects to answer questionsfor a survey on special parking facilities fordisabled drivers for a group called Friends ofthe Handicapped. The others were told thatthe experimenter represented Friends of theEnvironment and that the questions con-cerned special parking for compact cars. Twodays later, subjects were called on the tele-phone by a second experimenter and wereasked to participate in a 30-minute survey ontraffic safety for a different organization. Thehypothesis was partially supported. Fifty-sixpercent of the subjects approached by thehandicapped experimenter for Friends of theHandicapped complied with the second re-quest, whereas 40% of control subjects did so.However, only 41% of subjects approachedby the handicapped experimenter for Friendsof the Environment complied with the secondrequest.
In sum, evidence is generally consistentwith the proposition that high external justi-fication for initial acts of compliance can dam-pen the probability of subsequent compliance.
Only one study found no such effect at all(Dejong & Funder, 1977), and there is aplausible explanation for that failure. Littleevidence for an over justification effect exists;only Reingen and Kernan (1977) and Zuck-erman et al. (in press) reported results for ahigh-justification condition to be actually be-low those for a control group, and these differ-ences were small. One possible reason for notfinding overjustification effects is that subjectsin these studies were not preselected for theirhigh level of initial intrinsic or altruistic mo-tivation.
There may be circumstances under whichhigh external justification for initial compli-ance actually leads to higher levels of subse-quent compliance. Such an outcome might beanticipated if people believe that behaviorthat is truly intrinsically motivated is morehighly regarded than that which is thought toarise in response to external pressure (cf.Nemeth, 1970). Thus, when subjects' self-esteem (or their good name) is threatened bythe perception that they have complied in re-sponse to the dictates of external contin-gencies, they may be motivated to be morehelpful in the future when such pressures areabsent. A report by Upton (1973) is con-sistent with this notion. He found that pre-vious blood donors classified as intrinsicallymotivated did not respond well to an appealfor donations that offered money in exchangefor a pint of blood. When no such bribe wasoffered, the percentage of such subjects will-ing to donate was almost 30 percentage pointshigher. It is conceivable that subjects offeredthe bribe wished to avoid the perception thatthey were motivated to help others only as ameans of garnering rewards for themselves.
Effect of Social Labels
Sociological theories of deviance stress thatsociety encourages those whom it labels asdeviant to learn and accept a deviant roleidentity. Once such people come to share thisdefinition of themselves, the changed self-image is believed to sustain their deviant be-havior (see Schur, 1971). Although thesetheories focus almost exclusively on labels ofdeviance and generally ignore positively val-
FOOT-IN-THE-DOOR EFFECT 2233
enced labels, their basic propositions can stilleasily be translated into the language of self-perception theory. Though actual behavior(and the context in which it occurs) may pro-vide the clearest evidence about one's traitsor attitudes, self-perception theory recognizesthat a self-definition or label provided byothers may serve as an important source ofinformation about one's dispositions (cf.Miller, Brickman, & Bolen, 1975) or at leastmay signal to people that their behavior andits implications for their self-image should bereflected on, thereby energizing a self-percep-tion analysis.
This proposition was put to the test byKraut (1973). First, a male experimenterwent to subjects' homes and solicited con-tributions for the Heart Association. If sub-jects made a donation, they were randomlyassigned to be in a labeled or a nonlabeledcondition. Subjects receiving the charitablelabel were simply told, "You are a generousperson. I wish more of the people I met wereas charitable as you." Subjects who did notcontribute were also assigned to a labeled or anonlabeled condition. Subjects receiving thelabel were frankly told that they were "un-charitable." One or two weeks later, a secondexperimenter contacted the subjects and askedfor a donation to a fund-raising drive formultiple sclerosis. (Unfortunately, a controlgroup consisting of subjects receiving only thesecond solicitation was not run.) As Krautpredicted, 62% of those given the positivelabel donated the second time, whereas only47% of the nonlabeled subjects did so. Krautrecognized that this effect could be explainedin terms of a social reinforcement model ifthe charitable label constituted a reward forthe subjects. He also found that thos« who didnot donate the first time gave somewhat lesswhen they had been branded as uncharitableby the first experimenter, but this differencewas not significant.
