Journal ol Personality and Social Psychology1966, Vol. 4, No. 2, 155-202

COMPLIANCE WITHOUT PRESSURE:THE FOOT-IN-THE-DOOR TECHNIQUE3

JONATHAN L. FREEDMAN AND SCOTT C. FRASER2

Stanford University

2 experiments were conducted to test the proposition that once someone hasagreed to a small request he is more likely to comply with a larger request.The 1st study demonstrated this effect when the same person made bothrequests. The 2nd study extended this to the situation in which differentpeople made the 2 requests. Several experimental groups were run in an effortto explain these results, and possible explanations arc discussed.

How can a person be induced to do some-thing he would rather not do? This questionis relevant to practically every phase of sociallife, from stopping at a traffic light to stop-ping smoking, from buying Brand X to buy-ing savings bonds, from supporting the Marchof Dimes to supporting the Civil Rights Act.

One common way of attacking the problemis to exert as much pressure as possible onthe reluctant individual in an effort to forcehim to comply. This technique has been thefocus of a considerable amount of experi-mental research. Work on attitude change,conformity, imitation, and obedience has alltended to stress the importance of the degreeof external pressure. The prestige of thecommunicator (Kelman & Hovland, 1953),degree of discrepancy of the communication(Hovland & Pritzker, 19S7), size of the groupdisagreeing with the subject (Asch, 1951),perceived power of the model (Bandura,Ross, & Ross, 1963), etc., are the kinds ofvariables that have been studied. This im-pressive body of work, added to the researchon rewards and punishments in learning, hasproduced convincing evidence that greaterexternal pressure generally leads to greatercompliance with the wishes of the experi-menter. The one exception appears to besituations involving the arousal of cognitivedissonance in which, once discrepant behaviorhas been elicited from the subject, the greater

1 The authors are grateful to Evelyn Bless forassisting in the running of the second experimentreported here. These studies were supported in partby Grant GS-196 from the National Science Founda-tion. The first study was conducted while the juniorauthor was supported by an NSF undergraduatesummer fellowship.

2 Now at New York University.

the pressure that was used to elicit the be-havior, the less subsequent change occurs(Festinger & Carlsmith, 1959). But even inthis situation one critical element is theamount of external pressure exerted.

Clearly, then, under most circumstances themore pressure that can be applied, the morelikely it is that the individual will comply.There are, however, many times when forethical, moral, or practical reasons it is diffi-cult to apply much pressure when the goalis to produce compliance with a minimum ofapparent pressure, as in the forced-compliancestudies involving dissonance arousal. Andeven when a great deal of pressure is pos-sible, it is still important to maximize thecompliance it produces. Thus, factors otherthan external pressure are often quite criticalin determining degree of compliance. Whatare these factors?

Although rigorous research on the problemis rather sparse, the fields of advertising,propaganda, politics, etc., are by no meansdevoid of techniques designed to producecompliance in the absence of external pres-sure (or to maximize the effectiveness of thepressure that is used, which is really the sameproblem). One assumption about compliancethat has often been made either explicitly orimplicitly is that once a person has beeninduced to comply with a small request heis more likely to comply with a larger de-mand. This is the principle that is commonlyreferred to as the foot-in-lhc-door or grada-tion technique and is reflected in the sayingthat if you "give them an inch, they'll takea mile." It was, for example, supposed to beone of the basic techniques upon which theKorean brainwashing tactics were based

195

196 JONATHAN L. FRKKDMAN AND SCOTT C.

(Schcin, Schneier, & Barker, 1961), and, ina somewhat different sense, one basis forNazi propaganda during 1Q40 (Brunei",1941). It also appears to be implicit in manyadvertising campaigns which attempt to in-duce the consumer to do anything relating tothe product involved, even sending back acard saying he does not want the product.

