Threats to Internal and External Validity in the Field Setting

Alice M. Tybout, Northwestern University
Bobby J. Calder, Northwestern University
ABSTRACT - This paper discusses threats to internal validity which may occur when attempting to examine theories of human behavior in real world consumption settings. These threats and a methodology for overcoming them are illustrated by the examination of a field experiment in a consumption setting. Finally, limits to external validity which exist even in the field setting as the result of characteristics of both the research subjects and context are detailed.
[ to cite ]:
Alice M. Tybout and Bobby J. Calder (1977) ,"Threats to Internal and External Validity in the Field Setting", in NA - Advances in Consumer Research Volume 04, eds. William D. Perreault, Jr., Atlanta, GA : Association for Consumer Research, Pages: 5-10.

Advances in Consumer Research Volume 4, 1977   Pages 5-10

THREATS TO INTERNAL AND EXTERNAL VALIDITY IN THE FIELD SETTING

Alice M. Tybout, Northwestern University

Bobby J. Calder, Northwestern University

ABSTRACT -

This paper discusses threats to internal validity which may occur when attempting to examine theories of human behavior in real world consumption settings. These threats and a methodology for overcoming them are illustrated by the examination of a field experiment in a consumption setting. Finally, limits to external validity which exist even in the field setting as the result of characteristics of both the research subjects and context are detailed.

INTRODUCTION

Consumer researchers frequently draw on the theories and research of social scientists studying human behavior in their efforts to understand consumer behavior. Since the external validity of such theories is of critical concern, field experimental research testing these theories is done in consumption relevant settings. However, in moving to the field setting, the research may compromise internal validity. As a result, it may be concluded erroneously that a theoretical formulation does not apply in consumption settings. Furthermore, even if external validity is achieved without sacrificing internal validity, the consumption settings to which the research findings can be generalized must be delineated.

The purpose of this paper is twofold: (1) to identify and test a methodology for achieving external validity without compromising internal validity, and (2) to discuss and illustrate the limits of external validity in a field setting. The vehicle for achieving these objectives is a field experiment testing theories developed in the social psychology literature in a consumption setting. First, the relevant aspects of the experiment's purpose and procedure are briefly summarized. Then, a methodology for maintaining internal validity in a field setting which was developed and tested in the experiment is presented. Finally, limits to the external validity of the research findings resulting from characteristics of the field setting are discussed.

THE EXPERIMENT

Background

There are essentially two ways to influence individuals' behavior. One approach, persuasion, entails modifying behavior by modifying its cognitive precursors. Despite the vast literature indicating that persuasive appeals are successful in inducing attitude change, there is little evidence to suggest the efficacy of these strategies in changing individuals' behavior. Indeed, the predominant finding is that attitudes are not highly correlated with behavior (Wicker, 1969) and that attitude change does not necessarily cause behavior change (Fishbein and Ajzen, 1972).

An alternative approach entails the direct modification of behavior. The foot-in-the-door strategy operationalizes this approach. The foot-in-the-door technique entails inducing individuals to first comply with a small request, in the hope of increasing their compliance with subsequent larger requests. It has been demonstrated that this strategy is more effective in obtaining intentions to comply and actual compliance with the larger request than making only the larger request (Baron, 1972; Freedman and Fraser, 1966; Pliner, Hart, Kohl, and Saabi, 1974; Snyder and Cunningham, 1975). The efficacy of the foot-in-the-door technique has been explained in terms of attribution theory (Bem, 1972; Kelley, 1971); individuals reflect on their compliance with the small request and conclude that they must have a positive attitude toward such behaviors. This inferred attitude increases the probability of compliance with later requests.

Despite the evidence supporting the foot-in-the-door strategy, its applicability to numerous consumption situations is problematic. It has only been examined in contexts where the focal issue was familiar to experimental participants and was likely to have been considered socially beneficial by them (e.g., promoting safe driving, Freedman and Fraser, 1966; supporting a cancer drive, Pliner et al., 1974). It is unknown whether foot-in-the-door is effective in contexts where people are unfamiliar with the issue and where they would not necessarily perceive its benefits even if familiarized with them (e.g., situations involving the purchase of a new product or service). In such situations it is necessary that a spokesperson familiarize individuals with the issue and provide reasons to comply before foot-in-the-door can be used. Characteristics of this spokesperson may influence the effectiveness of the strategy.