The impact of labels on self-perception andsubsequent behavior was also examined in astudy conducted by Paulhus, Shaffer, andDowning (1977). Prior to their actual dona-tions, blood donors were asked to read com-munications that were designed to label theirmotives for giving blood. Half the subjects
read a communication that emphasized altru-istic motives, and as a cross-dimension, halfthe subjects were told about the personalbenefits they would receive (e.g., free bloodin case of emergency). After donating, sub-jects filled out a questionnaire that includedan item on their plans for giving again withinthe next year. The results showed a maineffect for salience of altruistic motives, suchthat subjects led to feel they had acted altru-istically reported a greater likelihood of fu-ture donation.
Me Arthur, Kiesler, and Cook (1969)labeled some subjects as doers (people whoknow what needs to be done and then takethe appropriate action) on the basis of bogustest results. This positive label increased theproportion of subjects willing to help dis-tribute antipollution leaflets only when sub-jects had been told that their doer personalityqualified them to be paid for their participa-tion in a second experiment. Apparently, sub-jects' belief that being a doer was a saleablequality strengthened the impact of the label,either because it made the feedback more con-vincing or simply because the subjectsthought about it more.
In contrast, Steele (1975) expected labelsto have effects different from those predictedby self-perception theory. He argued that anegative label (name-calling) that impugnedsubjects' character would motivate them totake action that would restore their self-esteem, thus increasing the probability thatthey would comply with a later request forhelp. Because positive labels would not dothis, Steele expected them to have little im-pact on help giving. In this study, an experi-menter claiming to be a pollster contactedsubjects by telephone. In the relevant-nega-tive-name condition, subjects were lambastedfor being "apathetic about the welfare ofothers," whereas subjects in the irrelevant-negative-name condition were criticized fortheir lack of concern for driving safety. Re-cipients of the relevant-positive-name messagewere praised for their desire to help theirfellow man. (No irrelevant-positive-name con-dition was run.) Two days later, subjectswere called by a second experimenter repre-senting a food cooperative in a lower income
2234 WILLIAM DEjONG
neighborhood. To help the cooperativeachieve its goal of "aiding the less fortunate,"subjects were asked to compile a detailed listof the quantities and brands of foods andhousehold items used in their homes. Com-pared with a control group never before con-tacted, subjects receiving the positive labelwere more likely to help, though not signif-icantly so (cf. Kraut, 1973). On the otherhand, virtually all subjects in both negative-name conditions promised to help out, andmost actually did so. This effect was repli-cated in a second study.
The name-calling procedure employed bySteele (1975) differs from the labeling proce-dure used by other researchers in one im-portant way. The labels used by others werebased either on recent behavioral evidence(Kraut, 1973) or on the results of phonypsychological tests (McArthur et al., 1969).In contrast, no such evidence substantiatedthe claim made by Steele's experimenter. Thissuggests that whether the negative label in-duces a change in self-perception may dependon the degree to which subjects' future be-havior can belie the label. If refutation of thelabel is still possible, it may not lead to achange in self-definition, but may lead to avigorous effort to defend self-esteem or toassuage guilt.
This possibility was explored in a recentstudy by Gurwitz and Topol (1978). Stu-dents at a large suburban university werefirst contacted by telephone and accused ofnot taking advantage of the opportunitiesafforded by a nearby city. This accusation wasdirected at subjects as members of a group(students at the university) or as individuals.Before the accusation was made, half the sub-jects were asked how many times in the pastfew months they had gone into the city, themodal answer being zero. The other subjectswere not led in this way to provide evidencein support of the accusation. Later that eve-ning, a second experimenter had subjects fillout a questionnaire concerning their interestin a number of student-organized activitiesin the city.
