Low-Ball Procedure for Producing Compliance: Commitment...

Journal of Personality and Social Psychology1978, Vol. 36, No. 5, 463-476

Low-Ball Procedure for Producing Compliance:Commitment then Cost

Robert B. CialdiniArizona State University

Rodney BassettRoberts Wesleyan College

John T. CacioppoUniversity of Notre Dame

John A. MillerOhio State University

The low-ball technique, a tactic often used by automobile sales dealers toproduce compliance from customers, was examined in a set of three experi-ments. In all three studies, a requester who induced subjects to make aninitial decision to perform a target behavior and who then made performanceof the behavior more costly obtained greater final compliance than a requesterwho informed subjects of the full costs of the target behavior from the outset.The low-ball phenomenon—that an active preliminary decision to take anaction tends to persevere even after the costs of performing the action havebeen increased—was found to be reliable (Experiment 1), different from thefoot-in-the-door effect (Experiment 2), and effective only when the preliminarydecision was made with a high degree of choice (Experiment 3). In competi-tion with three other conceptual explanations, a formulation based on the con-cept of commitment was seen to best account for the results. An ecologicallyderived strategy for the identification and investigation of research questionswas used and discussed.

Social psychologists have recently begun toexamine the effects of a variety of factors onthe likelihood that one person will complywith a request from another (cf. Cialdini &Schroeder, 1976). These investigations havegenerally used a similar epistemological se-quence in attempting to uncover the psycho-logical processes that influence compliancebehavior. Typically, factors likely to affectthe tendency to comply with a request havebeen identified on the basis of existing psy-chological theory. Once selected in this man-ner, the variables are submitted to experi-mental test to determine whether they doinfluence compliance probabilities accordingto prediction.1

The authors are grateful to Richard Cialdini,Julian Edney, Charles A. Kiesler, Darwyn E. Lin-der, and Peter Reingen for their valuable commentson an earlier version of this manuscript.

Requests for reprints should be sent to Robert B.Cialdini, Department of Psychology, Arizona StateUniversity, Tempe, Arizona 8S281.

While this deductive approach has provenfruitful in generating a considerable amountof evidence concerning a number of compli-ance-related factors, it could be argued thatsuch an orientation may be misleading as tothe ability of these factors to produce mean-ingful levels of compliance in naturally oc-curring situations. That is, in order to per-form an appropriate experimental test of atheoretical hypothesis, it is often necessary tocontrol away all sources of variance but theones under direct study. But it may well bethat these extraneous sources of variance areprecisely the ones that are prepotently strongin the normal pattern of the behavior beinginvestigated. By eliminating the action ofthese extraneous variables for design reasons,

1 Of course, this approach, wherein social phe-nomena are selected for study (and, consequently,for possible validation) by virtue of their relationto extant theory, is not unique to the complianceliterature; much of social psychological research canbe characterized similarly.

Copyright 1978 by the American Psychological Association, Inc. 0022-3S14/78/360S-0463$00.7S

463

464 CIALDINI, CACIOPPO, BASSETT, MILLER

then, the researcher may discover certainstatistically significant effects to exist in thecontrived, antiseptic experimental contextthat are so small as to be overwhelmed inreal behavior settings involving the naturalpresence of other, more powerful factors.Current social psychological research has re-cently been faulted for too frequently con-cerning itself with differences that, whilestatistically significant, are trivial or mean-ingless in the description of naturally occur-ring social behavior (Helmreich, 1975; Smith,1972; Bickman, Note 1).

A different orientation to the investigationof compliance behavior could greatly reducethe magnitude of the above-described diffi-culty. The techniques and procedures com-monly used in everyday compliance settings(e.g., sales, fund raising, collective bargain-ing) would be identified initially throughobservation. These procedures and techniquescould then be classified according to the con-ceptual variables they seem to embody. Aftera factor potentially linked to compliancebehavior has been selected in this fashion, itcould be put to experimental test to determinewhether it does mediate the effectiveness ofthe technique that brought it to light, as wellas to determine its generality in affecting avariety of other compliance decisions. Bybeginning with techniques that are alreadyused in real-world compliance contexts andthen moving to the examination of theirpsychological underpinnings, the conceptualand theoretical factors related to compliancemay be assessed with enhanced confidence intheir ecological validity.2

The Low-Ball Technique

It is the purpose of the present article touse such an investigatory sequence in an ex-amination of the relationship of certain socialpsychological concepts to compliance behaviorlike that typically obtained through a con-temporary sales practice. There is a tactic,reputedly widespread (Carlson, 1973; Con-sumer Reports, 1974), that is used by somesales organizations to produce compliancefrom their customers. The technique, called"throwing a low-ball" or "low-balling," isespecially prevalent among new-car dealers.The critical component of the procedure is

for the salesperson to induce the customer tomake an active decision to buy one of thedealership's cars by offering an extremely goodprice, perhaps as much as $300 below com-petitors' prices. Once the customer has madethe decision for a specific car (and has evenbegun completing the appropriate forms), thesalesperson removes the price advantage inone of a variety of ways. For example, thecustomer may be told that the originallycited price did not include an expensive op-tion that the customer had assumed was partof the offer. More frequently, however, theinitial price offer is rescinded when the sales-person "checks with the boss," who does notallow the deal because "we'd be losingmoney." Sometimes, the original agreement isvoided by the used-car manager, who offers atrade-in price substantially below the in-flated one suggested by the salesperson inthe initial negotiation. In each instance, theresult is the same: The reason that the cus-tomer made a favorable purchase decision isremoved, and the performance of the targetbehavior (i.e., buying that specific automo-bile) is rendered more costly. The increasedcost is such that the final price is equivalentto, or sometimes slightly above, that of thedealer's competitors. Yet, car dealership lorehas it that more customers will remain withtheir decision to purchase the automobile,even at the adjusted figure, than would havebought it had the full price been revealedbefore a purchase decision had been obtained.The essence of the low-ball procedure, then,is for a requester to induce another to make abehavioral decision concerning a target ac-tion. It is assumed that the decision will per-sist even after circumstances have changedto make performance of the target actionmore costly.