The most relevant piece of experimentalevidence comes from a study of conformitydone by Deutsch and Gerard (19SS). Somesubjects were faced with incorrect groupjudgments first in a series in which the stimuliwere not present during the actual judgingand then in a series in which they were pres-ent, while the order of the memory and visualseries was reversed for other subjects. Forboth groups the memory series produced moreconformity, and when the memory series camefirst there was more total conformity to thegroup judgments. It seems likely that thisorder effect occurred because, as the authorssuggest, once conformity is elicited at all it ismore likely to occur in the future. Althoughthis kind of conformity is probably somewhatdifferent from compliance as described above,this finding certainly lends some support tothe foot-in-the-door idea. The present re-search attempted to provide a rigorous, moredirect test of this notion as it applies tocompliance and to provide data relevant toseveral alternative ways of explaining theeffect.

EXPERIMENT I

The basic paradigm was to ask some subjects(Performance condition) to comply first witha small request and then 3 days later witha larger, related request. Other subjects (One-Contact condition) were asked to comply onlywith the large request. The hypothesis wasthat more subjects in the Performance condi-tion than in the One-Contact condition wouldcomply with the larger request.

Two additional conditions were included inan attempt to specify the essential differencebetween these two major conditions. The Per-formance subjects were asked to perform asmall favor, and, if they agreed, they did it.The question arises whether the act of agree-ing itself is critical or whether actually carry-ing it out was necessary. To assess this a

third group of subjects (Agree-Only) wasasked the first request, but, even if theyagreed, they did not carry it out. Thus, theywere identical to the Performance group ex-cept that they were not given the opportunityof performing the request.

Another difference between the two mainconditions was that at the time of the largerrequest the subjects in the Performance con-dition were more familiar with the experi-menter than were the other subjects. ThePerformance subjects had been contactedtwice, heard his voice more, discovered thatthe quesions were not dangerous, and so on.It is possible that this increased familiaritywould serve to decrease the fear and suspicionof a strange voice on the phone and mightaccordingly increase the likelihood of the sub-jects agreeing to the larger request. To con-trol for this a fourth condition was run(Familiarization) which attempted to givethe subjects as much familiarity with the ex-perimenter as in the Performance and Agree-Only conditions with the only difference beingthat no request was made.

The major prediction was that more sub-jects in the Performance condition wouldagree to the large request than in any ofthe other conditions, and that the One-Contact condition would produce the leastcompliance. Since the importance of agree-ment and familiarity was essentially un-known, the expectation was that the Agree-Only and Familiarization conditions wouldproduce intermediate amounts of compliance.

METHODThe prediction slated above was tested in a field

experiment in which housewives were asked to allowa survey team of five or six men to come into theirhomes for 2 hours to classify the household productsthey used. This large request was made under fourdifferent conditions: after an initial contact inwhich the subject had been asked to answer a fewquestions about the kinds of soaps she used, andthe questions were actually asked (Performancecondition) ; after an identical contact in which thequestions were not actually asked (Agree-Only con-dition) ; after an initial contact in which no requestwas made (Familiarization condition) ; or after noinitial contact (One-Contact condition). The depend-ent measure was simply whether or not the subjectagreed to the large request.

COMPLIANCE WITHOUT PUKSSUKE 197

ProcedureThe subjects were 156 Palo AIlo, California,

housewives, 36 in each condition, who were selectedat random from the telephone directory. An addi-tional 12 subjects distributed about equally amongthe three two-contact conditions could not bereached for the second contact and are not in-cluded in the data analysis. Subjects were assignedrandomly to the various conditions, except that theFamiliarization condition was added to the designafter the other three conditions had been com-pleted. AH contacts were by telephone by the sameexperimenter who identified himself as the sameperson each time. Calls were made only in themorning. For the three groups that were contactedtwice, the first call was made on either Mondayor Tuesday and the second always 3 days later.All large requests were made on either Thursdayor Friday.

At the first contact, the experimenter introducedhimself by name and said that he was from theCalifornia Consumers' Group. In the Performancecondition he then proceeded:

We arc calling you this morning to ask if youwould answer a number of questions about whathousehold products you use so that we couldhave this information for our public service publi-cation, "The Guide." Would you be willing togive us this information for our survey?

