Journal of Personality and Social Psychology1981, Vol. 40, No. 3, 492-500

Copyright 1981 by the American Psychological Association, Inc.0022-3514/81 /4003-0492S00.75

The Low-Ball Compliance Technique: Task or PersonCommitment?

Jerry M. BurgerWake Forest University

Richard E. PettyUniversity of MissouriColumbia

Three experiments were conducted to examine the mediating process involvedin the low-ball procedure for increasing compliance. In Experiment 1, subjectswho agreed to but were not allowed to perform an initial request complied witha more costly version of the same request to a greater extent than did controlsonly when the second request came from the same person as did the first requestand not when it came from a different person. In Experiment 2, subjects whoagreed to but were not allowed to carry out an initial low-cost request compliedwith a larger request from the same person to the same extent, whether the firstrequest was related or unrelated to the second. In Experiment 3, subjects wereallowed or not allowed to perform an initial small request after agreeing to doso. Later, these subjects were approached by either the same or a different personwith a larger second request. All groups showed increased compliance over acontrol cell. However, subjects not allowed to perform the initial request whowere approached by the same person for the second request showed a higher rateof compliance than subjects in the other experimental conditions. The resultsfrom the three experiments suggest that an unfulfilled obligation to the requester,rather than a commitment to the initial target behavior, is responsible for theeffectiveness of the low-ball technique.

Recent social psychological investigationshave examined the effectiveness of tech-niques designed to increase compliance torequests in the absence of any obvioussources of pressure (cf. DeJong, 1979). Theearliest compliance-without-pressure tech-nique, the foot-in-the-door, was introducedby Freedman and Fraser (1966). Subjectswho complied with an initial small requestwere found to be more likely to agree to asimilar but larger request from a differentperson than were subjects contacted onlywith the second request. Freedman andFraser suggested that as a result of perform-

Portions of this report were presented at the meetingof the Midwestern Psychological Association, Chicago,May 1979.

We would like to thank the many undergraduates whoassisted in the data collection for these experiments,especially Steve Cole, Dan Kaline, Paul Marklin, Mi-chael Oliveri, and Jim York. We are also indebted tothe Columbia chapter of the Multiple Sclerosis Societyfor their cooperation and provision of materials for Ex-periment 3.

Requests for reprints should be sent to Richard E.Petty, Department of Psychology, McAlester Hall,University of Missouri, Columbia, Missouri 65211.

ing the first request, the subject comes to seehim- or herself as the type of person whofavors such requests, resulting in an in-creased willingness to perform the secondrequest.

The most recent procedure for increasingcompliance without pressure, the low-balltechnique, was demonstrated by Cialdini,Cacioppo, Bassett, and Miller (1978). Sub-jects who agreed to but had not yet per-formed an initial small request (e.g., dis-playing a poster for a charity cause) werefound to comply with that same requestwhen it was subsequently made more costly(e.g., they had to pick up the poster at adistant location) at a higher rate than didsubjects approached only with the costly ver-sion of the request. Additionally, Cialdini etal. demonstrated that the low-ball procedurewas significantly more effective in increasingcompliance than the foot-in-the-door tech-nique. Although mere agreement to performan initial small request may be sufficient toengender the same self-perception processthat is operative with the foot-in-the-doortechnique (Bern, 1972), Cialdini et al. ar-

492

LOW-BALL EFFECT 493

gued that the low-ball procedure requires anadditional commitment to a particular be-havior, which is absent in the foot-in-the-door procedure. This additional "cognitivecommitment to the performance of the tar-get behavior" (1978, p. 468) was advancedto account for the increased effectiveness ofthe low-ball procedure beyond that foundwith the foot-in-the-door technique.

Although Cialdini et al. successfully dem-onstrated the effectiveness of the low-ballprocedure, a close examination of their ex-periments suggests an alternative interpre-tation of their findings. In all three of theCialdini et al. (1978) studies, the same ex-perimenter presented subjects with the firstand second request. The possibility exists,therefore, that the subjects experienced anunfulfilled obligation to the specific re-quester in addition to or instead of a com-mitment to the target behavior. Individualsagreeing to the initial request may have feltas if they owed something to the requester,because they were unable to carry out thefirst request. Therefore, when asked by thesame experimenter to engage in the behaviorat an increased cost, subjects may have com-plied with the second request to fulfill theirobligation to the requester.