The first step in an examination of thelow-ball technique and its relationship tocompliance involves a demonstration of theeffectiveness of the tactic. Does it reallywork, or have automobile dealers, in theabsence of evidence from controlled pro-

2 The first step of looking to natural settings forexamples of widely used tactics has also been sug-gested (McGuire, 1973) as a means for generatingnew hypotheses. McGuire labeled the procedure"analyzing the practitioner's rule of thumb."

LOW-BALL PROCEDURE 465

cedures, deluded themselves as to the com-pliance-producing power of low-balling? Inorder to provide experimental evidence con-cerning the reliability of low-ball proceduresin enhancing compliance, a small field studywas conducted. The study implemented thelow-ball strategy by obtaining a decision fromsubjects to execute a target behavior andthen raising the cost of performing that be-havior. The low-ball procedure was contrastedwith a control procedure in which subjectswere informed of the full cost of the targetbehavior before being requested to perform it.

Experiment 1Method

Subjects and procedure. Subjects were 63 stu-dents of both sexes enrolled in introductory psy-chology classes at a large state university. The sub-jects were randomly selected from class rolls andphoned by an experimental confederate. The con-federate, who was blind to the experimental hy-pothesis, solicited subject participation in a psychol-ogy experiment by using either low-ball or controlprocedures. In both conditions, the confederate in-troduced herself as follows:

My name is . _. I'm calling for the PsychologyDepartment to schedule Psychology 100 studentsfor an experiment on thinking processes. The ex-periment concerns the way people organize facts.We can give you 1 hour of credit for your par-ticipation in this experiment.

At this point, the experimental script diverged forthe two conditions.

Control condition. Before they were asked ifthey would be willing to participate, subjects inthe control condition were informed that the ex-periment would take place at 7:00 a.m. Specifically,the confederate said:

The room in which the experiment is being heldis used during the day and evening by other peo-ple in the department; so we are running thisexperiment at 7:00 in the morning on Wednesdaysand Fridays. Can I put you down for Wednes-day or Friday morning at 7:00?

If a subject said "No," he or she was debriefed andthanked. If the subject said "Yes," an appointmentwas made and the subject's name taken.

Low-ball condition. Subjects in this conditionwere asked if they wished to participate after theexperimental requirements were only partially de-scribed. If a subject agreed, he or she was theninformed of the 7:00 a.m. starting time and wasagain asked if he or she was willing to participate.Specifically, the confederate said, "Would you be

willing to participate?" A subject who said "No,"was debriefed and thanked.3 If a subject inquiredabout a time, the confederate replied,

Well, we have more than one time during theweek, but right now I'm just interested in findingout if you wish to participate.

If the subject said "Yes," the confederate continuedas follows:

The room in which the experiment is being heldis used during the day and evening by other peo-ple in the department; so we are running thisexperiment at 7:00 in the morning on Wednesdaysand Fridays. Can I put you down for Wednesdayor Friday morning at 7:00?

Unless the experimental appointment was for thefollowing day, all subjects agreeing to participatewere called again the night before the appointmentto remind them of the experiment. If a subject wasnot home, a message was left.

Two dependent measures were taken. The first,measuring verbal compliance, was the percentage ofsubjects who made an appointment to participate inthe study at one of the two specified times; theother, measuring behavioral compliance, was thepercentage of subjects who actually appeared fortheir 7:00 a.m. appointments. Those subjects whoappeared did indeed participate in an experiment onthinking processes at that time. At the completionof the thinking-processes experiment, subjects werefully informed of the procedures of both studies.

Results

The data on verbal compliance with theconfederate's request confirm the effectivenessof the low-ball strategy, in that 56% (19/34)of the low-ball condition subjects made anappointment to participate, whereas only31% (9/29) of the control condition sub-jects did so, X

2(l)-4.14, p < .05. Thesuperiority of the low-ball procedure wasdemonstrated to an even greater extent onthe more important, practical measure ofbehavioral compliance, in that 53% (18/34)of the low-ball condition subjects who werecalled actually appeared at the appointedtime as compared to only 24% (V29) of

3 To avoid the possibility that differential subjectloss could account for our results, any subject whorefused even this request was included in our dataanalysis and categorized as noncompliant. Such aprocedure was carried out in the subsequent experi-ments as well.


the control condition subjects, x2U) — 5.40,p < .024. A high percentage of subjects inboth conditions who complied verbally withthe request complied behaviorally as well;this percentage was somewhat, but not sig-nificantly, higher in the low-ball condition,95% (18/19) versus 79% (7/9). Therewere no significant differences due to sex ofsubject.

Discussion

The results of our analogue of the low-balltechnique indicate that the sequence of ob-taining an active decision from a target per-son to perform an action and only then pro-viding information about the full costs of theaction is an effective way to produce com-pliance with a request to perform the fullydescribed action. Armed with such evidencefrom controlled, experimental procedures, ourconfidence that the low-ball strategy doesreally work and is not merely a sales mythcan be increased. Given that increased confi-dence, the next step in the examination ofthe technique would seem to be the demon-stration of its generality to a naturalisticcontext unlike that of Experiment 1. That is,although Experiment 1 was a field study andthe experimenter was blind to the hypothesis,the study did use as subjects introductorypsychology students who interacted with arequester whom they perceived as an experi-menter; consequently, experimental demandinfluences (Orne, 1962) may conceivablyhave been implicated. Further, if the tech-nique is cross-situationally robust, it shouldinduce compliance in settings and with typesof behavior quite unlike those to which it iscustomarily applied; for example, in con-trast to the typical sales context, we shouldbe able to establish the tactic's effectivenessin a charity context and on altruistic action.To these ends, it was decided to conduct asecond study that was wholly unrelated, inthe subject's mind, to psychological experi-mentation and that used a form of benevo-lence as the target behavior.