If the subject agreed, she was asked a series of eightinnocuous questions dealing with household soaps(e.g., "What brand of soap do you use in yourkitchen sink?") She was then thanked for hercooperation, and the contact terminated.

Another condition (Agree-Only) was run to assessthe importance of actually carrying out the requestas opposed to merely agreeing to it. The only dif-ference between this and the Performance conditionwas that, if the subject agreed to answer the ques-tions, the experimenter thanked her, but said thathe was just lining up respondents for the surveyand would contact her if needed.

A third condition was included to check on theimportance of the subject's greater familiarity withthe experimenter in the two-contact conditions.In this condition the experimenter introducedhimself, described the organization he worked forand the survey it was conducting, listed the ques-tions he was asking, and then said that he wascalling merely to acquaint the subject with the ex-istence of his organization. In other words, thesesubjects were contacted, spent as much time onthe phone with the experimenter as the Performancesubjects did, heard all the questions, but neitheragreed to answer them nor answered them.

In all of these two-contact conditions some sub-jects did not agree to the requests or even hungup before the requests were made. Every subjectwho answered the phone was included in the analysisof the results and was contacted for the secondrequest regardless of her extent of cooperativeness

during the first contact. In other words, no subjectwho could be contacted the appropriate numberof times was discarded from any of the fourconditions.

The large request was essentially identical for allsubjects. The experimenter called, identified himself,and said cither that his group was expanding itssurvey (in the case of the two-contact conditions)or that it was conducting a survey (in the One-Contact condition). In all four conditions he thencontinued:

The survey will involve five or six men from ourstaff coming into your home some morning forabout 2 hours to enumerate and classify all thehousehold products that you have. They will haveto have full freedom in your house to go throughthe cupboards and storage places. Then all thisinformation will be used in the writing of thereports for our public service publication, "TheGuide."

If the subject agreed to the request, she was thankedand told that at the present time the experimenterwas merely collecting names of people who werewilling to take part and that she would be contactedif it were decided to use her in the survey. If shedid not agree, she was thanked for her time. Thisterminated the experiment.

RESULTSApparently even the small request was not

considered trivial by some of the subjects.Only about two thirds of the subjects inthe Performance and Agree-Only conditionsagreed to answer the questions about house-hold soaps. It might be noted that none ofthose who refused the first request lateragreed to the large request, although asstated previously all subjects who were con-tacted for the small request are included inthe data for those groups.

Our major prediction was that subjectswho had agreed to and carried out a smallrequest (Performance condition) would subse-

TABLE 1PKRCKNTAOE oj' SUBJECTS COMPLYING WITH

LARGE REQUEST IN EXPERIMENT t

Condition

PerformanceAgree-OnlyFamiliarizationOne-Contact

52.833.327.8*22.2:"«

Note.—-N — 36 for each group. Significancedifferences from (he IVifonnance condition.

* J> < .07.:'=i p < .02.

198 JONATHAN F K K K U M A N AND SCOTT C. FUASKU

queutly be more likely to comply with alarger request than would subjects who wereasked only the larger request (One-Contactcondition). As may be seen in Table 1, theresults support the prediction. Over 50% ofthe subjects in the Performance conditionagreed to the larger request, while less than2<3°/f) of the One-Contact condition agreed toit. Thus it appears that obtaining compli-ance with a small request does tend to in-crease subsequent compliance. The questionis what aspect of the initial contact producesthis effect.

One possibility is that the effect wasproduced merely by increased familiarity withthe experimenter. The Familiarization controlwas included to assess the effect on compli-ance of two contacts with the same per-son. The group had as much contact withthe experimenter as the Performance group,but no request was made during the first con-tact. As the table indicates, the Familiariza-tion group did not differ appreciably inamount of compliance from the One-Contactgroup, but was different from the Perform-ance group (x~ — 3.70, /J < .07). Thus, al-though increased familiarity may well leadto increased compliance, in the present situa-tion the differences in amount of familiarityapparently were not great enough to produceany such increase; the effect that was ob-tained seems not to be due to this factor.