The purpose of this experiment was to examine the relative efficacy of foot-in-the-door and straight persuasion strategies when the focal issue was the purchase of a new service and the credibility of the source communicating information about this service was varied. Subjects were introduced to an experimenter who gave a brief description of herself which manipulated source credibility (high, low), and were then shown an audio-visual communication describing the new service. Following the message, requests were made of subjects according to their experimental condition. Individuals in the foot-in-the-door treatment were asked to comply with a small request, while individuals assigned to the straight persuasion condition were not asked to comply with any request. Next, a moderate request was made of all individuals - they were asked to subscribe to the new service. Finally, subjects' attitudes toward the new service were measured.

On the basis of the extant literature, the following predictions were made. First, on the basis of previous findings (Freedmen and Fraser, 1966; Pliner et al., 1974; Snyder and Cunningham, 1975), foot-in-the-door was expected to result in greater "purchase" of the new service than straight persuasion. Second, significant interactions between source credibility and influence strategy were predicted. Based on the repeated finding of a source credibility main effect, it was expected that the high credibility source would induce significantly great subscription to the new service than the low credibility source when a straight persuasion strategy was employed (Hovland and Weiss, 1951; Kelman and Hovland, 1953; Miller and Basehart, 1969; Watts and McGuire, 1964;Whittakerand Meade, 1968). However, when the foot-in-the-door technique was employed, source was hypothesized to operate as a discounting cue, re-suiting in greater effectiveness when source credibility was low rather than high. This prediction was made on the basis of research which has demonstrated that the presence of plausible external causes, such as incentives, for compliance with a small request undermines attribution of the behavior to internal causes and results in low persistence of the behavior (Lepper, 1973; Scott, 1975). Additional support for the predicted interaction between level of source credibility and persuasion and foot-in-the-door influence strategies comes from Powell's (1965) findings. Powell reports that under straight persuasion, high credibility resulted in greater attitude change than low credibility. However, when the source made a voluntary request following the persuasive message (foot-in-the-door), low credibility resulted in greater attitude change than high credibility.

Research Issue

In 1975 the State of Illinois was preparing to institute a new alternative to the fee-for-service Medicaid prop gram which had historically met the public health care needs of Illinois recipients of Aid to Families with Dependent Children (AFDC). The new plan was a prepaid health program which would enable the recipient to receive comprehensive health care at one location and would stress preventive medicine. The prepaid plan also would reduce administrative costs to the State. To conform to the federal government's stipulation that the State offer both health care plans and allow public aid recipients to choose freely between them, a film which explained the characteristics of each of the two health care plans was developed. The experiment reported here was conducted in the field, just prior to official introduction of the new plan.

Method

Subjects. Three hundred and thirty female public aid recipients from the Chicago area served es subjects. These individuals were ones who were receiving AFDC and who came to the Western District Office of Public Aid to see their caseworker. They were already enrolled in the Medicaid or Green Card plan and qualified for membership in the new prepaid plan. The public mid recipients in the Western District were homogeneous in racial background, 96% were Black, and the majority had some high school education.

Procedure. Public aid recipients who wished to see their caseworker came to their District Office of Public Aid end registered with the receptionist in the waiting area. After an individual gave her name to the receptionist, she waited anywhere from fifteen minutes to several hours to see the caseworker and resolve her problems. During this waiting period, a research assistant approached all female public aid recipients (excluding the elderly who did not qualify for the plan) and asked them several screening questions to insure that they were eligible for the new plan. Eligible women were informed that a film was being shown in the District Office that day on a new health care plan that would soon be available to them and asked if they would be willing to step to the back of the waiting room for s few minutes to see it. The vast majority of eligible women approached agreed to view the film.

Women who agreed to see the film were escorted, one at a time, to the viewing cubicle and introduced to the experimenter. The experimenter gave a brief description of herself which manipulated source credibility (high, low). Following the source description, the subject was shown an 11-minute film with narration that described the new prepaid plan and compared it to the Green Card plan. After the film the experimenter-source stated her support for the new plan and then asked the subject to comply with a request appropriate to her experimental condition. Subjects in the foot-in-the-door conditions were asked to endorse the prepaid plan (small request). No small request was made of subjects in the straight persuasion condition. Then, the experimenter asked all subjects to comply with a moderate request by enrolling in the new prepaid plan.

Following the enrollment request, approximately one-third of the subjects in each treatment were administered a questionnaire by the research assistant measuring their attitudes toward both plans, their comprehension of the information in the film, and demographic characteristics. A manipulation check for the independent variable source was also administered. After the questionnaire was completed, subjects were told when official enrollment in the new plan would occur (it was not possible to actually enroll anyone at the time of the research), questions were answered, and subjects were dismissed. Subjects not receiving the questionnaire were informed of official enrollment procedures immediately after the enrollment request and were dismissed.