Another purpose of Experiment 2 was toprovide evidence that the low-ball technique

was different from an established compliancetactic that could be seen as similar, if notidentical, to low-balling. That is, it might beargued that the low-ball sequence in which arequester secures an active decision to per-form a target behavior and then raises thecost of performing the behavior is just aversion of the foot-in-the-door procedure(Freedman & Fraser, 1966). A requesteruses the foot-in-the-door technique by in-ducing an individual to perform an initialfavor and subsequently asking that individualto comply with a larger, second request.While both techniques seek to gain perform-ance of a costly action by first obtainingaccession to an apparently less costly re-quest, there is at least one important dif-ference. With the low-ball tactic, the behaviorrequested initially (e.g., buying a certaincar, participating in a certain experiment) isin fact the target behavior; only the cost ofcarrying out that specific behavior changes.With the foot-in-the-door procedure, thebehavior requested initially may be relatedto the larger favor desired by the requester,but it is not the target behavior itself.Whether the above-described difference be-tween the two techniques is just a semanticone or is a genuine one that would manifestitself in the differential ability of the twotactics to produce compliance in given situ-ations is, of course, an empirical question. Itwas our feeling that the procedural differencebetween the two techniques would empowerthe low-ball technique as the more effectivecompliance inducer. That is, an individualwho has already decided to perform the tar-get behavior may experience a greater senseof cognitive commitment to the performanceof thai behavior than would an individualwho has already decided to perform a dif-ferent, though related, action. Consequently,the low-ball technique might be expected toproduce more performance of the target be-havior than the foot-in-the-door technique.To test this possibility, a field study wasconducted in which three sets of procedures—low-ball, foot-in-the-door, and control—wereused to induce subjects to perform a chari-table action.


Experiment 2Method

Subjects and procedure. Subjects were 30 malegraduate students who resided in dormitory roomsat a large state university and who were contactedin their rooms by a college-age male experimenterposing as a United Way worker.4 When subjectsanswered the experimenter's knock at their doors,the experimenter introduced himself uniformly asfollows:

Hi, my name is . _, and I'm working with theUnited Way. I'm going around asking people todisplay United Way posters for us.

At this point, the experimental script differed inaccord with the randomly alternated set of pro-cedures designed to obtain performance of the targetbehavior. The target behavior sought from eachsubject was the display of a pair of United Wayposters.

Control condition. Control subjects learned fromthe outset that agreeing to display the posters wouldrequire that they procure a "poster packet" at thedownstairs dorm desk within an hour of experi-mental contact. After the standard introduction, theexperimenter shuffled through his briefcase and an-nounced:

I don't have any posters with me now; I musthave given my last one to the last person I talkedwith. But there are packets at the dorm deskdownstairs which contain a window poster and adoor poster. They'll only be there for the nexthour; then they'll be taken to another area.Would you be willing to pick up a packet withinthe hour and put a poster on your window andone on the outside of your door and leave themup for a week?

Low-ball condition. Subjects experiencing thelow-ball procedures were first asked to display theposters. Those who agreed were then informed thatthey would have to pick up the posters downstairswithin an hour. Thus, after the standard introduc-tion, the experimenter continued, "Would you putup a pair of United Way posters?" If the subjectsaid, "No," he was thanked and the experimenterleft; if the subject said, "Yes," the experimentershuffled through his briefcase and announced:

I don't have any posters with me now; I musthave given my last one to the last person I talkedwith. But, if you still want to do this, there arepackets at the dorm desk downstairs which con-tain a window poster and a door poster.

The remainder of the script was identical to thatof the control condition.

Foot-in-the-door condition. Subjects in the foot-in-the-door condition initially complied with therequest to accept for display a window poster thatthe experimenter carried with him; they were then

asked to perform the more costly target behaviorrequiring a trip to the downstairs desk within thehour. Thus, after giving the standard introduction,the experimenter continued, "Would you put up aUnited Way window poster?" All subjects agreed andwere provided with a poster. "Thanks. We're alsoasking people to help by putting up a door poster."The experimenter shuffled through his briefcase andannounced:

I don't have any door posters with me now; Imust have given my last one to the last person Italked with. But there are packets at the dormdesk downstairs which contain a door poster.

The remainder of the script was identical to thecontrol condition.

Results

A high and approximately equal percentageof verbal compliance occurred across all con-ditions. With 10 subjects per group, 80% ofthe subjects in the low-ball condition and70% of the subjects in the other two condi-tions agreed to display the posters after beinginformed of the full cost of so doing. Ourmajor interest, however, concerned perform-ance of the target behavior, which was mea-sured the following day as the percentage ofsubjects' rooms displaying the posters. Itwas on this measure of behavioral compliancethat the superiority of the low-ball techniqueasserted itself clearly. We had hypothesizedthat the greatest performance of the targetbehavior would occur in the low-ball condi-tion. That expectation was supported by acontrast that pitted behavioral compliance inthe low-ball condition (6/10) against that inthe combination of the foot-in-the-door (I/10) and control (2/10) conditions; by Fisherexact test, p < .02. Further, the superiorityof the low-ball treatment maintained itselfwhen only those subjects who verbally com-plied were considered; that is, 75% (6/8) ofverbally compliant subjects in the low-ballcondition complied behaviorally, whereas sub-stantially fewer, 28.6% (2/7) and 14.3%(1/7), did so in the control and foot-in-the-

4 Graduate students were chosen because theywere the only dormitory residents who were allowedsingle rooms. The potentially contaminating effectsof the presence of a roommate on subjects' charitablebehavior were thereby eliminated.


door conditions, respectively; by Fisher exacttest, p < .025. It was also our expectationthat the foot-in-the-door condition subjectswould show more compliance than the controlcondition subjects. The difference betweenthose conditions, as tested by a secondorthogonal contrast, was clearly nonsignificantand slightly opposite to the predicted direc-tion.

Discussion

The results of Experiment 2 provide evi-dence of the effectiveness of the low-ball pro-cedure in a novel and naturalistic context.Charitable action, a form of behavior quiteunlike the target behaviors to which thetactic is standardly applied, was significantlyinfluenced by the technique in a setting thatdid not contain (for the subject) the experi-mental trappings present in Experiment 1.It appears, then, that the low-ball techniquehas some robustness across target behaviorsas well as power in naturalistic situations.