Another possibility is that the criticalfactor producing increased compliance issimply agreeing to the small request (i.e.,carrying it out may not be necessary). TheAgree-Only condition was identical to thePerformance condition except that in theformer the subjects were not asked the ques-tions. The amount of compliance in thisAgree-Only condition fell between the Per-formance and One-Contact conditions andwas not significantly different from either ofthem. This leaves the effect of merely agree-ing somewhat ambiguous, but it suggests thatthe agreement alone may produce part ofthe effect.

Unfortunately, it must be admi t ted thatneither of these control conditions is an en-t i rely adequate test of the possibility it wasdesigned to assess. Both conditions are insome way quite peculiar and may have made

a very different and extraneous impressionon the subject than did the Performance con-dition. In one case, a housewife is asked toanswer some questions and then is not askedthem; in the other, some man calls to tellher about some organization she has neverheard of. Now, by themselves neither of theseevents might produce very much suspicion.But, several days later, the same man callsand asks a very large favor. At this pointit is not at all unlikely that many subjectsthink they are being manipulated, or in anycase that something strange is going on. Anysuch reaction on the part of the subjectswould naturally tend to reduce the amountof compliance in these conditions.

Thus, although this first study demon-strates that an initial contact in which a re-quest is made and carried out increases com-pliance with a second request, the questionof why and how the initial request producesthis effect remains unanswered. In an attemptto begin answering this question and to ex-tend the results of the first study, a secondexperiment was conducted.

There seemed to be several quite plausibleways in which the increase in compliancemight have been produced. The first wassimply some kind of commitment to or in-volvement with the particular person makingthe request. This might work, for example,as follows: The subject has agreed to thefirst request and perceives that the experi-menter therefore expects him also to agreeto the second request. The subject thus feelsobligated and does not want to disappoint theexperimenter; he also feels that he needs agood reason for saying "no"—a better reasonthan he would need if he had never said"yes." This is just one line of causality—theparticular process by which involvement withthe experimenter operates might be quite dif-ferent, but the basic idea would be similar.The commitment is to the particular person.This implies that the increase in compliancedue to the first contact should occur prima-rily when both requests are made by thesame person.

Another explanation in terms of involve-ment centers around the particular issue withwhich the requests arc concerned. Once thesubject has taken some action in connection

COMPLIANCE WITHOUT PRESSURE 199

wilh an area of concern, bo it surveys, polili-cal activity, or highway safely, there is prob-ably a tendency lo become somewhat moreconcerned with the area. The subject beginsthinking about it, considering its importanceand relevance to him, and so on. This tendsto make him more likely to agree to takefurther action in the same area when he islater asked to. To the extent that this is thecritical factor, the initial contact should in-crease compliance only when both requests arerelated to the same issue or area of concern.

Another way of looking at the situationis that the subject needs a reason to say"no." In our society it is somewhat difficultto refuse a reasonable request, particularlywhen it is made by an organization that isnot trying to make money. In order to refuse,many people feel that they need a reason—simply not wanting to do it is often not initself sufficient. The person can say to therequester or simply to himself that he doesnot believe in giving to charities or tippingor working for political parties or answeringquestions or posting signs, or whatever he isasked to do. Once he has performed a par-ticular task, however, this excuse is no longervalid for not agreeing to perform a similartask. Even if the first thing he did wastrivial compared to the present request, hecannot say he never does this sort of thing,and thus one good reason for refusing is re-moved. This line of reasoning suggests thatthe similarity of the first and second requestsin terms of the type of action required is animportant factor. The more similar they are,the more the "matter of principle" argumentis eliminated by agreeing to the first request,and the greater should be the increase incompliance.