Cialdini et al. (1978) provide the exampleof an automobile salesperson who secures anagreement to purchase a car and then in-creases the cost of the vehicle. The consumeris said to be committed to the action of pur-chasing the car and is more likely to pur-chase the car at an increased cost than if noinitial commitment to the car had beenmade. However, it is reasonable to assumethat an obligation to the salesperson has alsodeveloped in this situation and may be suf-ficient to account for the increased willing-ness to buy the car at the higher price. If theCialdini et al. reasoning is correctthat asa result of the initial decision, persons be-come committed to the carthen even if adifferent salesperson were to return with thedeal, increased compliance would result.However, if the low-ball procedure is effec-tive because the initial agreement producesan unfulfilled obligation to the salesperson,then subjects would more easily reject thecar at an increased price if the more costlydeal were presented by a new salesperson.

Experiment 1Our first experiment was designed to test

the obligation-to-the-requester interpreta-tion of the low-ball effect. Subjects whoagreed to but did not perform an initial re-quest were asked to perform the same be-havior at a higher cost either by the samerequester or by a different individual. Sub-jects in a control condition were presentedonly with the second request. It was reasonedthat if a commitment to the target behavioris solely responsible for the low-ball effect,as suggested by Cialdini et al., then whetherthe second (more expensive) request is pre-sented by the same or a different personshould not affect the rate of compliance.Both conditions should show more compli-ance than the control condition. On the otherhand, if an unfulfilled obligation to the initialrequester is solely responsible for the low-ball effect, as suggested here, then personsreceiving the first and second requests fromthe same individual should comply with thesecond request at a higher rate than personswho receive the requests from two differentpeople. In fact, the latter condition shouldnot differ from the control cell.

MethodSubjects

Sixty male and female undergraduate psychology stu-dents served individually as subjects. All subjects hadjust finished participating in an experiment on "adver-tising," for which they received class credit. Five un-dergraduate males served as experimenters for thestudy. Which experimenter made which request wasdetermined randomly for each subject.

ProcedureSubjects were randomly assigned to one of three con-

ditions. Upon completion of the "advertising" experi-ment, the experimenter gave the subject his or her creditpoint and then told the subject that he was also workingon a class project and needed some students to serve asvolunteers. The experimenter presented the subject withseveral sheets of long division problems and explainedthat he was conducting research on numeric skills. Eachsheet contained 12 problems that require^gie subjectto divide a multidigit number into another multidigitnumber. The first problem was completed to illustratethat several steps were required in answering each prob-lem. The experimenter explained that the subject's taskwas to work on the problems until getting them all cor-rect. He further explained that the task took about anhour to complete for the average college student.

494 JERRY M. BURGER AND RICHARD E. PETTY

All subjects except those in the control condition werethen told that they could receive another hour of ex-perimental credit by completing the division problems.The experimenter asked the subject if he or she wouldlike to stay and participate. If the subject asked if heor she could complete only part of the problems, comeback to participate, or finish participating in the exper-iment at a later time, the experimenter explained thathe needed to have all of the problems completed at onesitting and the data collected that day.

If subjects declined to participate in the experimenton numeric skills, they were dismissed. If they agreedto participate (27 out of 40 agreed, see Table 1), theexperimenter announced that he would have to go to hisoffice to pick up the answer sheets. The experimenterasked the subject to wait in the room for a few minutes,and then he left with the problem sheets.

Same-requester condition. If subjects were assignedto the same-requester condition, the same experimenterreturned approximately 2 minutes later with the exper-imental materials. The experimenter also held a memo,which he said he had just received from the chairmanof the psychology department's Human Subjects Com-mittee. He explained that due to a shortage of subjects,he would not be allowed to give any experimental creditfor participating in class projects like the numeric skillsstudy. The experimenter then briefly described the taskagain and asked the subject if he or she would still bewilling to participate in the study for no credit. If thesubject declined, he or she was debriefed and excused.If the subject agreed, he or she was handed the problemsheet and allowed to work on one problem before beingstopped and debriefed.