A METHODOLOGY FOR MAINTAINING VALIDITY IN A FIELD SETTING

It has been noted that conducting research in a field setting to increase the external validity of the findings may also increase threats to internal validity. As a result, if previous findings are not replicated in the field setting, two alternative explanations are plausible: (1) the theory is not any good, or (2) threats to internal validity present in the field setting made it impossible to test the theory adequately. A methodology for identifying and overcoming threats to internal validity is needed before theories can be tested appropriately in field settings. Here, the experiment described in the previous section is first used to illustrate the problem of alternative explanations for non-replication of previous research results end then employed to demonstrate a methodology for reducing threats to internal validity.

To examine the external validity of the finding in the social psychology literature that a direct behavior influence strategy based on self-perception theory, known as foot-in-the-door, is more effective than a strategy not focusing on behavior directly (i.e., persuasion), a field experiment in a consumption setting was conducted. This experiment investigated the effectiveness of foot-in-the-door end straight persuasion in obtaining enrollment in a new health care plan when the strategy source was of high or low credibility. On the basis of the theory and previous research it was anticipated that: (1) foot-in-the-door would result in a greater enrollment in the new health care plan than straight persuasion; and (2) that type of influence strategy and source credibility would interact in their effect on enrollment such that high credibility would be more effective than low credibility under persuasion and low credibility would be more effective than high credibility under foot-in-the-door.

The results for the experiment appear in Table 1 below.

TABLE 1

% YES RESPONSES TO THE ENROLLMENT REQUEST

Counter to the predictions, there was no significant main effect of influence strategy (c2 = 1.221, df = 1, p > .05) and no significant interaction of source credibility and influence strategy (c2 = 2.860, df - l, p > .05) [To enable an analysis of variance for the overall significant c2 between experimental treatment and enrollment (c2 = 11.250, df = 4, p <.05), an arc sin transformation was performed on the data in Table 1. (See Rao, 1951, p. 211-215, for discussion and illustration of this procedure.)]. In fact, the only significant effect was a main effect of source credibility which had not been predicted (c2 = 6.132, df = 1, p > .05).

At first glance the findings appear to suggest that self-perception theory is not a viable explanation for individuals' behavior in consumption settings. However, a closer examination is necessary to insure that the discrepancy between these results and those reported by other researchers is not attributable to threats to internal validity present in the field setting.

The prediction that foot-in-the-door would be more effective than persuasion was based on results of research where the two requests - small request and moderate request - employed in this strategy were made in two separate interactions.

In the present research both requests were made in the same interaction. In two other studies in non-consumption settings where both requests were made in the same interaction, no difference between foot-in-the-door and a condition where no small request was made was obtained (Cialdini, Cacioppo and Bassett, 1974; Cialdini, in review). Cialdini (in review) speculates that the lack of effectiveness of foot-in-the-door under these conditions may be attributable to inequity which individuals feel when a second request is made immediately after they have complied with the small request. Thus, the lack of difference between straight persuasion and foot-in-the-door appears to be the result of making both requests in one interaction rather than the field setting. Consequently, the theory may still be applicable in consumption settings provided that both requests are not made in the same interaction. In fact, Scott (1976) has demonstrated the appropriateness of foot-in-the-door in a consumption setting where the issue was an existing product and the two requests were made in separate contacts.

More damaging to the theory is the finding that the type of influence strategy and source credibility did not interact. Although it was predicted that high credibility would result in greater enrollment than low credibility under persuasion and low credibility would result in greater enrollment than high credibility under foot-in-the-door, high credibility was significantly more effective than low credibility in both instances. This finding is particularly detrimental to self-perception theory since it calls into question the belief-inference process underlying the theory. The prediction was based on the proposition that individuals reflect on their past behavior and its circumstances in an attempt to infer their internal states which, in turn, direct future behavior. When plausible external causes for the behavior exist (e.g., the high credibility of the source caused the compliance with the small request), the behavior will be attributed to these causes rather than internal states and will not persist (e.g., enrollment behavior not likely). However, when no plausible external causes for the behavior exist (e.g., the source is of low credibility and therefore not a likely cause for the behavior), the behavior will be attributed to internal states (e.g., liking for the new health care plan), and will persist (enrollment will be likely). The issue which must be addressed is whether or not there were any factors present in the field setting which threatened internal validity and thus made a test of the theory impossible.