There arc probably many other mecha-nisms by which the initial request might pro-duce an increase in compliance. The secondexperiment was designed in part to test thenotions described above, but its major pur-pose was to demonstrate the effect unequivo-cally. To this latter end it eliminated one ofthe important problems with the first studywhich was that when the experimenter madethe second request he was not blind as towhich condition the subjects were in. In thisstudy the second request was always made

by someone other than the person who ma.deIhe first request, and the second experimenterwas blind as to what condition the subjectwas in. This eliminates the possibility thatthe experimenter exerted systematically dif-ferent amounts of pressure in different experi-mental conditions. If the effect of the firststudy were replicated, it would also rule outthe relatively uninteresting possibility thatthe effect is due primarily to greater famili-arity or involvement with the particularperson making the first request.

EXPERIMICNT II

The basic paradigm was quite similar tothat of the first study. Experimental subjectswere asked to comply with a small requestand were later asked a considerably largerrequest, while controls were asked only thelarger request. The first request varied alongtwo dimensions. Subjects were asked eitherto put up a small sign or to sign a petition,and the issue was either safe driving or keep-ing California beautiful. Thus, there werefour first requests: a small sign for safedriving or for beauty, and a petition for thetwo issues. The second request for all subjectswas to install in their front lawn a very largesign which said "Drive Carefully." The fourexperimental conditions may be defined interms of the similarity of the small and largerequests along the dimensions of issue andtask. The two requests were similar in bothissue and task for the small-sign, safe-drivinggroup, similar only in issue for the safe-driving-petition group, similar only in taskfor the small "Keep California Beautiful"sign group, and similar in neither issue nortask for the "Keep California Beautiful"petition group.

The major expectation was that the threegroups for which either the task or the issuewere similar would show more compliancethan the controls, and it was also felt thatwhen both were similar there would prob-ably be the most compliance. The fourthcondition (Different Issue-Different Task)was included primarily to assess the effectsimply of the initial contact which, althoughit was not identical to the second one oneither issue or task, was in many ways qui te

200 JONATHAN L. KUKKDMAN AND SCOTT C. KHASICR

similar (e.g., a young student asking for co-operation on a noncontrovcrsial issue). Therewere no clear expectations as to how thiscondition would compare to the controls.

METHODThe subjects were 114 women and 13 men living

in Palo Alto, California. Of these, 9 women and6 men could not be contacted for the second re-quest and are not included in the data analysis.The remaining 112 subjects were divided aboutequally among the live conditions (see Table 2).All subjects were contacted between 1:30 and 4:30on weekday afternoons.

Two experimenters, one male and one female,were employed, and a different one always made thesecond contact. Unlike the first study, the experi-menters actually went to the homes of the subjectsand interviewed them on a facc-to-face basis. Aneffort was made to select subjects from blocks andneighborhoods that were as homogeneous as possible.On each block every third or fourth house wasapproached, and all subjects on that block were inone experimental condition. This was necessarybecause of the likelihood that neighbors would talkto each other about the contact. In addition, forevery four subjects contacted, a fifth house waschosen as a control but was, of course, not con-tacted. Throughout this phase of the experiment,and in fact throughout the whole experiment, thetwo experimenters did not communicate to eachother what conditions had been run on a givenblock nor what condition a particular house was in.

The small-sign, safe-driving group was told thatthe experimenter was from the Community Commit-tee for Traffic Safety, that he was visiting a numberof homes in an attempt to make the citizens moreaware of the need to drive carefully all the lime,and that he would like the subject to take a smallsign and put it in a window or in the car so thatit would serve as a reminder of the need to drivecarefully. The sign was 3 inches square, said "Bea safe driver," was on thin paper without a gummedbacking, and in general looked rather amateurishand unattractive. If the subject agreed, he was giventhe sign and thanked; if he disagreed, he was simplythanked for his time.

The three other experimental conditions werequite similar with appropriate changes. The otherorganization was identified as the Keep CaliforniaBeautiful Committee and its sign said, appropriatelyenough, "Keep California Beautiful." Both signswere simply black block letters on a white back-ground. The two petition groups were asked tosign a petition which was being sent to California'sUnited States Senators. The petition advocated sup-port for any legislation which would promote cithersafer driving or keeping California beautiful. Thesubject was shown a petition, typed on heavy bondpaper, with at least 20 signatures already affixed.If she agreed, she signed and was thanked. If shedid not agree, she was merely thanked.