In addition to investigating the generalityof the low-ball tactic, a second purpose ofExperiment 2 was to determine whether andhow it differed from the foot-in-the-doortechnique. We hypothesized that more per-formance of the target behavior would occurfrom the low-ball strategy because it initiallyinduced subjects to decide to enact the targetbehavior itself rather than a different, thoughrelated, behavior. Such a decision was thoughtto produce a cognitive commitment to theperformance of the target behavior thatwould manifest itself in especially high levelsof compliance. Of the two types of compli-ance measured in the study, it was only inbehavioral compliance that the low-ball pro-cedure proved to be clearly superior. Thefailure of either experimental compliancetechnique to elicit substantially greaterverbal compliance than the control procedureis probably best interpreted as the result of aceiling effect caused by the very high levelof baseline verbal compliance (70%) thatoccurred in the control condition.

It is instructive, nonetheless, to examinethe relationship between verbal and behavioralcompliance among the three conditions of thedesign. The behavioral superiority of the low-

ball technique appears to be primarily due tothe significantly greater tendency of verballycompliant subjects in that condition to exe-cute the target behavior they had agreed toperform. One plausible explanation for theform of these data supports our prior hy-pothesizing and underscores the need for com-pliance researchers to obtain behavioral mea-sures of compliance. It is our feeling that ahigh percentage of the verbally compliantsubjects in the foot-in-the-door and controlconditions never intended to perform the tar-get behavior; since the full cost of the targetaction was known to these subjects beforethey were asked to agree to do it, manymay have privately decided not to performthe costly action but rather to provide onlythe impression that they would, in order toavoid immediate social disapproval. Thecomparable subjects of the low-ball proce-dures, on the other hand, were induced todecide to execute the target behavior whenit seemed to involve a minimal cost; theresultant cognitive commitment to the per-formance of the target action should haveexisted privately from the outset for them andmay have mediated the behavioral superi-ority of the technique. Whether or not theabove account is the correct one, the formof the results does indicate that the low-ballprocedure is not identical in effect to thefoot-in-the-door procedure, in that the twotechniques produced distinct data patterns.

Possible Mediators

To this point we have, for the most part,avoided a discussion of possible conceptualmediators of the low-ball effect, although wehave hinted at the causal relevance of cogni-tive commitment. Now that we have deter-mined via Experiments 1 and 2 that the low-ball phenomenon is both real and robust, itseems appropriate to turn our attention totheoretical implications. There are four for-mulations that could explain the low-ball effectthat an active decision to behave in a certainway tends to endure even after the behaviorhas become more costly to execute.

Behavior engulfs the field. The first for-mulation, suggested initially by Heider (19S8),is that a behavior (e.g., an active decision)


often exerts a disproportionally strong in-fluence in the cognitive arena and, conse-quently, on the direction of subsequent be-haviors: "It seems that behavior in particularhas such salient properties it tends to engulfthe total field rather than be confined to itsproper position as a local stimulus whoseinterpretation requires the additional data ofa surrounding field" (p. 54). Perhaps themost noteworthy illustration of how behaviormay "engulf the total field" and overwhelmthe influence of other relevant factors can beseen in the data of Jones and Harris (1967).Subjects who read an essay favorable to FidelCastro attributed a pro-Castro attitude tothe author of the essay, even when it wasclear that the author had no choice in beingassigned to write in favor of Castro. Jonesand Harris contended that in deciding onthe correct attitude to ascribe in the situa-tion, these subjects were unduly influencedby the perception that a pertinent behaviorhad been performed. This tendency forbehavior to overpower the effect of othersituational variables may account for the suc-cess of the low-ball technique. Once an indi-vidual has behaved by making an activedecision, the individual's perception that thedecisional behavior had occurred could swampthe influence of other relevant considerationssuch as subsequent changes in the cost ofimplementing the decision, thereby increasingthe probability that the decision would becarried out.

Self-perception theory. A related formula-tion, having more of the status of theory,also emphasizes the importance of behaviorin the determination of subsequent events.Self-perception theory (Bern, 1967) suggeststhat a person who perceives him- or herselfas freely behaving in a certain manner towardsome object will self-attribute an attitudetoward the object on the basis of that be-havior. One who selects chocolate ice creamin a shop that offers 31 flavors, then, shouldlogically infer strong positivity toward choco-late ice cream on the basis of that selection.This attributed positivity toward chocolateice cream could be expected to cause the indi-vidual to retain the decision, even if it werelater learned that chocolate ice cream wasslightly more expensive than originally

thought. An analogous process could explainthe effectiveness of the low-ball procedures inExperiments 1 and 2. The perception of thebehavioral decision to take specified actionmay have caused our subjects to self-at-tribute personal favorability toward theaction that would enhance the likelihoodthat the action would occur, even after ithad become more costly. A major distinctionbetween Heider's (19S8) notion of behaviorengulfing the field and self-perception theoryis that in the latter case the perception ofbehavior is seen as an important, but notoverwhelming, influence on the cognitive field;other factors such as volition play a crucialrole in the inference process. In contrast tothe notion that behavior engulfs the field,self-perception theory contends that an actionthat was performed without freedom ofchoice would have little impact on self-attri-butions.

Dissonance theory. It might be arguedthat dissonance theory (Festinger, 1957)could also explain the low-ball effect. That is,postdecisional dissonance resulting from theinitial decision to perform an action could beexpected to dispose the actor to become morefavorable toward the chosen action (e.g.,Davidson & Kiesler, 1964); that enhancedfavorability would then work to increase thechance that the action would be performed,even if it were rendered more costly by sub-sequent events. As with self-perceptiontheory, the variable of volition is a centralone in dissonance theory; little or no disso-nance would be generated by a decision madewithout high choice (Brehm & Cohen, 1962).In fact, the dissonance and self-perceptionformulations are sufficiently similar that, aswe will see, each makes identical predictionsconcerning the occurrence of low-ball effects.