Two factors in the field setting appear that may have influenced the relationship between the independent and dependent variables: (1) The low level of education and poor communication skills of the participants, and (2) the low salience of respondents' compliant behavior in the field setting. Although the majority of the participants had had some high school education, early phases of the research such as personal interviews and pre-experimental surveys indicated their communication skills were poor and they had difficulty understanding new concepts. As a result, it may have been necessary for the subjects to focus all of their attention on the message about the health care plans in order to comprehend them and they, therefore, may have had little opportunity to reflect on their own compliance with the small request and engage in the belief inference posited by self-perception theory. This explanation is consistent with the observation that research demonstrating the effectiveness of the foot-in-the-door has only been done using more highly educated subjects (e.g., Palo Alto residents, college students).

Alternatively, internal validity may have been threatened not as a result of subject characteristics, but rather as a function of the low salience of the compliant behavior in the field setting. In order to employ foot-in-the-door for a new service a lengthy message was required prior to the small request. As a result, the message rather than the individual's own compliance behavior with the small request may have been the salient cue and thus attributes of the message guided subsequent behavior.

In view of these two potential threats to internal validity it is difficult to draw any conclusions about the utility of the theory. However, if these threats had been anticipated, procedures to measure or control their effects could have been taken. Education could have been assessed and then treated as a covariate in the analysis of the effects of the independent variables on enrollment. The impact of the salience of compliance with the small request on the effectiveness of foot-in-the-door could be determined by treating this factor as an independent variable.

In fact, these threats to internal validity were anticipated and the measures suggested above implemented. Participants' educational backgrounds were determined during a post-experimental interview. Unfortunately, the uniformly low level of education prevented this variable from being a useful covariate in the analysis of enrollment. Additionally, a variant on the foot-in-the-door strategy, high salience foot-in-the-door was added to the experimental design. Individuals in this treatment followed the same procedure as those in the foot-in-the-door condition except that after they complied with the small request their compliant behavior was made salient by asking them to explain why they complied. The salience induction both insured that participants focused on their behavior and provided data which could be used to check the hypothesized attribution of behavior to internal causes under low credibility and external causes under high credibility. These modifications enabled retaining internal validity while gaining external validity from the field setting. If the self-perception theory predictions were not upheld in this high salience foot-in-the-door group, it would be quite damaging to the theory.

The results for all three types of influence strategies appear in Table 2.

TABLE 2

% YES RESPONSES TO THE ENROLLMENT REQUEST

The interaction between source credibility and influence strategy was highly significant when high salience font-in-the-door was included in the analysis (_c2 = 11.981, df = 2, p < .01). More specifically, consistent with self perception theory, low credibility was significantly more effective than high credibility under high salience foot-in-the-door (Z = 2.049, p < .05).

It was hypothesized that low credibility would he more effective than high under foot-in-the-door because high credibility would serve as a discounting cue for internal attribution of compliance with the small request. The accuracy of this explanation was examined by analyzing subjects' responses to the salience induction which asked them to explain why they had complied with the small request. The hypothesis was supported. Individuals in the high credibility, high salience foot-in-the-door condition attributed their compliance with the small request to external causes to a significantly greater extent than those in the low credibility, high salience foot-in-the-door condition (8/26 high credibility vs. 1/19 low credibility; Z = 2.090, p < .05).

In summary, the results for the high salience foot-in-the-door group provide strong support for self-perception theory in a consumption setting. Further research is needed to determine whether a salience induction is necessary for the self-perception process to occur in all situations where the issue is a new product or service or only when the participants have little education. However, this conclusion is dramatically different from one which might have been made if potential threats to internal validity in the field setting had been ignored and no high salience foot-in-the-door condition included in the design.

When moving to the field setting the researcher must carefully examine that setting for threats to internal validity. After identifying any potential threats to internal validity, efforts can be made to measure and/or control them. Variables which cannot be controlled by the researcher such as demographic and personality characteristics can be measured and then used as covariates in the analysis of relationships between the independent and dependent variables. Controllable variables identified as likely to mediate the effect of. the independent variables may themselves be included in the design as independent variables. In addition, manipulation checks may be included in the research to insure that the effects of the manipulations are not swamped by factors in the field setting. Only when threats to internal validity have been eliminated can a field setting provide a strong test of a theory.