Different-requester condition. If subjects were as-signed to the different-requester condition, a second ex-perimenter returned to the room approximately 2 min-utes after the first experimenter had left. The secondexperimenter introduced himself as one of the personsworking on the numeric skills project with the first ex-perimenter. In addition to the experimental materials,he held the memo from the Human Subjects Committee.He explained that the first experimenter had to leavesuddenly and would not be back. The experimenter saidthat he didn't know if the first experimenter had ex-plained the project and briefly repeated the task re-quirements. He then announced that although they hadinitially planned to give subjects experimental credit forparticipating in the numeric skills study, he had justreceived a memo from the chairman of the Human Sub-jects Committee that did not allow credit for partici-pating in such projects. Subjects were then asked if theywould be willing to participate in the study.

Control condition. Subjects in the control conditionwere not told about the possibility of earning experi-mental credit. Instead, when the experimenter presentedthe initial request he explained that although they hadwanted to give 1 hour of credit, the Human SubjectsCommittee would not allow it. These subjects were thuspresented with only the no-credit (more expensive) re-quest.

ResultsThe number of subjects complying with

the first and second requests in each condi-

tion are presented in Table 1. As can be seenin the table, subjects in the two experimentalconditions did not differ significantly in theirrate of compliance with the initial request,X2( 1) = .11, ns. The investigation was con-cerned, however, with the rate of compliancewith the second, more costly request. Twoorthogonal contrasts were conducted. First,the compliance rate of subjects in the dif-ferent-requester condition (including thosedeclining the initial request) was comparedwith the compliance rate in the control con-dition. It was found that the conditions didnot differ significantly in their rate of com-pliance, x2(0 = .17, ns. Next, the rate ofcompliance in the same-requester conditionwas compared with that in the other twoconditions. It was found that subjects re-ceiving both requests from the same individ-ual complied with the second request at agreater rate than did subjects in the othertwo conditions, x20) = 8.93, p < .005.

When the subjects who declined the initialrequest are dropped from the analyses, sim-ilar results are obtained. Different-requestersubjects did not comply significantly differ-ently (21%) than control subjects, x2U) =.01, ns., and same-requester subjects com-plied at a rate significantly higher (85%)than that for subjects in the other two con-ditions, x2(0 = 10.78. p < .005.

DiscussionThe results of Experiment 1 successfully

replicated the increase in compliance overa control condition found with the low-balltechnique, as operationalized by Cialdini etal. (1978). Subjects complied with a costlyrequest more often when first presented with

Table 1Percentages of Subjects Complying With Initialand Second Requests

Initialrequest

Secondrequest

Group

Same requesterDifferent requesterControl

6570

1314

551520

1134

Note, n = 20 for each cell.

LOW-BALL EFFECT 495

the same request at a lower cost than didsubjects presented with only the more costlyrequest. More importantly, however, thiseffect was found only when the same re-quester presented the first and second re-quests. When a different experimenter pre-sented the second request, subjects were nomore likely to comply with this more costlyrequest than were subjects not receiving theinitial request.

The findings thus fail to support Cialdiniet al.'s commitment-to-the-behavior expla-nation for the low-ball effect. Subjects shouldhave been equally committed to performingthe target behavior in both the same- anddifferent-requester conditions. The findingthat subjects increased their rate of compli-ance beyond that found in the control con-dition only when the same person presentedboth requests suggests that an obligation tothe person instead of or in addition to a com-mitment to the behavior is responsible forthe low-ball effect.

Experiment 2

One question that remains unanswered bythe first experiment is whether a commit-ment to the target behavior is at all necessaryfor producing the low-ball effect. It may bethat when a person agrees to but is unableto perform an initial task, the unfulfilledobligation to the requester makes the personmore susceptible to a second, more costlyrequest from the same person, even if thesecond task is unrelated to the first one. Onthe other hand, if a commitment to performthe target behavior is necessary for the low-ball effect, as suggested by Cialdini et al.,then it would be expected that increasedcompliance would be found only when thesecond request is a more costly version of theinitial request and not when it is unrelated.This question was examined in Experi-ment 2.

Method

Subjects

ProcedureSubjects were contacted by telephone in the evening.