Commitment. A final explanation for thelow-ball phenomenon lies in the commitmentformulation presented by Kiesler (1971).Kiesler suggests that a major function ofcommitment is to impart resistance to change.To the extent that one is committed to adecision, for instance, the decision will beless changeable; the decision itself and thecognitions representing it will be "frozen,"to use Lewin's (1947) terminology. Further,Kiesler contends that one way to bring about


commitment is through volitional action.According to this analysis, then, one whomakes a decision behaviorally and withchoice will be cognitively committed to thedecision and will be unlikely to change it,perhaps even after the circumstances thatbrought about the decision have themselveschanged to make the decision less appropri-ate. The results oi Deutsch and Gerard's(1955) classic conformity study support sucha possibility. They showed that subjects whohad committed themselves to a decisionthrough the mere act of writing it downanonymously were less inclined to modifythe decision when new evidence (i.e., theresponses of other persons) suggested thatthe decision might not have been a correctone. The applicability of the commitmentformulation to the low-ball tactic seemsstraightforward: An active initial decision tobehave positively toward some object willtend to make the decision persist, not neces-sarily because the decision will produce amore favorable attitude toward the objectin question, as self-perception and dissonancetheories would assert, but because the activedecision will create commitment, that is, aresistance to change that will tend to be im-pervious to the influence of subsequent dataconcerning the wisdom of the decision.

Factors Allowing jor Mediational Tests

Volition. In order to test the four poten-tial mediators of the low-ball effect, a thirdexperiment was considered necessary. Experi-ments 1 and 2 were not designed for thepurpose of differentiating among those possi-bilities and, hence, lacked several compo-nents necessary for a proper test. One suchcomponent would be a manipulation of voli-tion of decision. That is, one of the possibili-ties—Heider's (1958) notion of behaviorengulfing the field—would predict that thelow-ball phenomenon should occur when anindividual makes an initial, active decision,whether or not the decision was made witha high degree of choice. The perception thata behavior had been performed should, bythis account, work to trivialize the effect ofother relevant variables like volition. Recallthat exactly such a process seemed to occur

in the Jones and Harris (1967) study, andsimilar results have been obtained wheresubjects have rated themselves (e.g., Cial-dini, 1971). Each of the other three formu-lations, however, does stress the importanceof the factor of volition and would expectthe low-ball procedure to be effective onlywhen freedom of choice existed for the pre-liminary behavioral decision.

Attitude shifts. A second experimentalcomponent that would allow a discriminationamong the alternative formulations would bethe inclusion of an attitude measure. Two ofthe formulations—self-perception and disso-nance—argue that an individual who hasmade a favorable behavioral decision towardan object will tend to persevere in the deci-sion, even in the face of newly negative cir-cumstances, because the individual has cometo view the object more positively than be-fore as a result of the decision itself. Thus,these two theories would expect a low-balleffect to be accompanied by a mediating at-titude-change effect. Although the idea thatbehavior engulfs the field is nonpredictiveconcerning this point, the commitment formu-lation is clear in its expectation that a shiftin attitude would not accompany a simpleincrease in commitment. Kiesler (1971) hasargued that commitment itself does not pro-duce attitude change and has presented datato support his position (e.g., Kiesler & Saku-mura, 1966).

An alternate operationalization. A finalexperimental component that would allow usto assess the predictive power of the alterna-tive explanations for the effect would involvea somewhat different method of putting thelow-ball sequence into operation than wasused in Experiments 1 and 2. In those stud-ies, a subject was initially induced to agreeto perform an action, and the cost of theaction was then increased when the requesteradded some rather noxious further conditions.Although legitimate, such a way of enactingthe low-balling tactic does not differentiateamong the possible mediators of the phenome-non. Another, equally legitimate operational-ization, in which the requester initially ob-tains an agreement by describing specificpositive properties of the target action thatmake it more attractive than alternative ac-


tions and then increases the cost of the targetaction by removing those positive properties,would provide such a differentiation, how-ever. This latter method of operation can beseen to be comparable to that of the car-salescontext, where the customer is induced todecide to purchase a particular car on thebasis of a clear price advantage, which is thenremoved after the active decision has beenmade. Such a procedural sequence would al-low a test among the various explanatoryformulations because the clear advantageoffered initially by the requester provides awholly external justification for the decisionto perform the target action. Self-perceptionand dissonance theories would not expect thelow-ball effect to occur under such condi-tions. That is, since the target behavior wasinitially presented in a way that made iteasily the most attractive alternative, de-ciding to perform it would not result in post-decisional dissonance (Brehm, 1956) or in arevised self-perception of attitude (Bern,1967); consequently, without the mediatingeffect of initial attitude change, the decisionto perform the target behavior should not beexpected to persist after circumstances havechanged to make the behavior a more costlyone. Neither of the other two possible medi-ators of the phenomenon would make a simi-lar prediction: A behavioral decision wouldbe expected to engulf the field, and a behav-ioral decision would be expected to producecommitment whether or not the decision wasperfectly justified.

In order to provide evidence concerning theconceptual mediator of the low-ball phenome-non, a laboratory study was conducted thatincluded the above-described form of low-balling, an attitude-change measurement, anda manipulation of choice. The study con-tained three conditions, each of which re-quired subjects to make an initial decision totake one of two alternative personality tests.In the two experimental conditions, sub-jects were informed before the initial selec-tion that one of the tests would producetwice as much experimental credit as theother. In one of these experimental conditions(low-ball/high volition), the subjects weregiven free choice in making the initial selec-tion between the tests. In the other experi-

mental condition (low-ball/low volition), thesubjects were initially required to select thetest that provided the greater amount ofcredit. After the initial selection, subjects inboth experimental conditions recorded theirattitudes toward the alternatives; they thenlearned that the earlier credit informationhad been in error and that they would re-ceive the smaller amount of credit for takingeither personality test. At that point, sub-jects were allowed to decide again, freely inall cases, which test they wished to have ad-ministered to them. A control condition, inwhich subjects made an initial free selectionbetween the alternatives at the lower amountof credit production, provided the baselineinformation necessary to assess the effective-ness of the two low-ball conditions.