The second contact was made about 2 weeks a f t e r( l i e initial one. Each experimenter was armed wi tha list of houses which had been compiled by theother experimenter. This list contained all fourexperimental conditions and the controls, and, ofcourse, there was no way for the second experi-menter to know which condition the subject hadbeen in. At this second contact, ail subjects wereasked the same thing: Would they put a large signconcerning safe driving in their front yard? Theexperimenter identified himself as being from theCitizens for Safe Driving, a different group fromthe original safe-driving group (although it is likelythat most subjects who had been in the safe-drivingconditions did not notice the difference). The subjectwas shown a picture of a very large sign reading"Drive Carefully" placed in front of an attractivehouse. The picture was taken so that the sign ob-scured much of the front of the house and com-pletely concealed the doorway. It was rather poorlylettered. The subject was told that: "Our men willcome out and install it and later come and removeit. It makes just a small hole in your lawn, butif this is unacceptable to you we have a specialmount which will make no hole." She was asked toput the sign up for a week or a week and a half.If the subject agreed, she was told that more namesthan necessary were being gathered and if her homewere to be used she would be contacted in a fewweeks. The experimenter recorded the subject'sresponse and this ended the experiment.

RKSULTS

First, it should be noted that there wereno large differences among the experimentalconditions in the percentages of subjectsagreeing to the first request. Although some-what more subjects agreed to post the "KeepCalifornia Beautiful" sign and somewhatfewer to sign the beauty petition, none ofthese differences approach significance.

The important figures are the number ofsubjects in each group who agreed to thelarge request. These arc presented in Table 2.The figures for the four experimental groupsinclude all subjects who were approached thefirst time, regardless of whether or not theyagreed to the small request. As noted above,a few subjects were lost because they couldnot be reached for the second request, and,of course, these are not included in the table.

It is immediately apparent that the firstrequest tended to increase the degree ofcompliance with the second request. Whereasfewer than 20% of the controls agreed toput the large sign on their lawn, over 55%of the experimental subjects agreed, with over

COMPLIANCE WITHOUT PRKSSUKIC 201

TAIJLE 2J ' E K C K N T A G K oji' SUBJECTS COMPLYING w i r u

LARGE REQUEST I N KXFKRISIF.M' J I

Issue-1

SimilarDifferent

- —Similar

76.0**•47.6*

Ta

jV

2521

sk»— —Diileient47.8*47.4*

iV

2319

One-Contact 16.7 (;V = 24)

Denotes relationship between firs t and second requests.

45% being the lowest degree of compliancefor any experimental condition. As expected,those conditions in which the two requestswere similar in terms of either issue or taskproduced significantly more compliance thandid the controls (x2's range from 3.67, p < .07to IS.01, p < .001). A somewhat unexpectedresult is that the fourth condition, in whichthe first request had relatively little in com-mon with the second request, also producedmore compliance than the controls (x2 —3.40, p < .08). In other words, regardless ofwhether or not the two requests are similarin either issue or task, simply having the firstrequest tends to increase the likelihood thatthe subject will comply with a subsequent,larger request. And this holds even when thetwo requests are made by different peopleseveral weeks apart.

A second point of interest is a comparisonamong the four experimental conditions. Asexpected, the Same Issue-Same Task condi-tion produced more compliance than any ofthe other two-contact conditions, but the dif-ference is not significant (^2!s range from 2.7to 2.9). If only those subjects who agreed tothe first request are considered, the samepattern holds.

DISCUSSIONTo summarize the results, the first study

indicated that carrying out a small requestincreased the likelihood that the subjectwould agree to a similar larger request madeby the same person. The second study showedthat this effect was quite strong even whena different person made the larger request,

and Hie (wo requests were quite dissimilar.How may these results be explained?