If a subject could not be contacted after three attempts,he or she was dropped from the sample and replacedby another randomly selected subject.

Related-request condition. Subjects in the related-request condition heard the following initial request:

Hi. My name is. . and I'm calling for the Corn-

Sixty university students were randomly selected fromthe university phone directory and assigned to one ofthree experimental conditions.

mittee for Student Awareness. We are a group ofconcerned students interested in demonstrating ouropposition to further increases in student tuition andfee costs. What I would like to know from you nowis, if we send a student by with a petition stating ourposition, would you be willing to sign it?

If subjects asked for further information about the pe-tition, they were told, "The petition merely says thatyou, along with other signers, are in opposition to furthertuition and fee increases." If subjects agreed to the ini-tial request (only two refused and were replaced), theexperimenter replied:

Good. Oh, wait a minute. I'm sorry. I was looking atthe wrong list. It looks like we have already filledseveral petitions. What I meant to call you about wasnot the petition, but a related task.

Subjects were then presented with the second, morecostly request. They were told that in addition to thepetitions,

We would also like to get a large number of studentsto write letters or postcards during this next week tothe Student Opinion Administrator here on campusto further demonstrate our opposition. If I give youthe address right now, would you be willing to writea short postcard or letter just stating in a line or twothat you, too, are opposed to any further increases intuition and fees?

If subjects inquired about what to write in their letters,they were told, "Just write something like: I am writingto say that I am opposed to further increases in studenttuition and fees at [the University of] Missouri." Theexperimenter waited until receiving a verbal reply to therequest before ending the conversation. Thus, in thisexperimental condition, the cost (in terms of time, en-ergy, and the expense of a stamp) of protesting an in-crease in tuition was increased. Subjects agreeing to therequest were given a mailing address, which would allowthe experimenter to record how many subjects in eachcondition showed behavioral as well as verbal compli-ance with the more costly request.

Unrelated-request condition. When the experi-menter initially contacted subjects in the unrelated-re-quest condition, he also identified himself as a memberof the Committee for Student Awareness. Subjects inthis condition were presented with the following initialrequest:

We're calling students to get your opinion about thenew campus shuttle bus system. Could you spare afew minutes now to answer about four or five shortquestions for us?

496 JERRY M. BURGER AND RICHARD E. PETTY

When subjects agreed to this request (none refused),the experimenter replied:

Good. Oh, wait a minute. I'm sorry. I was looking atthe wrong list. What I meant to call you about wasnot the survey, I've already reached my quota on that,but I wanted to ask you about helping us show ouropposition to the suggested further increases in stu-dent tuition and fees here at [the University of] Mis-souri. We have already filled several petitions dem-onstrating our opposition to these increases.

This was followed by the same letter-writing requestthat was presented to the subjects in the same-requestcondition.

Control condition. Subjects in the no-initial-requestcontrol condition were also contacted by the experi-menter, who identified himself as a member of the Com-mittee for Student Awareness. These subjects were told:

We're calling university students about helping usshow our opposition to the suggested further increasesin student tuition and fees here at [the University of]Missouri. We have already filled several petitionsdemonstrating our opposition to these increases.

This was followed by the letter-writing request as pre-sented to subjects in the other two conditions.

ResultsSubjects' responses were assigned the fol-

lowing values: 2 = verbal and behavioralcompliance (agreed on phone and letter re-ceived), 1 = verbal compliance only, and0 = no compliance. To test the specific hy-potheses of interest, two orthogonal plannedcomparisons were performed.

First, as anticipated, a comparison be-tween the related-request condition (M =.70) and the unrelated-request condition(M = .65) failed to find a significant differ-ence between the compliance scores for thesetwo groups (F < 1). However, as expected,when the combined compliance scores ofsubjects in the related-request condition andthe unrelated-request condition (M=.68)were compared with the scores of subjectsin the no-initial-request control condition(M - .35), a significant effect emerged, F(l,57) = 4.70, p < .05.'