Each of the four possible explanations ofthe low-ball effect would make different pre-dictions in the experimental situation. Fromthe simple notion that behavior engulfs thefield, we should expect that subjects whoinitially selected the personality test provid-ing the larger amount of credit would remainwith that selection after the credit superiorityof the test had been eliminated; further,this would be the case whether or not theinitial selection was made with a high degreeof volition. The self-perception and disso-nance formulations would predict that sincethe initial decision for the test offering morecredit was wholly justified by the credit dif-ference, the experimental subjects should notexperience more favorable attitudes towardthe chosen alternative; consequently, thesesubjects should not be expected to maintaintheir initial decisions to any enhanced de-gree after the credit level of the selected testwas reduced to that of the nonselected test.The commitment formulation, on the otherhand, would expect that subjects who ini-tially selected the test offering the higheramount of credit would not experience posi-tive attitude shifts toward the selected testbut would, nevertheless, retain that decisionafter the test's credit advantage had beenremoved; additionally, the commitment inter-pretation would predict that this persever-ance of the first decision would result onlywhen subjects had been allowed free choicein making that decision.


MethodExperiment 3

Subjects. The subjects were 144 students of bothsexes enrolled in an introductory psychology coursethat required four hours of experimental participa-tion during the term. Subjects signed up for anexperiment concerning "social personality tests" thatadvertised a single hour of credit for their partici-pation.

Procedure. Subjects were met at the experimentalroom by one of three college-age experimenters whodescribed the experiment as concerning what stu-dents think of different kinds of personality tests.The experimenter then gave the subject a bookletand left the subject alone, with instructions to com-plete the booklet. The booklet contained shortdescriptions of two personality tests, the NationalInvolvement Scale and the California F Scale. Thedescriptions informed subjects what each test wassaid to measure and what a score on the test wouldindicate about one's personality. The booklet thenasked subjects to rate, along a series of three evalua-tive dimensions (informative, valuable, enjoyable),how attractive they would find taking each test andfinding out their score. At this point, the bookletsdiffered according to the condition to which thesubjects had been randomly assigned.

Low-ball conditions. After rating the personalitytests, low-ball subjects read that the National In-volvement Scale (the target test) would providethem with twice the authorized amount of creditbecause the experimenter needed more respondentson that test. Subjects were then allowed to selectbetween the two tests (high volition) or wereassigned to take the target test (low volition):

Since more prior research has been conducted hereon the California F Scale than on the NationalInvolvement Scale, we would like more people inthis experiment to take the National Involve-ment Scale. Thus, we are willing to give 2 hoursof experimental credit to subjects who take theNational Involvement Scale.

In the high volition cell, the instructions continued:

From the following personality tests, please circlethe one you have decided to take. Please under-stand that the choice is completely your own.

In the low volition cell, the instructions said:

Therefore, from the following personality tests,you have been assigned to take the test starred(*) below.

The National Involvement Scale was always thestarred alternative. After a selection had been madeor assigned, the experimenter returned and askedsubjects to rate the tests again, but this time on adifferent set of evaluative dimensions (meaningful,interesting, pleasant) than before. Pilot testing had

shown the two sets of dimensions to elicit fromsubjects comparable favorability scores concerningthe personality scales ( F < 1 ) . After the tests hadbeen rerated, the experimenter looked at the sub-ject's booklet, pretended to notice an error, andremoved the credit advantage of the target test:

Oh! Please disregard this statement in the book-let about getting 2 credits for taking the NationalInvolvement Scale. We were giving 2 credits lastterm when we needed a desired number of stu-dents to take that test. However, during thisterm we are offering only 1 credit for whichevertest is chosen. It makes no difference to us whichtest you choose, and they both take about thesame time to complete.

The experimenter then allowed subjects the chance tochange their test selection:

Now just to be sure, since the tests take the sametime and are worth the same credit, do youwant to stay with the test you have now, or doyou want to change to the other test?

At this point, subjects were allowed to decide be-tween the tests again.

Control condition. Control subjects were neverinformed that the personality tests differed in termsof experimental credit or in the experimenter'sdesire to administer them; consequently, after theirinitial selection of a test had occurred and theyhad rerated the tests, the experimenter merely al-lowed them the opportunity to change their deci-sion as he had in the low-ball conditions. In allother ways, the control condition was identical tothe low-ball/high volition condition.

All subjects then received a postexperimentalquestionnaire inquiring into (a) their perceptionthat the final selection of a test was their ownchoice, (b) their estimates of how they and mostothers would score on the personality tests, and(c) their hypotheses and suspicions concerning theexperiment. Finally, subjects received the person-ality test of their choice along with instructions onhow to score it and what a score was purported tomean. After subjects had taken and privately scoredtheir tests, they were debriefed and thanked.

Results

Postexperimental questionnaire. As a post-test check on the effects of suspicion of themanipulation or motives involved in the ex-periment, all subjects were asked to describe inwritten form what they thought the hypothe-sis of the experiment was and what suspicionsthey may have had. No subject was able tostate the hypothesis. None of the few sub-jects expressing suspicions concerning aspectsof the experiment said that their experimental


behaviors were affected by these doubts. Sub-jects indicated their feeling that their finalselection of a test was freely their own on a7-point scale ranging from "Not at all" (1)to "Completely" (7 ) . Subjects in all condi-tions reported equally high perceptions thatthe choice was their own (Ms = 5.9, 6.1, 6.0;F < I ) ; and the modal response in all condi-tions was "Completely." Thus, when thecrucial final test selection was made, it ap-pears that subjects in all conditions felt ahigh and equal degree of freedom in thatdecision.

Attitude ratings. Attractiveness ratings forthe two personality tests were taken beforeand after the initial selection was made.Neither an analysis of the pre- versus post-test change scores nor analysis of the post-test scores alone showed any significant atti-tude-change effects. There were no significantchanges in attractiveness of the tests as a re-sult of the initial selection in any condition;nor were there differences in attractiveness ofthe tests among the three conditions eitherbefore or after the initial selection (all Fs <2) . Despite the absence of attitude effects,there is evidence that the attraction scalesused were sensitive and valid. A point-biserialcorrelation showed that attraction to the testswas significantly related to the final behav-ioral selection of a test, r — .58, p < .001.This evidence is important, as it could other-wise be argued that a problem with the sensi-tivity of our attraction scales accounts for thelack of observed attitude change.