When subjects had not been led to provideevidence in support of the accusation, thedegree to which they later belied the label
depended on whether the accusation had beenmade about them as individuals or as mem-bers of a group. When they were seen as partof a group, they indicated more interest in theactivities, thus belying the label. In contrast,when the subjects had been led to providesuch evidence prior to the accusation, theywere more likely to disconfirm the label whenit had been directed at them as individuals.Similar results were found in a laboratorystudy conducted by Gurwitz and Topol(1978) in which subjects were labeled as be-ing low in self-confidence. Why subjects' re-sponses to the label depended so greatly onwhether it was applied to them individuallyor as members of a group is not clear.
In sum, although the effect of positivelabels has been fairly consistent across sev-eral studies, the impact of negative labels onsubsequent behavior presents a more compli-cated picture.
Measurement of Altered Self-Perceptions
Perhaps the most convincing evidence insupport of the self-perception explanation ofthe foot-in-the-door effect would come fromstudies that actually measured the changes inself-perception that are thought to occur. Un-fortunately, this kind of direct evidence ishard to come by. A long catalog of excusesfor this failure can be offered, most of thempointing to possible inadequacies in designand measurement (see Bern, 1972; Lepper,1973). But the major difficulty may be thatthe self-perception changes that follow a per-son's initial compliance with a small requesthave never been clearly specified and may bemuch more complicated than has beenacknowledged.
First, it may be unreasonable to expect thesubjects' general perception of themselves tobe altered by the kinds of brief experiencesinvolved in these foot-in-the-door studies. Infact, particular subjects may only come toinfer something about their attitudes or mo-tives for a fairly limited set of situations(e.g., telephone solicitations from strangers).This notion is consistent with recent state-ments of self-concept theory (Gergen, 1971;
FOOT-IN-THE-DOOR EFFECT 2235
McGuire & Padawer-Singer, 1976), whichemphasize that a person may harbor a varietyof self-definitions that differ in salience acrossvarious situation contexts.
Second, measurement of these supposedchanges in self-perception is made difficultby the fact that individuals are likely to codetheir behavior along different dimensions orinto different groupings ('Bern & Allen, 1974).Which behaviors constitute actions that arepsychologically similar to the initial act ofcompliance is a function of each person's viewof the world. To give one example, some per-sons might code their behavior as complianceor as participating in a survey, whereas othersmight code it as helping someone in need.
For these two reasons, investigators whohave hoped to show a general change in sub-jects' self-definitions seem to have had littlechance of success. Still, two experiments havesuccessfully demonstrated that extrinsic in-centives for help giving can lead helpers todescribe themselves as less altruistically mo-tivated. During the course of an experimenton "first impressions," Batson, Coke, Jas-noski, and Hanson (1978) asked male under-graduates to help an experimenter code data.Payment for this help was offered before sub-jects agreed to help (payment condition),after the subjects' agreement (paymentafter), or payment was not mentioned at all(no payment). The request for help was al-ways made in the presence of a male con-federate who never volunteered his services.A control group received no such request.Subjects were then asked to rate themselvesand the confederate on several dimensions, in-cluding helpfulness and cooperativeness. Asexpected, subjects who agreed to aid the ex-perimenter in exchange for money rated them-selves as less altruistic than the confederate,whereas subjects in the other three conditionsrated themselves as more altruistic.
In an experiment conducted by Smith, Gel-fand, Hartmann, and Partlow (1979), second-and third-grade children played a marbledrop game in which they could earn penniestoward the purchase of a prize. During thefirst part of the experiment, the children weregiven a number of opportunities, signaled bya light, to donate pennies to another child
playing the game in a nearby room. The ex-perimenter also created situations in whichthe children could not make a donation (butwould think they should have done so) byhaving the signal light go off before theycould respond. Whereas some subjects weremerely praised each time they donated apenny, others were given a monetary rewardfor doing so, and the relationship betweentheir help giving and the reward was spelledout to them. Other subjects were scolded bythe experimenter each time they did notavail themselves of an opportunity to help;another group of children were fined eachtime they failed to donate, and the contin-gency between their behavior and the punish-ment was made explicit. Finally, a controlgroup did not receive any type of reward orpunishment from the experimenter. Intensiveinterviews conducted by a second experi-menter with each subject showed that chil-dren who were rewarded or fined were morelikely to attribute their help giving to externalpressures than were children in the controlgroup or than were those who received praiseor a scolding.