DiscussionThe results of Experiment 2 once again

replicated the low-ball effect as operation-alized by Cialdini et al. (1978). Subjectscomplied with a costly request at a higherrate when first presented with a similar low-cost request that they were not allowed toperform. More importantly though, the in-crease in compliance rate did not vary as a

function of the specific behavior that the in-dividual promised to perform in the initialrequest. When an unrelated behavior wasrequested by the same experimenter, sub-jects complied at a rate significantly higherthan that of subjects presented with only onerequest but not significantly different fromthat of subjects presented with a similar firstand second request.

The results thus suggest that a commit-ment to a specific task or issue (e.g., stoppingtuition increases) may not be necessary forthe increase in compliance found with thelow-ball procedure. Instead, it appears thatan unfulfilled obligation to the requestermay be responsible for the low-ball effect.This unfulfilled obligation to the requesterappears to increase compliance to a secondrequest even when a very different issue isinvolved.

Experiment 3Taken together, the results of Experi-

ments 1 and 2 suggest that Cialdini et al.'s(1978) interpretation of the low-ball effectmay not be correct. Instead of a commitmentto the target activity, these two experimentssuggest that an unfulfilled obligation to therequester may be responsible for the increasein compliance found with this technique. Animportant aspect of the Cialdini et al. studiesand Experiments 1 and 2 is that subjectsagreeing to the initial request were not al-lowed to perform that behavior at the orig-inal cost. If the unfulfilled-obligation-to-the-requester interpretation is correct, then itwould be expected that if individuals areprovided with an opportunity to fulfill theinitial obligation, they will not feel a needto help the requester further and will there-fore fail to show the increased rate of com-pliance found in the low-ball procedure.

1 If the verbal and behavioral compliance frequencies

are analyzed separately, then the same pattern emergeson each measure, though only the verbal measure pro-duces a significant low-ball effect. Specifically, the ratesof verbal compliance for the related (13/20) and un-related (12/20) conditions did not differ significantly,but the combined rate was greater than the compliancerate in the control cell (7/20), x20) = 4.05, p < .05.Only two letters in opposition to the tuition increasewere received (behavioral compliance measure), oneeach from subjects in the two experimental cells. Theseletters were forwarded to the University chancellor.

LOW-BALL EFFECT 497

Experiment 3 was designed to examinethis possibility. Subjects were either allowedor not allowed to perform the initial low-costrequest. The same or a different experi-menter later presented the subject with arelated but more costly request. It was an-ticipated that (a) when the subject was al-lowed to perform the initial request, com-pliance to the second request would not beaffected by whether the same person madethe request (since no unfulfilled obligationexisted), and (b) when the subject was notallowed to perform the initial request, com-pliance to the second request would begreater when the same person made the re-quests than when the requests came fromdifferent people (since it is hypothesized thatthe unfulfilled obligation is to a particularperson).

MethodSubjects

Seventy-five college-age males and females served assubjects. Subjects were residents of on-campus dormi-tories or apartment complexes near the universitycampus.

ProcedureSubjects were randomly preassigned to one of the five

conditions. That is, a random order of doors to be ap-proached and a random order of conditions for subjects'answering their doors were predetermined.

Performance manipulation. In the first part of theexperiment, the individual answering the door was pre-sented with the following request by the experimenter:

Hello. My name is . . and I'm working for theNational Multiple Sclerosis Society. We are inter-ested in publicizing our current fund-raising drive andI was wondering if you would be interested in helpingus by displaying a poster like this one on your doorfor the next two weeks.

All subjects agreed to this request. In the perform con-ditions, the experimenter thanked the person for helpingand then taped an Wi in. X 11 in. (21.5 cm X 27.9 cm)poster onto the front door. In the no-perform conditions,the experimenter reached into an envelope as if to pullout a poster but stopped and announced:

I'm sorry. It looks like I'm all out of posters. I guessI gave my last one away to the last person I talkedto. I'm really sorry. But thanks anyway for offeringto help.

Thus in the perform conditions, subjects were able toperform the initial small request, whereas in the no-perform conditions, subjects agreed to but were unable

to perform the initial small request. In both conditions,the experimenter recorded the room number, first name,and experimental condition for each subject.