Two possibilities were outlined previously.The matter-of-principle idea which centeredon the particular type of action was not sup-ported by the data, since the similarity ofthe tasks did not make an appreciable dif-ference in degree of compliance. The notionof involvement, as described previously, alsohas difficulty accounting for some of the find-ings. The basic idea was that once someonehas agreed to any action, no matter howsmall, he tends to feel more involved than hedid before. This involvement may centeraround the particular person making the firstrequest or the particular issue. This is quiteconsistent with the results of the first study(with the exception of the two control groupswhich as discussed previously were ratherambiguous) and with the Similar-Issue groupsin the second experiment. This idea of in-volvement does not, however, explain the in-crease in compliance found in the two groupsin which the first and second request did notdeal with the same issue.

It is possible that in addition to or insteadof this process a more general and diffusemechanism underlies the increase in compli-ance. What may occur is a change in theperson's feelings about getting involved orabout taking action. Once he has agreed toa request, his attitude may change. He maybecome, in his own eyes, the kind of personwho does this sort of thing, who agrees torequests made by strangers, who takes actionon things he believes in, who cooperates withgood causes. The change in attitude could betoward any aspect of the situation or towardthe whole business of saying "yes." The basicidea is that the change in attitude need notbe toward any particular issue or person oractivity, but may be toward activity or com-pliance in general. This would imply that anincrease in compliance would not depend uponthe two contacts being made by the sameperson, or concerning the same issue or in-volving the same kind of action. The simi-larity could be much more general, such asboth concerning good causes, or requiring asimilar kind of action, or being made bypleasant, attractive individuals.

It is not being suggested that this is the

202 JONATHAN I,. FKKKDMAN AND SCOTT C. FRASKR

only mechanism operating here. The idea ofinvolvement, continues to be extremely plausi-ble, and there are probably a number ofother possibilities. Unfortunately, the presentstudies offer no additional data with which tosupport or refute any of the possible explana-tions of the effect. These explanations thusremain simply descriptions of mechanismswhich might produce an increase in compli-ance after agreement with a first request.Hopefully, additional research will test theseideas more fully and perhaps also specifyother manipulations which produce an in-crease in compliance without an increase inexternal pressure.

It should be pointed out that the presentstudies employed what is perhaps a veryspecial type of situation. In all cases the re-quests were made by presumably nonprofitservice organizations. The issues in the secondstudy were deliberately noncontroversial, andit may be assumed that virtually all subjectsinitially sympathized with the objectives ofsafe driving and a beautiful California. Thisis in strong contrast to campaigns which aredesigned to sell a particular product, politicalcandidate, or dogma. Whether the techniqueemployed in this study would be successful inthese other situations remains to be shown.

REFERENCES

ASCII, S. E. Effects of group pressure upon the modification and distortion of judgments. In H. duct/.-kow (Ed.), Groups, leadership and men; researchin human relations. Pittsburgh: Carnegie Press,1951. Pp. 177-190.

BANDURA, A., Ross, D., & Ross, S. A. A comparativetest of the status envy, social power, and second-ary reinforcement theories of idenlificalory learn-ing. Journal of Abnormal and Social Psychology,1963, 67, 527-534.

BRUNEK, J. The dimensions of propaganda: Germanshort-wave broadcasts to America. Journal ojAbnormal and Social Psychology, 1941, 36, 311-337.

DF.UTSCII, M., & GERARD, H. B. A study of norma-tive and informational social influences upon indi-vidual judgment. Journal of Abnormal and SocialPsychology, 1955, 51, 629-636.

FESTINGER, L., & CARLSMITIT, J. Cognitive conse-quences of forced compliance. Journal of Abnormaland Social Psychology, 1959, 58, 203-210.

HOVLAND, C. I., & PBITZKER, H. A. Extent of opin-ion change as a function of amount of changeadvocated. Journal of Abnormal and Social Psy-chology, 1957, 54, 257-261.

KELMAN, H. C., & HOVLAND, C. I. "Reinstatement"of the communicator in delayed measurement ofopinion change. Journal oj Abnormal and SocialPsychology, 1953, 48, 327-335.

SCHEIN, E. H., SCHNEIER, I., & BARKER, C. H. Co-ercive pressure. New York: Norton, 1961.

(Received August 2, 1965)