Alternative Explanations for theFoot-in-the-Door Effect
Surprisingly few alternative explanationsfor the foot-in-the-door effect have been sug-gested. In part, this is because the self-per-ception analysis has proven to be a heuristicexplanation, leading investigators to examinea wide variety of phenomena such as over-justification effects and the impact of social,labels. Other explanations have not generatedmuch research, but they do resurface period-ically. Three of these are considered.
Adaptation Level
It has been argued that the small initialrequest to which people agree establishes anew baseline against which the subsequentlarger request is compared (e.g., Schmidt,1973). In other words, the prior requestcauses the magnitude of the second to be re-defined, making it seem less extreme than it
2236 WILLIAM DEjONG
otherwise would. It is not clear that this ex-planation would lead one to predict a differ-ence in subsequent compliance between thosewho refuse and those who agree to the initialrequest. The perceived relative size of thesecond request should not depend very muchon whether people have complied with thefirst request. However, several experimentsshow foot-in-the-door effects to be strongerwhen those who refuse the first request areexcluded from the data analysis (e.g., Snyder& Cunningham, 1975; Dejong & Musilli,Note 3). It should be noted that this explana-tion cannot easily account for the results ofstudies looking either at extrinsic justificationfor the initial act of compliance or at the im-pact of social labels.
To lay this explanation permanently torest, two types of studies could be done. First,subjects could be asked to rate the relativesize of various requests under differing experi-mental conditions. Second, the subsequentcompliance of those who are merely informedof a request being made of others and thosewho actually comply with that request couldbe compared. An adaptation level explanationwould predict higher compliance from both ofthose groups, compared with a control groupnever before contacted, but self-perceptiontheory would not.
Salience oj Social Norms
Harris (1972) suggested that being askedto perform an initial small request makespeople more aware of the norm of social re-sponsibility, a norm that prescribes that oneshould help those who are in need. Like theadaptation level explanation, this argumentalso leads to the prediction that those whoare only informed about the first request willbe more compliant with the second demand.It would also lead us to expect a high rate ofcompliance from those who refuse the initialrequest, in contrast with available evidence.The major problem with this explanation,however, is that it, too, cannot account forthe studies on extrinsic constraints or labels,which are better handled by the self-percep-tion analysis.
Behavioral Consistency
It has also been argued that foot-in-the-door demonstrations show that people try toact consistently with the way they have be-haved in the past. Does a person induced toagree to a first request agree to a second inorder to sustain a consistent public image?If the effect were due to subjects' attempts atsuch impression management, then a largerfoot-in-the-door effect should result when thesecond request is made by the same experi-menter or involves the same issue or action.The importance of these kinds of variableshas not been adequately tested, but the avail-able evidence suggests that similarity betweenthe requests on these dimensions is not ter-ribly important (e.g., Freedman & Fraser,1966).
A second version of this argument was sug-gested by Brock (1969), who argued, in es-sence, that people desire to be psychologicallyconsistent for the sake of a self-image, not apublic one. People have behaved in a certainway in the past and continue to behave thatway in the future simply because they valueconsistency. However, it is clear that to man-age a self-image in the fashion Brock sug-gested, people must first code their behaviorand understand its attributional meaning.They must decide, essentially, whether the be-havior reflects the pressures of extrinsic con-straints or reflects their own dispositions orattitudes. Of course, it may be that the self-rewarding consequences of being consistentwith one's self-image are what mediate therelationship between changes in self-percep-tion and subsequent behavior. Evidence onthis point is lacking.
Conclusions
1. It is concluded that the foot-in-the-dooreffect can be reliably obtained. Although thenumber of studies that have failed to demon-strate it unequivocally is surprising, it mustbe reemphasized that almost all of the studiesreviewed showed experimental results in thepredicted direction. Furthermore, a numberof plausible explanations for failures to repli-cate can be offered.