Requester manipulation. Approximately 10-15minutes after the first request, either the same or a dif-ferent experimenter returned to the door. In the same-requester condition, the experimenter explained "I for-got to mention last time I was here that. . . ." In thedifferent-requester condition, the experimenter intro-duced himself with, "Hello. My name is and I'mworking for the National Multiple Sclerosis Society."In both conditions, the experimenter then said:

The National Multiple Sclerosis Society is looking forpeople to serve as volunteers distributing envelopesfor MS. There will be an MS representative at [aroom in the lobby of the dorm or apartment] between[a 2-hour time interval] tonight handing out collectionenvelopes in which contributions can be made. Wouldyou be interested in helping MS by going tobetween and tonight and picking upa collection envelope and either collecting from othersor taking the envelope for your own contribution?Experimenters waited for an affirmative or negative

reply before asking subjects for their first names, thank-ing them, and leaving. The experimenter then recordedthe room number, the subject's first name, and whetherthe subject had agreed to the second request.

Control condition. In the control condition, subjectswere approached only once. The experimenter presentedthe subject with the different-experimenter request topick up a volunteer packet.2

ResultsThe names and room numbers of subjects

who picked up packets were recorded. As inExperiment 2, subjects were assigned a scoreof 2 (verbal and behavioral compliance), 1(verbal compliance only), or 0 (no compli-ance) for their responses to the second re-quest. The mean scores for the five condi-tions are presented in Table 2. Fourorthogonal contrasts in the context of a one-way analysis of variance were performed totest the specific hypotheses of interest. First,the four experimental conditions were com-pared with the control condition. This con-trast revealed that experimental subjectscomplied more than controls, F(l, 70) =13.34, p < .05. Next, no effect for the same-different requester variable within the per-form conditions was found, F(l, 70) =.37, ns.

2 Four male experimenters who were blind to the ex-

perimental hypotheses were rotated through the rolesrequired in the various conditions. A fifth male actedas the MS representative in the dorm or apartmentlobby.

498 JERRY M. BURGER AND RICHARD E. PETTY

Table 2Effects of Same or Different Requester andPerformance of Initial Request on ComplianceRates

Condition VCF BCF CCI

No perform-same requesterNo perform-different requesterPerform-same requesterPerform-different requesterControl

138

11105

41210

1.13.60.87.73.33

Note. VCF = verbal compliance frequency; BCF = be-havioral compliance frequency; CCI = combined com-pliance index, n = 15 for each cell.

If the commitment-to-the-behavior inter-pretation of the low-ball effect is correct,subjects in the no-perform-different-re-quester condition should have complied withthe second request more often than did sub-jects in the two perform conditions, becauseof their unfulfilled obligation to the task,helping MS. However, this effect failed toemerge, F(l, 70) = 1.01, ns. Finally, if anunfulfilled obligation to the person is re-sponsible for the increased compliance withthe low-ball procedure, then subjects in theno-perform-same-requester condition shouldhave complied significantly more often thansubjects in the other three conditions (be-cause this is the only condition in which anunfulfilled obligation to a person exists). Asignificant contrast supported this viewpoint,F(l, 70) = 4.51, p < .05.3

Discussion

Experiment 3 once again demonstratedthe effectiveness of the low-ball procedurefor increasing the rate of compliance witha costly request by first securing agreementto perform but not allowing performance ofa similar behavior at a lower cost. The resultsalso replicated the foot-in-the-door effect,again demonstrating the effectiveness of thiscompliance technique. Subjects approachedby a different individual with a large request,after agreeing to perform or performing asmaller request, agreed to the second requestat a higher rate than subjects approachedonly with the large request. The results alsoreplicated the Cialdini et al. finding that thelow-ball procedure (with the same requester)

was more effective in inducing compliancethan the foot-in-the-door procedure.