THE LIMITS TO EXTERNAL VALIDITY

When problems of internal validity are overcome, it is often assumed the results obtained in a field setting have broad ecological validity. In fact, the generalizability of such findings is often limited by characteristics of both the research sample and context. A careful analysis of these factors is necessary to delineate the settings to which research findings apply. To illustrate this point, the impact of the research subjects and the research context on the external validity of the field experiment presented in this paper are discussed.

Research Subjects

Participants in the field experiment were Black, female, welfare recipients who agreed to view a health care film while waiting to see their caseworker in a district office of public aid. The applicability of the research findings to different samples is unknown since characteristics, such as race, sex, socio-economic background and willingness to view the film, may all be related to the nature and magnitude of the effects obtained. For example, it is conceivable that the results obtained are only valid for females because (say) they are more sensitive than males to external cues for their behavior such as source credibility. Alternatively, the finding that a salience induction was necessary in the foot-in-the-door condition to insure that subjects used their compliance with the small request and its circumstances to infer internal states and direct future behavior may be attributable to the low education and poor communication skills of the research participants. When more sophisticated individuals serve as subjects the support for self-perception theory obtained in the high salience foot-in-the-door condition may be obtained without the salience induction.

Finally, the applicability of the results obtained may be limited to Black female welfare recipients who are not negatively predisposed toward either the new health care plan in particular, or state programs in general. It was observed in recruiting participants for the study that a small group of individuals who had either heard negative things about the new health care plan from the media, doctors, family, friends, etc., or who had a generally hostile attitude toward State-sponsored services, refused to participate in the study. This hostility might well influence the effectiveness of the strategies examined in this study. For example, hostile individuals might be unwilling to accept the small request of endorsing the new health care plan employed in foot-in-the-door condition. As a result, when they reflected on their refusal of this request, they would infer that they must have refused because of their dislike for the plan. This negative attitude would, in turn, guide their response to the enrollment request, thereby resulting in a decrease in enrollment relative to straight persuasion, rather than the no difference between foot-in-the-door and straight persuasion observed with the current sample.

Research Context

The research addressed enrollment in a new health care plan by public aid recipients and was conducted in a district office of public aid by white, State employed, researchers. Several aspects of this research setting may limit the generality of the findings obtained. First, the issue was of considerable consequence to the public aid recipients. If they chose to enroll in the new health care plan, they would be committed to receiving all their health care at a specified clinic for at least one year. In most cases this meant that enrollees would have to discontinue seeing their current doctors. The magnitude of this decision may have resulted in subjects carefully attending the cues in the environment (e.g., source credibility, compliance with the small request) in an effort to determine the appropriate decision. In situations where the new product or service is of little consequence to the participants these factors may not influence behavior.

Second, the physical environment in which the research was conducted contained numerous distractions. There was noise from children playing and adults conversing in the waiting room. In addition, mothers who participated in the experiment often brought their children with them to the viewing room and the children talked, squirmed, etc. While these distractions increase the external validity of the findings for this setting since they will continue to exist during the actual introduction of and enrollment in the new health care plan, they may limit the generalizability of the findings to settings where distraction does not exist. For example, it could be argued that distraction increased the effect of source credibility obtained in the persuasion and foot-in-the-door conditions.

This contention is based on cognitive dissonance which states that the greater the effort expended to engage in a counterattitudinal behavior, the greater the dissonance (Zimbardo, 1965). It is reasoned that choosing to hear a counterattitudinal attack by a low credibility source may actually be consonant with an individual's belief (i.e., it is positive to have a low credibility individual disagree with you). Therefore, expending effort to hear such an attack may heighten commitment to one's own beliefs. Conversely, choosing to listen to a counterattitudinal message from a high credibility source should evoke dissonance, which would increase when effort is expended to do so. Thus, distraction should increase persuasion under conditions of high credibility and decrease persuasion when credibility is Low.

However, the effort explanation assumes that the behavior, in this case viewing the film on the prepaid plan, is counterattitudinal. Subjects who participated in the present research tended to have no initial opinion toward the plan because they lacked knowledge of it. Baron, Baron, and Miller (1973) do suggest that the effort hypothesis might he extended to predict increased attitudinal extremity among pro-attitudinal subjects who are distracted. Thus, it is conceivable that distraction may have enhanced persuasion in this research, particularly when source credibility was high, and consequently these findings might not hold when distraction is reduced.

Finally, the fact that the researchers were white and apparently employed by the State may limit the generality of findings. Public aid recipients are in a low power position relative to. the State and may have enrolled to please State employees who stated they believed the plan was a good one. The results may be altered in situations where the communicator has little power over his target.