FOOT-IN-THE-DOOR EFFECT 2237
2. In addition to the self-perception ex-planation first offered by Freedman andEraser (1966), three alternative explanationsfor the effect have been suggested. It is con-cluded that these explanations are inadequateaccounts of the foot-in-the-door literature.
3. A number of theoretical derivations fromthe self-perception explanation of the foot-in-the-door effect have been outlined, and theevidence for most is found to be generallysupportive. However, investigators have notclearly specified the exact nature of the self-perception changes that are said to mediatethe foot-in-the-door effect. Self-perceptiontheory must concern itself with how peopleclassify or code their own behavior and howthey are led to form a broad or highly specificinference about their attitudes, traits, or dis-positions. Echoing Bern's (1972) admission,it must also be underscored that self-percep-tion theory does not fully explain the link be-tween self-attribution and subsequent behav-ior. Further work is needed to explore thesegaps in the self-perception analysis of thefoot-in-the-door effect.
Reference Notes
1. Crano, W. D., & Sivacek, J. Social reinforcement,self-perception and the foot-in-tke-door phenom-enon. Unpublished manuscript, 1978. (Availablefrom W. D. Crano, Department of Psychology,Olds Hall, Michigan State University, East Lans-ing, Michigan 48824.)
2. Dejong, W. Consensus information and the foot-in-the-door effect. Unpublished manuscript, 1978.(Available from W. Dejong, Department of Psy-chology, Brandeis University, Waltham, Massa-chusetts 02154.)
3. Dejong, W., & Musilli, L. Handicapped versusnon-handicapped requesters: The effect of pressureto comply with an initial request on the foot-in-the-door phenomenon. Paper presented at themeeting of the Eastern Psychological Association,Washington, D.C., March 1978.
References
Arbuthnot, J., et al. The induction of sustained re-cycling behavior through the foot-in-the-doortechnique. Journal of Environmental Systems,1976-1977, 6, 353-366.
Baer, R., Goldman, M., & Juhnke, R. Factors affect-ing prosocial behavior. Journal of Social Psychol-ogy, 1977, 103, 209-216.
Baron, R. A. Foot-in-the-door phenomenon: Mediat-ing effects of size of first request and sex of re-quester. Bulletin of Psychonomic Science, 1973, 2,113-114.
Batson, C. D., Coke, J. S., Jasnoski, M. L., & Han-son, M. Buying kindness: Effect of an extrinsicincentive for helping on perceived altruism. Per-sonality and Social Psychology Bulletin, 1978, 4,86-91.
Beaman, A. L., Svanum, S., Manlove, S., & Hamp-ton, C. An attribution theory explanation of thefoot-in-the-door effects. Personality and SocialPsychology Bulletin, 1974, 1, 122-123.
Bern, D. J. Self-perception theory. In L. Berkowitz(Ed.), Advances in experimental social psychology(Vol. 6). New York: Academic Press, 1972.
Bern, D. J., & Allen, A. On predicting some of thepeople some of the time: The search for cross-situational consistencies in behavior. PsychologicalReview, 1974, 81, 506-520.
Brock, T. C. On interpreting effects of transgressionupon compliance. Psychological Bulletin, 1969, 72,138-145.
Cann, A. A. The effects of performing a small con-sonant behavior on subsequent attitudes and be-havior (Doctoral dissertation, Indiana University,1975). Dissertation Abstracts International, 1976,36, 5859B. (University Microfilms No. 76-11,359)
Cann, A., Sherman, S. J., & Elkes, R. Effects of in-itial request size and timing of a second request oncompliance: The foot in the door and the door inthe face. Journal of Personality and Social Psy-chology, 1975, 32, 774-782.
Cialdini, R. B., & Ascani, K. Test of a concessionprocedure for inducing verbal, behavioral, andfurther compliance with a request to give blood.Journal of Applied Psychology, 1976, 61, 295-300.
