Two More Self-Prophecy Experiments

ABSTRACT - The self-prophecy effectCa predictable and robust behavior modification techniqueCsimply requires an individual to make a prediction about some behavior, with this prediction increasing their likelihood of performing that behavior in the future. This article extends the self-prophecy literature with two new studies. Study 1 is a field experiment applying the self-prophecy effect to recycling behavior, the results of which suggest promise concerning application of the effect in this domain. As with several self-prophecy studies, however, the measurement of real world behaviors such as recycling is both time consuming and quite expensive. Study 2 explores, but fails to find support for, a self-prophecy effect with a surrogate measure of behavior (i.e., behavioral self-reports).

Citation:

David E. Sprott, Eric R. Spangenberg, and Andrew W. Perkins (1999) ,"Two More Self-Prophecy Experiments", in NA - Advances in Consumer Research Volume 26, eds. Eric J. Arnould and Linda M. Scott, Provo, UT : Association for Consumer Research, Pages: 621-626.

Advances in Consumer Research Volume 26, 1999      Pages 621-626

TWO MORE SELF-PROPHECY EXPERIMENTS

David E. Sprott, Washington State University

Eric R. Spangenberg, Washington State University

Andrew W. Perkins, University of Washington

[The authors thank the Recycling Program, participating dormitories, and undergraduate research assistants (Becky, Chris, Dave, Jen, Mark, Matt, Sean, Tamara, and Terri) at Washington State University-without their help, this paper would not have been possible. Special thanks are extended to David Knuff for his diligent data collection and entry efforts.]

ABSTRACT -

The self-prophecy effectCa predictable and robust behavior modification techniqueCsimply requires an individual to make a prediction about some behavior, with this prediction increasing their likelihood of performing that behavior in the future. This article extends the self-prophecy literature with two new studies. Study 1 is a field experiment applying the self-prophecy effect to recycling behavior, the results of which suggest promise concerning application of the effect in this domain. As with several self-prophecy studies, however, the measurement of real world behaviors such as recycling is both time consuming and quite expensive. Study 2 explores, but fails to find support for, a self-prophecy effect with a surrogate measure of behavior (i.e., behavioral self-reports).

Prior research has demonstrated that simply predicting one’s future behavior, in itself, can influence the outcome of that behavior (e.g., Sherman 1980). Originally labeled the "self-erasing nature of errors of prediction" (Sherman 1980), this "self-prophecy" phenomenon should have important implications for society in its application. [Sherman's (1980) term is based on the observation that individuals' behavioral predictions are generally different from normally observed actions and, thus, can be considered in error. Because this prediction alters an individual's behavior (such that the prediction and future behavior are consistent), the originally observed "error" is erased by the future behavior. We adopt the term "self-prophecy," from earlier research by Greenwald et al. (1988), in that it captures the essence of the phenomenon in a more simple and strightforward manner.] Spangenberg (1997), for example, demostrated the technique’s ability to increase exercising by individuals at a health club over a 6-month period. One characteristic that makes this influence technique so compelling is its straightforward application: as Greenwald et al. (1987, p. 315) note, it simply "involves asking people to predict whether they will perform the target action."

Prior research suggests that self-prophecy could be a useful strategy for influencing socially relevant behaviors like curbing drug use, increasing volunteer labor participation (Sherman 1980), and reducing intolerant behaviors based on stereotypes (Spangenberg and Greenwald 1999). In that only 14 known studies have been conducted to date, questions remain regarding the range of applicability for this technique. Without a clear understanding of the types of behaviors for which the effect occurs and under what conditions it operates, researchers in the area are faced with considerable opportunity.

This article contributes to self-prophecy research with two new studies. Study 1Ca field experiment conducted in a university dormitory systemCapplies the self-prophecy effect to recycling behavior in a natural setting. Study 2 explores whether a self-report measure of recycling, among other behaviors, can serve as a surrogate indicator of actual behavior when assessing the effects of self-prediction.

SELF-PROPHECY: AN OVERVIEW

The self-prophecy effect provides a simple influence strategyCmerely ask individuals to predict their behavior regarding a target action, and the likelihood of performing the behavior will significantly increase. As noted above, the effect offers great promise for society in its ability to alter the performance of socially relevant behaviors. This position is strengthened by a recent meta-analysis of the literature (Spangenberg and Greenwald 1999) that found the self-prophecy effect to be of small to moderate magnitude (per Cohen 1988) for the eleven known studies reviewed (the average effect size was r=.134). By deleting two studiesCwhose authors provided valid methodological and theoretical reasons for their failures to detect self-prophecy effectsCthe average effect size increased to r=.214.