In summary, while the field setting significantly improved external validity by reducing demand character -subjects were unaware they were participating in an experiment and therefore unlikely to search for clues regarding the experimental hypotheses or attempt to confirm or disconfirm hypotheses - and incorporating real world factors in the setting (e.g., distractions), the ecological validity of the findings is still limited by characteristics of the research subjects and content.

REFERENCES

Robert A. Baron, "The Foot-in-the-Door' Phenomenon: Mediating Effects of the Size of the First Request," Bulletin of Psychonomic Society, 2 (1973), 113-114.

R. Baron, P. Baron and N. Miller, "The Relation Between Distraction and Persuasion," Psychological Bulletin, 80 (1973), 310-323.

Daryl J. Bem, "Self-Perception Theory," in L. Berkowitz (ed.), Advances in Experimental Social Psychology, (New York: Academic Press, 1972), 1-62.

Robert B. Cialdini, "A Test of Two Techniques for Inducing Verbal, Behavioral, and Further Compliance with a Request to Give Blond," working paper, Arizona State University, in review.

Robert B. Cialdini, J. T. Cacioppo, and R. Bassett, "Throwing the Low-Ball for Charity: A Field Study," paper presented at the meeting of the Midwestern psychological Association, Chicago, 1974.

Robert B. Cialdini, Joyce E. Vincent, Stephen K. Lewis, Jose Catalan, Diane Wheeler, and Betty Lee Darby, "Reciprocal Concessions Procedure for Inducing Compliance: The Door-in-the-Face Technique," Journal of Personality and Social Psychology, 31 (1975).

Martin Fishbein and Icek Ajzen, "Attitudes and Opinions," Annual Review of Psychology (1972), 487-544.

Jonathan L. Freedmen and Scott C. Fraser, "Compliance Without Pressure: The Foot-in-the-Door Technique," Journal of Personality and Social Psychology, 4 (1966), 195-202.

C. I. Hovland and W. Weiss, "The Influence of Source Credibility on Communicator Effectiveness," Public Opinion Quarterly, 15 (1951), 635-650.

Harold H. Kelley, "The Processes of Causal Attribution," American Psychologist, 28 (1973), 107-128.

H. C. Kelman and C. I. Hovland, "Reinstatement of the Communicator in Delayed Measurement of Opinion Change," Journal of Abnormal and Social Psychology, 48 (1963), 327-335.

Mark R. Lepper, "Dissonance, Self-Perception, and Honesty in Children," Journal of Personality and Social Psychology, 25 (1973), 65-74.

G. R. Miller and J. Basehart, "Source Trustworthiness, Opinionated Statements, and Response to Persuasive Communication,', Speech Monographs, 36 (1969), 1-7.

Patricia Pliner, Heather Hart, Joanne Kohl, and Dory Saabi, "Compliance Without Pressure: Some Further Data on the Foot-in-the-Door Technique," Journal of Experimental Social Psychology, 10 (1974), 17-22.

Frederick A. Powell, "Source Credibility and Behavioral Compliance as Determinants of Attitude Change,"

Journal of Personality and Social Psychology, 2 (1965), 669-676.

C. R. Rao, Advanced Statistical Methods in Biometric Research (New York: Wiley, 1952).

Carol Scott, "Past Behavior, Incentives and Maintained Behavior Change: A Field Study of Self Attribution Theory," unpublished doctoral dissertation, Northwestern University, 1975.

Mark Snyder and Michael R. Cunningham, "To Comply or Not Comply: Testing the Self-Perception Explanation of the 'Foot-in-the-Door' Phenomenon," Journal of Personality and Social Psychology, 31 (1975), 64-67.

Alice Tybout, "The Relative Effectiveness of Three Social Influence Strategies in a Novel Situation: Foot-in-the-Door, Door-in-the-Face, and Persuasion," in review.

W. Watts and W. J. McGuire, "Persistence of Induced Opinion Change and Retention of the Inducing Message Contents," Journal of Abnormal and Social Psychology, 68 (1964), 233-241.

J. O. Whittaker, and R. D. Meade, "Retention of Opinion Change as a Function of Differential Source Credibility," International Journal of Psychology, 3 (1968), 103-108.

Allen W. Wicker, "Attitudes Versus Actions: The Relationship of Verbal and Overt Behavioral Responses to Attitude Objects," The Journal of Social Issues, 15 (1969), 41-78.

P. G. Zimbardo, "The Effect of Effort and Improvisation on Self-Persuasion Produced by Role Playing," Journal of Experimental Social Psychology, 1 (1965), 103-120.

----------------------------------------