Cialdini, R. B., Cacioppo, J. T., Basset, R., & Miller,J. A. The low-ball procedure for producing com-pliance: Commitment then cost. Journal of Per-sonality and Social Psychology, 1978, 36, 463-476.
Cialdini, R. B., et al. A reciprocal concessions proce-dure for inducing compliance: The door-in-the-face technique. Journal of Personality and SocialPsychology, 1975, 31, 206-215.
Cook, D. A., Pallak, M. S., & Sogin, S. R. The effectof consensus on attitude change and attributionof causality. Personality and Social PsychologyBulletin, 1976, 2, 248-251.
Cooper, J., Jones, E. E., & Tuller, S. M. Attribution,dissonance, and the illusion of uniqueness. Journalof Experimental Social Psychology, 1972, 8, 45-47.
Dejong, W., & Funder, D. Effect of payment forinitial compliance: Unanswered questions aboutthe foot-in-the-door phenomenon. Personality andSocial Psychology Bulletin, 1977, 3, 662-665.
Dutton, D. G., & Lennox, V. L. Effect of prior"token" compliance on subsequent interracial be-havior. Journal of Personality and Social Psy-chology, 1974, 29, 65-71.
Even-Chen, M., Yinon, Y., & Bizman, A. The doorin the face technique: Effects of the size of the
2238 WILLIAM DsJONG
initial request. European Journal of Social Psy-chology, 1978, 8, 135-140.
Fish, B., & Kaplan, K. J. Does a "foot-in-the-door"get you in or out? Psychological Reports, 1974, 34,35-42.
Freedman, J. L., & Fraser, S. C. Compliance withoutpressure: The foot-in-the-door technique. Journalof Personality and Social Psychology, 1966, 4, 195-202.
Gergen, K. J. The concept of self. New York: Holt,Rinehart & Winston, 1971.
Gurwitz, S. B., & Topol, B. Determinants of con-firming and disconfirming responses to negativesocial labels. Journal of Experimental Social Psy-chology, 1978, 14, 31-42.
Harris, M. B. The effects of performing one altruisticact on the likelihood of performing another. Jour-nal of Social Psychology, 1972, 88, 65-73.
Harris, M. B., Liguori, R., & Joniak, A. Aggression,altruism, and models. Journal of Social Psychology,1973, 91, 343-344.
Harris, M. B., Liguori, R. A., & Stack, C. Favors,bribes, and altruism. Journal of Social Psychology,1973, 89, 47-54.
Harris, M. B., & Samerotte, G. The effects of ag-gressive and altruistic modeling on subsequent be-havior. Journal of Social Psychology, 1975, 95,173-182.
Harris, M. B., & Samerotte, G. The effects of actualand attempted theft, need, and a previous favoron altruism. Journal of Social Psychology, 1976,oo, 193-202.
Kelley, H. H. Attribution theory in social psychol-ogy. In D. Levine (Ed.), Nebraska Symposium onMotivation (Vol. 15). Lincoln: University ofNebraska Press, 1967.
Kraut, R. E. Effects of social labeling on giving tocharity. Journal of Experimental Social Psychology,1973, 9, 551-562.
Lepper, M. R. Dissonance, self-perception, and hon-esty in children. Journal of Personality and SocialPsychology, 1973, 24, 65-74.
Lowman, R. P. Recycling refuse: The effect of thefoot-in-the-door technique on attitude and repeti-tive behavior (Doctoral dissertation, ClaremontGraduate School, 1972). Dissertation Abstracts In-ternational, 1973, 33, 6114B. (University Micro-films No. 73-14,258)
McArthur, L. A. Perceiving the cause of one's ownbehavior: An unexpected reversal of Kelley's at-tribution theory. Psychological Reports, 1973, 32,983-988.
McArthur, L. A., Kiesler, C. A., & Cook, B. P. Actingon an attitude as a function of self-concept andinequity. Journal of Personality and Social Psy-chology, 1969, 12, 295-302.
McGuire, W. J., & Padawer-Singer, A. Trait saliencein the spontaneous self-concept. Journal of Per-sonality and Social Psychology, 1976, 33, 743-754.