Of most importance in Experiment 3 wasthe finding that subjects not allowed to per-form the initial request and approached bythe same person for the second request hada rate of compliance significantly higherthan that of subjects in the other three ex-perimental conditions. This finding suggeststhat the higher rate of compliance for theno-perform-same-requester subjects is dueto an unfulfilled obligation to the requesterthat developed from the inability to performthe initial request. Subjects approached bya different experimenter were not able tofulfill their obligation to the initial requesterby complying with the second request. Sub-jects allowed to perform the initial requestwere able to fulfill their obligation to therequester and thus were not as likely to com-ply with the second request as were the no-perform-same-requester subjects. The re-sults thus provide additional support for theunfulfilled-obligation-to-the-requester inter-pretation of the low-ball effect.4

General DiscussionSummary

The three experiments presented here sug-gest that the low-ball procedure, as opera-

3 These same four contrasts were computed separately

on the frequencies of verbal and behavioral compliance(see frequencies in Table 2). Analyses on the verbalcompliance measure produced one significant effect:There was greater verbal compliance in the experi-mental cells (42/60) than in the control cell (5/15),X2(l) = 6.89, p < .05. Analyses on the behavioral com-pliance measure yielded one marginally significant re-sult: Behavioral compliance was greater in the no-per-form-same-requester condition (4/15) than in the otherthree experimental cells combined (4/45), x20) = 3.07,p< .10.

4 In addition to the unfulfilled obligation to the re-

quester, subjects in the same-requester conditions mayhave also been motivated by impression-managementconcerns. Tedeschi, Schlenker, and Bonoma (1971) haveargued that people are motivated to present themselvesin a consistent manner to others. However, a motive toappear consistent cannot by itself account for the in-crease in compliance found for the same-requester sub-jects not allowed to perform the behavior over that ofthe perform-same-requester subjects. Because the ex-perimenter saw them perform the initial behavior, per-form-same-requester subjects should have been equallyif not more motivated to present themselves in a con-sistent manner than the no-perform subjects.

LOW-BALL EFFECT 499

tionalized by Cialdini et al. (1978), is a suc-cessful means of increasing compliance witha request. The effectiveness of the techniqueacross three different types of requests in-dicates the robust nature of the effect. Con-sistent with the Cialdini et al. findings, thelow-ball procedure also appears to be moreeffective in increasing compliance than thefoot-in-the-door technique.

However, the effectiveness of the low-ballprocedure does not appear to be due to theexplanation proposed by Cialdini et al.,namely, that a commitment to the targetbehavior develops. Instead, the results of allthree experiments presented in the presentresearch suggest that an unfulfilled obliga-tion to the requester may be responsible forthe effectiveness of the low-ball procedure.In Experiment 1, which provided the mostdirect test of the task versus person com-mitment hypotheses, the low-ball techniquewas effective only when the same personraised the cost of the initial request. Whenthe higher cost version of the initial requestwas presented by a different person, no in-creased compliance over the control condi-tion was found. Experiment 2 found that thesecond request did not have to be a highercost version of the initial request. When sub-jects were unable to comply with the initiallow-cost request, they showed enhancedcompliance with a higher cost request fromthe same person, even when the second re-quest was unrelated to the first. Finally, Ex-periment 3 demonstrated the importance ofthe nonperformance of the initial request.Individuals who were not allowed to performthe initial request were more likely to complywith the second request presented by thesame person than were those allowed to per-form the initial request. The results of allthree studies provide support for the viewthat the effectiveness of the low-ball com-pliance technique is dependent on an unful-filled obligation to a particular person ratherthan a commitment to a specific target be-havior.

Low-Ball Versus Foot-in-the-DoorGiven the above analysis, it seems neces-

sary to clarify the relationship between thelow-ball procedure and the foot-in-the-doortechnique as initially identified by Freed-

man and Fraser (1966). Cialdini et al. sug-gested that the difference between the twocompliance techniques was that the low-ballprocedure relied on inducing a commitmentto the initial target behavior, whereas thefoot-in-the-door procedure did not. How-ever, Experiment 1 strongly indicated thatthe effectiveness of the low-ball proceduredid not depend on inducing a commitmentto the initial target behavior. If it did, en-hanced compliance should have occurredeven when the second request was presentedby a different person, but it did not. Also,Experiment 2 demonstrated that the effec-tiveness of the low-ball procedure was notaffected when the initial request was for abehavior very different from the second,more costly request. Thus, a commitment tothe initial target behavior does not appearnecessary for producing the low-ball effectand therefore cannot distinguish the low-ballfrom the foot-in-the-door technique.