Of interest in the present context is that the self-prophecy effect has been demonstrated in numerous #real life’ settings of substantive importance. For example, the self-prophecy effect (elicited by a single prediction request question) has been shown to: (1) increase voting in an election (Greenwald et al. 1987), (2) reduce cheating on a take-home college exam (Spangenberg and Obermiller 1997), and (3) increase attendance at a health club (Spangenberg 1997). [Other self-prophecy studies have been conducted in natural settings that have found inconsistent effects of the manipulation. These include registering to vote in Experiment 1 of Greenwald et al. (1987), voting behavior in Greenwald et al. (1988), and pledging for a university fund raiser in Obermiller, Spangenberg, and Atwood (1992). Detailed review of these and other studies is provided by Spengenberg and Greenwald (1999).]

Representative of this stream of research is Spangenberg’s (1997) study of self-prophecy effects on health club attendance. A sample of registered members at a local health club was contacted by phone and administered a self-prophecy manipulation. There was no effect of the manipulation on health club attendance during the ten-day period following contact; there was, however, a significant change in behavior over the six months after contact. The average number of visits for members making a prediction (M=10.25) was twice that of members making no prediction (M=5.10; F_(1,93)=3.78, p=0.05).

While previous self-prophecy studies have made progress in establishing the effect’s robustness, to date, no cogent theoretical account has been offered for the effect. In fact, only recently have attempts been made to determine the conditions under which the effect operates (cf. Spangenberg and Greenwald 1999). A review of extant research indicates that the effect has been demonstrated primarily when target behaviors (e.g., voting, cheating, exercising, etc.) have a socially normative component (Spangenberg and Greenwald 1999). Although social desirability may have facilitated the occurrence o the self-prophecy effect, it is unclear whether it is a necessary condition for it to be manifested.

There are a number of potential theoretical accounts of the self-prophecy effect, the most compelling of which are script evocation (cf. Sherman 1980) and cognitive dissonance (e.g., Festinger 1957). The script evocation account assumes that the prediction request forms a "good person script" that once activated guides performance of the future behavior. Alternatively, self-prophecy has a number of similarities with research demonstrating cognitive dissonance effects. Aronson (1992) has provided a theoretical interpretation of cognitive dissonanceCi.e., induced hypocrisyCthat is useful to consider when discussing self-prophecy (e.g., Stone et al. 1994). In this research, subjects are prompted to advocate a socially desirable action (e.g., condom usage and AIDS prevention). Subjects are reminded of failing to perform such behaviors in their past and then given the opportunity to perform the behavior in the present. Behaviors are changed in the socially desirable direction in an attempt, according to Stone et al. (1994), to reduce cognitive dissonance. The self-prophecy effect may operate in a similar fashions, such that subjects are motivated to diminish the "values-action discrepancy" made salient by the prediction request accompanied with the reminder of subjects’ past failures to perform the behavior in the predicted direction (Spangenberg and Greenwald 1999). A more detailed discussion of these and other competing theories is provided by Spangenberg and Greenwald (1999), unfortunately, published research has not exhaustively examined either the limiting conditions or theoretical accounts for the effect.

The self-prophecy effect holds important implications for a variety of substantive areas, especially considering its manifestation in natural settings approximating moderate-sized effects (Spangenberg and Greenwald 1999). Indeed, and as suggested above, the self-prophecy effect should hold for most settings where a clear social norm exists. However, this conjecture can only be validated with additional empirical evidence demonstrating the effect’s ability to manifest in a variety of settings. Study 1 of this article was designed to further this end by employing self-prophecy in a yet to be investigated contextCrecycling. Before providing details, we set the stage for Study 1 by briefly reviewing other research aimed at altering consumer recycling behavior.

RECYCLING BEHAVIOR MODIFICATION RESEARCH

Prior research examining strategies for altering recycling behavior can be broadly categorized in four groups: (1) fear, (2) material incentives, (3) societal recognition, and (4) public commitment. Comprehensive reviews of this literature are provided by Cook and Berrenberg (1981) and Bryce, Day and Olney (1997).

Fear. Many studies have focused on using fear to increase arousal and subsequent action toward environmental ends. For example, Higbee (1969) notes that evidence found in most studies indicates that the level of fear increases, subjects’ attitudes and behaviors increase concomitantly. If an event is perceived as a crisis on a societal level, for example, individuals are more likely to participate in group-sponsored prevention behavior, but less likely to undertake individual measures (Rogers 1975). It is theorized that this effect occurs because individuals do not believe that individual effort will be effective in addressing the problem, making group- or societal-level behavior the only feasible option.