Miller, R. L. Facilitating compliance by manipulat-ing the nature of the comparison: Relative cost vs.
reciprocal concession. Personality and Social Psy-chology Bulletin, 1974, 1, 160-162.
Miller, R. L., Brickman, P., & Bolen, D. Attributionversus persuasion as a means for modifying be-havior. Journal of Personality and Social Psy-chology, 1975, 31, 430-441.
Miller, R. L., Seligman, C., Clark, N. T., & Bush, M.Perceptual contrast versus reciprocal concession asmediators of induced compliance. Canadian Journalof Behavioural Science, 1976, 8, 401-409.
Miller, R. L., & Suls, J. Helping, self-attribution,and size of an initial request. Journal of SocialPsychology, 1977, 103, 203-208.
Nemeth, C. Effects of free versus constrained be-havior on attraction between people. Journal ofPersonality and Social Psychology, 1970, 15, 302-311.
Paulhus, D. L., Shaffer, D. R., & Downing, L. L.Effects of making blood donor motives salientupon donor retention: A field experiment. Per-sonality and Social Psychology Bulletin, 1977, 3,99-102.
Pendleton, M. G., & Batson, C. D. Self-presentationand the door-in-the-face technique for inducingcompliance. Personality and Social PsychologyBulletin, 1979, 5, 77-81.
Pliner, P., Hart, H., Kohl, J., & Saari, D. Compli-ance without pressure: Some further data on thefoot-in-the-door technique. Journal of Experi-mental Social Psychology, 1974, 10, 17-22.
Reingen, P. H. Inducing compliance via door-in-the-face and legitimization of paltry contributions.Psychological Reports, 1977, 41, 924.
Reingen, P. H., & Kernan, J. B. Compliance with aninterview request: A foot-in-the-door, self-percep-tion interpretation. Journal of Marketing Research,1977, 14, 365-369.
Schmidt, L. G. The foot-in-the-door effect: An in-quiry. Unpublished doctoral dissertation, Univer-sity of South Carolina, 1973.
Schur, E. M. Labeling deviant behavior: Its so-ciological implications. New York: Harper & Row,1971.
Seligman, C., Bush, M., & Kirsch, K. Relationshipbetween compliance in the foot-in-the-door para-digm and size of first request. Journal of Per-sonality and Social Psychology, 1976, 33, 517-520.
Seligman, C., Miller, R., Goldberg, G., Gelberd, L.,Clark, N., & Bush, M. Compliance in the foot-in-the-door technique as a function of issue similarityand persuasion. Social Behavior and Personality,1976, 4, 267-271.
Smith, C. L., Gelfand, D. M., Hartmann, D. P., &Partlow, M. E. Y. Children's causal attributionsregarding help-giving. Child Development, 1979,SO, 203-210.
Snyder, M., & Cunningham, M. R. To comply or notcomply: Testing the self-perception explanation ofthe "foot-in-the-door" phenomenon. Journal ofPersonality and Social Psychology, 1975, 31, 64-67.
Steele, C. M. Name-calling and compliance. Journal
FOOT-IN-THE-DOOR EFFECT 2239
of Personality and Social Psychology, 1975, 31,361-369.
Tipton, R. M., & Browning, S. Altruism: Reward orpunishment. Journal of Psychology, 1972, 80, 319-322.
Upton, W. E. Altruism, attribution, and intrinsicmotivation in the recruitment of blood donors(Doctoral dissertation, Cornell University, 1973).Dissertation Abstracts International, 1973, 34,6260B. (University Microfilms No. 74-12,652)
Uranowitz, S. Helping and self-attributions: A fieldexperiment. Journal of Personality and SocialPsychology, 1975, 31, 852-854.
Zuckerman, M., Lazzaro, M. M., & Waldgeir, D.Undermining effects of the foot-in-the-door tech-nique with extrinsic reward. Journal of AppliedSocial Psychology, in press.
Received December 1, 1978 •
STATEMENT OF OWNERSHIP. MANAGEMENT AND CIRCULATION