Another apparent difference between thetwo procedures concerns the subject's per-formance of the initial request. WhereasCialdini et al.'s low-ball subjects were notallowed to perform the request at the initialcost, foot-in-the-door subjects typically per-form the initial request. Consistent with theself-perception interpretation of the foot-in-the-door phenomenon (Bern, 1972), the per-formance of this behavior can be seen asenhancing the subject's self-perception thathe or she is the type of person who engagesin such behaviors. However, some investi-gators (Snyder & Cunningham, 1975; Zuck-erman, Lazzaro, & Waldgeir, 1979) havesuccessfully produced a foot-in-the-door ef-fect without the performance of the initialbehavior (cf. DeJong, 1979). Consistent withthis finding, subjects in Experiment 3 whoagreed to but were not allowed to performthe initial request from one person compliedsignificantly more often than the control sub-jects to a second, larger request presentedby a different individual.5 Thus, whether thesubject is allowed to perform the initial re-

5 The foot-in-the-door effect did not emerge in Ex-

periment 1 when a different requester presented the sec-ond request, because in that study, subjects had an ex-ternal reason for agreeing to the initial request (i.e.,getting an extra hour of credit), so there was no needto search for an internal explanation. Thus, the self-perception process would be unlikely to occur.

500 JERRY M. BURGER AND RICHARD E. PETTY

quest does not appear to be the crucial dis-tinction between the foot-in-the-door and thelow-ball procedures.

The foot-in-the-door effect appears to oc-cur when an individual who agrees to per-form (though does not necessarily perform)a small request comes to view him- or herselfin a manner that makes him or her moresusceptible to a second, more costly requestfrom either the same or a different person.Thus, it is likely that there is a componentof the foot-in-the-door effect in each of thefour experimental conditions in Experi-ment 3.

However, one of the experimental condi-tions in Experiment 3 produced significantlymore compliance than the others. Somethingunique appears to occur when (a) the persondoes not perform the initial agreed-uponsmaller request, and (b) the second, largerrequest is presented by the same person.Joint occurrence of these two conditions pro-duces an unfulfilled obligation to the initialrequester, which is not present in the otherthree experimental cells (i.e., when the ini-tial behavior is performed and/or the secondrequest comes from a different person). Thisadditional psychological process (an unful-filled commitment to a person), called thelow-ball effect, appears to be responsible for

producing compliance above that whichwould be expected from the foot-in-the-doorprocedure alone.

References

Bern, D. J. Self-perception theory. In L. Berkowitz(Ed.), Advances in experimental social psychology(Vol. 6). New York: Academic Press, 1972.

Cialdini, R. B., Cacioppo, J. T., Bassett, R., & Miller,J. A. The low-ball procedure for producing compli-ance: Commitment then cost. Journal of Personalityand Social Psychology, 1978, 36, 463-476.

DeJong, W. An examination of the self-perception me-diation of the foot-in-the-door effect. Journal of Per-sonality and Social Psychology, 1979, 37, 2221-2239.

Freedman, J. L., & Eraser, S. C. Compliance withoutpressure: The foot-in-the-door technique. Journal ofPersonality and Social Psychology, 1966,4,195-202.

Snyder, M., & Cunningham, M. R. To comply or notcomply: Testing the self-perception explanation of the"foot-in-the-door" phenomenon. Journal of Person-ality and Social Psychology, 1975, 31, 64-67.

Tedeschi, J. T., Schlenker, B. R., & Bonoma, T. V.Cognitive dissonance: Private ratiocination or publicspectacle? American Psychologist, 1971, 26, 685-695.

Zuckerman, M., Lazzaro, M. M., Waldgeir, D. Un-dermining effects of the foot-in-the-door techniquewith extrinsic reward. Journal of Applied Social Psy-chology, 1979, 9, 292-296.

Received July 28, 1980

Manuscripts Accepted for Publication in the SectionInterpersonal Relations and Group Processes

The Relationship of Ability to Decode the Spouse and Ability to Decode Opposite Sex Others. PatriciaNoller (Animal Behaviour Unit, Department of Psychology, University of Queensland, St. Lucia,Queensland 4067, Australia).

Vocal and Verbal Assertiveness in Same- and Mixed-Sex Groups. Charles E. Kimble (Department ofPsychology, University of Dayton, Dayton, Ohio 45469), and Joyce C. Yoshikawa.