Behavioral data. Table 1 shows the per-centages of subjects choosing to take thetarget test, National Involvement Scale, bothat the initial decision point (when low-ballsubjects thought that test produced twice thestandard allotment of experimental credit) aswell as at the final decision point (when thecredit advantage of the target test had al-ready been removed). As the table indicates,the low-ball/high volition procedures weremore successful in inducing subjects to makean initial selection of the target test than con-trol procedures (81% vs. 31%), X

2U) = 24.4,p < .001. Of course, 100% of the low-ball/low volition subjects initially selected thetarget test, since it was assigned to them. The

Table 1Percentages of Initial and Final Selectionsof the Target Test

Initial selection Final selection

Low-ball/high volition 81 (39/48) 61 (29/48)

Low-ball/low volition Assigned (48/48) 42 (20/48)

Control 31 (15/48) 31 (15/48)

critical test of the low-ball technique's supe-riority, however, must be seen in the data ofthe final decision, when all subjects selectedfreely between the tests that by then hadbeen rendered comparable in credit produc-tion. Once again, the low-ball/high volitioncondition resulted in greater selection of thetarget test than the control condition (61%vs. 31%), x 2( l ) = 8.22, p < .005. The low-ball/low volition condition (42%), however,did not prove to be superior to the control con-dition, x 2 ( l ) = 1.12, ns.5 Further, the low-ball/high volition condition produced sub-stantially more complicance than its low voli-tion counterpart, x 2 ( l ) = 3.38, p - .065° No

5 Besides reflecting the importance of the percep-tion of choice in the effectiveness of low-balling, thepattern of these means also eliminates an alternativeexplanation of the basic effect that derives from thedemonstrated tendency of individuals to prefer toperform behaviors they had initially expected to per-form (Aronson, Carlsmith, & Darley, 1963). Hadthis expectancy interpretation accounted for theeffect, both of the low-ball conditions would havebeen significantly different from the control, sincesubjects in both low-ball conditions were given anequivalent initial expectation that they would takethe target test.

0 Although it might be argued that this differencebetween the low-ball conditions was due to a reac-tance tendency among the low volition subjects thatsuppressed selection of the target test, a close scrut-iny of the data does not support the possibility. Thelow-ball/low volition condition was different fromthe control condition on neither attraction towardthe tests nor final behavioral selection of the tests,as a reactance-based interpretation would predict.Further, the ratings of the perception of free choicein making the final behavioral selection indicatedthat low-ball/low volition subjects did not feelrestricted choice freedom, suggesting that reactancepressures were not present when the crucial selec-tion occurred.


significant effects due to sex of the subjectwere found.

General Discussion

Although each of the possible explanationswe have considered for the low-ball effectcould account for aspects of the data of Ex-periment 3, the findings are wholly consistentonly with the commitment interpretation.That is, only Kiesler's (1971) commitmentformulation would have predicted entirelythe obtained pattern of results that (a) thelow-ball technique would produce greater finalselection of the target activity than a con-trol treatment, (b) the technique would beeffective despite a lack of attitude-change ef-fects associated with the initial decision, and(c) the technique would only be more effec-tive than a control treatment when the initialdecision was made with free choice. Of course,it should be recalled that Experiment 3 wasdesigned specifically to eliminate the explana-tory relevance of dissonance and self-percep-tion theories. In other settings, such as thoseof Experiments 1 and 2 as well as a varietyof naturally occurring situations, dissonanceor self-perception factors may well play a rolein enhancing the efficacy of the low-ball tac-tic. Thus, while it appears that commitmentto perform an action is the determining con-dition for the occurrence of a low-ball effect,such a commitment might well produce aug-mented effects if it also engaged the actionof dissonance or self-perception principles.One might suspect, then, that the low-ballmethod used in Experiments 1 and 2 (i.e.,increasing the cost of a selected behavior),which allowed for the influence of those prin-ciples, would be more successful than theversion used in Experiment 3 (i.e., reducingthe benefits of a selected behavior), which didnot. A test of this hypothesis must await sub-sequent research wherein the two forms oflow-balling are assessed in comparable con-texts.

Together, the results of the three studiesindicate that the low-ball phenomenon is re-liable, robust, and mediated by a commitmentto an initial, uncoerced decision to perform abehavior. It is interesting to speculate as to

why the factor of volition plays a crucial rolein the technique's effectiveness if, as Experi-ment 3 showed, it does not necessarily resultin attitude shifts. Kiesler (1971) has sug-gested that the perception of choice carrieswith it the perception of responsibility. Thus,one who freely chooses to perform an actionshould feel responsible for the action. In fact,Kiesler goes on to suggest that the entirebasis of commitment may lie within the con-cept of self-responsibility: One is committedto something when one feels responsible for it.The reason we are less likely to change oncecommitted is that reversing what we feel re-sponsible for results in a variety of negativeself-perceptions (e.g., hastiness, a lack of in-telligence or judgment or appropriate cau-tion, etc.); consequently, we will resist suchchange. Another possibility is that situationsrequiring free choice between alternatives in-volve a certain amount of strain; that is,abnormally high degrees of information vigi-lance and awareness are necessary so thatthe proper decision can be made. Once aresolution has been attained, especially in theinstance of a large-scale decision like an auto-mobile purchase, we may find ourselves loathto destroy the resultant sense of completionand begin again the stressful task of process-ing relevant information. A test of the valid-ity of these speculations was, of course, be-yond the intended scope of the present re-search; nonetheless, they do offer plausibleunderlying conceptual frameworks withinwhich to place our results.

Conclusion

At the outset, we described a sequence forthe investigation of compliance-related fac-tors that was designed to increase confidencein the ecological validity of these factors. Thesequence suggests that a researcher shouldbegin with a period of observation, perhapseven participant observation, of the proce-dures, techniques, and tactics that are regu-larly used in natural compliance settings. Onthe assumption that the psychological factorsembodied in such common strategies are al-ready demonstrated to work well in real-world compliance contexts, doubts about the


power of one's conceptual variables to ac-count for more than trivial amounts of vari-ance in normal forms of behavior may bereduced. Variables for study, so selected,should then undergo experimental investiga-tion to determine their reliability, generality,and theoretical relevance. The present re-search used such an approach to examine thelow-ball technique, a tactic commonly usedto produce compliance in sales settings. Thetechnique was found to be a strong and re-liable motivator of compliance in a variety ofnonsales contexts. Further, the effectivenessof the tactic appeared to stem from its rela-tion to the psychological concept of commit-ment, which was seen to be greatly affected,and perhaps mediated, by a manipulation offree choice.

A Full Cycle Approach

The ecologically derived strategy for theidentification and examination of researchquestions is certainly not limited to thestudy of compliance behavior. Such an ap-proach could be applied to any of a numberof research domains in social psychology. Assuch, it should be noted that while there areobvious pragmatic advantages, the foremostconcern of this orientation is conceptualrather than applied. Our principal intent inlooking to the car-sales setting, for instance,was not to understand more fully how au-tomobiles are sold, although that was a posi-tive by-product. The primarily conceptualpurpose was implicit in the decision to ex-amine the low-ball technique in situationsother than the car showroom so that ques-tions of generality and conceptual mediationcould be adequately addressed. The naturalconsequence of such a decision was that manyof the rich set of features characteristic ofnew-car sales settings were lost; the com-pensation, of course, was the enhanced aware-ness of the cross-situational robustness of thelow-ball technique and of its relation to exist-ing social psychological theory.

A final point is that we would advocatethe use of naturally occurring instances notonly to identify variables suitable for experi-mental study but also to check on the validity

of the findings from that experimentation.For example, the present work appears toshow that the action of the variable of com-mitment was limited to a situation involvingthe perception of free choice. If we look backto the car-sales context that originally broughtthe low-ball technique to light for us, we cansee that a customer's perception of decisionalvolition is represented in that setting. Had itnot been -the case, we should have doubted theecological validity of our laboratory findings.What we are proposing, then, is a "full cycle"approach, wherein initial natural observa-tion gives direction to subsequent controlledexperimentation, the outcomes of which canthen be given external validation throughfurther natural observation. Systematic re-course to the evidence of the real world bothbefore and after the performance of experi-mental work may thereby reduce the extent towhich current social psychological researchcan be criticized as artificial and epiphe-nomenal.

Reference Note

1. Bickman, L. The gap between basic research find-ings and applications: Can it be closed? Paperpresented at the meeting of the Midwestern Psy-chological Association, Chicago, May 1976.

References

Aronson, E., Carlsmith, J. M., & Darley, J. M. Theeffects of expectancy on volunteering for an un-pleasant experience. Journal of Abnormal and So-cial Psychology, 1963, 66, 220-224.

Bern, D. J. Self-perception: An alternative interpreta-tion of cognitive dissonance phenomena. Psycho-logical Review, 1967, 74, 183-200.

Brehm, J. Postdecision changes in the desirability ofalternatives. Journal of Abnormal and Social Psy-chology, 1956, 52, 384-389.

Brehm, J., & Cohen, A. Explorations in cognitive dis-sonance. New York: Wiley, 1962.

Carlson, M. D. How to get your car repaired with-out getting gypped. New York: Harrow Books(Harper& Row), 1973.

Cialdini, R. B. Attitudinal advocacy in the verbalconditioner. Journal of Personality and Social Psy-chology, 1971, 17, 350-358.

Cialdini, R. B., & Schroeder, D. A. Increasing com-pliance by legitimizing paltry contributions: Wheneven a penny helps. Journal of Personality and So-cial Psychology, 1976, 34, 599-604.

Consumer Reports, May 1974, 39, 368.Davidson, J., & Kiesler, S. Cognitive behavior before


and after decisions. In L. Festinger (Ed.), Con-flict, decision, and dissonance. Stanford, Calif.:Stanford University Press, 1964.

Deutsch, M., & Gerard, H. B. A study of normativeand informational social influences upon individualjudgment. Journal of Abnormal and Social Psy-chology, 195S, 51, 629-636.

Festinger, L. A theory of cognitive dissonance.Stanford, Calif.: Stanford University Press, 1957.

Freedman, J. L., & Fraser, S. Compliance withoutpressure: The foot-in-the-door technique. Journalof Personality and Social Psychology, 1966, 4,195-202.

Heider, F. The psychology of interpersonal relations.New York: Wiley, 1958.

Helmreich, R. Applied social psychology: The un-fulfilled promise. Personality and Social Psychol-ogy Bulletin, 1975, .7, 548-560.

Jones, E. E., & Harris, V. E. The attribution of at-titudes. Journal of Experimental Social Psychology,1967, 3, 1-24.

Kiesler, C. A. The psychology of commitment. NewYork: Academic Press, 1971.

Kiesler, C. A., & Sakumura, J. A test of a model forcommitment. Journal oj Personality and SocialPsychology, 1966, 3, 349-353.

Lewin, K. Group decision and social change. In T.M. Newcomb & E. L. Hartley (Eds.), Readings insocial psychology. New York: Holt, 1947.

McGuire, J. W. The yin and yang of progress in so-cial psychology: Seven Koan. Journal of Personal-ity and Social Psychology, 1973, 26, 446-456.

Orne, M. T. On the social psychology of the psy-chology experiment: With particular reference todemand characteristics and their implications.American Psychologist, 1962, 17, 776-783.

Smith, M. B. Is experimental social psychology ad-vancing? Journal oj Experimental Social Psychol-ogy, 1972, 8, 86-96.

Received September 7, 1977 •

Hendrick Appointed Acting Editor ofthe Journal oj Personality and Social Psychology in 1979

The Publications and Communications Board of the American PsychologicalAssociation announces the appointment of Clyde Hendrick as Acting Editorof the Journal of Personality and Social Psychology in 1979. Hendrick suc-ceeds incumbent Editor Anthony Greenwald, who has resigned. As of July 1,1978, manuscripts should be directed to the Acting Editor:

Clyde HendrickDepartment of PsychologyBox 8185University of MiamiCoral Gables, FL 33124

Low-Ball Procedure for Producing Compliance: Commitment...

Documents

Transcript of Low-Ball Procedure for Producing Compliance: Commitment...