Material Incentives (Disincentives). Previous research has shown that monetary incentives or punishments can affect recycling or energy consumption behavior. Based on behavior reinforcement theory, behavior leading to a reward (punishment) will be repeated (extinguished). Unfortunately, it seems that the effects of threats of fines or other punishment on consumption behavior is comple and unpredictable, and often short lived. Bandura (1969) notes that the unpredictability of response may be due to the temporary nature of the punishment or the individual’s perceived ability to avoid the source of the punishment. Conversely, there is evidence supporting a relationship between financial incentives and behavior modification. This effect, however, is likely to be moderated by the perceived size of the incentive when compared to the cost of the behavior, price of the resource involved, and/or the socio-economic status of the individual (e.g., Cook and Berrenberg 1981). Overall, attempting to modify behavior using either incentives or disincentives may have some effect, but the size and duration of said effect is generally indeterminate.

Societal Recognition. The use of societal recognition has been studied extensively in the conservation literature. Seaver and Patterson (1976), for example, found that visible social recognition (in the form of a window sticker) significantly lowered the amount of oil consumed by participating homes. While inconclusive, results from this research indicate that social recognition plays at least some role in inducing conservation behavior. Although not directed toward recycling, results in this area of research suggest that social norms play a part in motivating environmentally conscious behavior.

Public Commitment. Research has also focused on using public commitment to influence recycling behavior. For example, Wang and Katzev (1990) found that consumers who made a four-week commitment to recycle continued to do so even after the four-week period was up, and recycled significantly more paper than consumers who had received incentives to recycle (see also Katzev and Pardini 1987-1988). Maloney and Ward (1973) found that actual commitment was significantly related to verbal commitment, depending on the personal characteristics of the individuals involved in the study. [This research is differentiated from self-prophecy studies in two important ways. In self-prophecy, (1) there is a single anonymous prediction and (2) subjects don't know that their behavior is being monitored subsequent to prediction.]

In summary, a variety of techniques has been used attempting to influence recycling behaviors of individuals. While these techniques have resulted in varying levels of success, it is clear that additional influence techniques could be useful in this area of substantive importance. Study 1 of the present research tests one such alternate method for affecting recycling behavior. Specifically, the study explores self-prophecy in a large field experiment conducted in a university dormitory setting.

STUDY 1

Method

Experimental Design and Sample. Study 1 was conducted as a between-subjects experimental design. Twenty-seven floors of five dormitories at Washington State University were randomly assigned to two experimental conditions (prediction and no prediction) in such a manner that a relatively equal number of floors in each dorm appeared in both conditions. The self-prophecy manipulationCa survey (described below) either incorporating a question requesting a prediction of recycling behavior, or notCwas administered to residents on participating floors.

The percentage of dorm residents completing the survey varied by floor and ranged from approximately fifteen to one hundred percent. In that a significant proportion of residents needed to be contacted to provide the opportunity for a self-prophecy effect to emerge, floors with less than sixty percent contact rate (n=13) were dropped from the sample. The final sample, therefore, included fourteen floors (seven in each experimental condition) containing approximately 326 residents (148 in the prediction condition and 178 in the no prediction condition). The number of dorm residents per floor completing the survey was equivalent between conditions (p=.72).

Self-Prophecy Manipulation. The self-prophecy manipulation (via the administration of a survey) was conducted throug early semester floor meetings or on a door-to-door basis during the first three weeks of the Spring 1997 semester. Specifically, dorm residents were asked to complete a one-page survey (ostensibly a graduate sociology study) requesting them to make four forced-choice decisions based on written scenarios. Three scenarios were consistent across experimental conditions and included making predictions about (1) taking one of two jobs with differing salaries and advancement opportunities, (2) eating at either an American or local restaurant while in Europe, and (3) reporting or not reporting an observed shoplifter (cf. Spangenberg and Obermiller 1997).

In the prediction condition, the survey also included a scenario (appearing in the second position) regarding recycling. The scenario requested subjects to predict future recycling behavior using the following format:

You are using a lot of products that come in containers/packages/cans that can be recycled.

Q. Do you predict that

You will not recycle

You will recycle

Order of prediction (i.e., will or will not recycle) was counterbalanced to preclude order effects. In the no prediction condition, the recycling scenario was replaced with one concerning tipping (or not tipping) the server in an expensive restaurant with poor service.

Procedures and Dependent Measure. [Although data were collected for both can and paper recycling, only can results are reported here. The paper data were unstable due to a number of factors including unreliable measurement (e.g., students had other paper disposal/recycling opportunities that may have been more attractive) and uncontrollable differences across experimental conditions (e.g., some floors had subscriptions to the local daily paper, while others did not).] Variation existed in the dorm recycling programs prior to start of the studyCsome floors had recycling while others had none. To promote consistency across participating dormitories, a unified recycling program was instituted and presented (independent of the self-prophecy manipulation) to dorm residents as a new program. The new program was different from those already in existence before launch of our study. The program included an announcement at the beginning of the semester informing all dorm residents about the new recycling program and provision of recycling facilities (receptacles for cans and paper) at a central location on each floor.

Data were collected in each dorm on Friday mornings throughout the semester by a supervised team of undergraduate research assistants. The dependent measure for Study 1 was the number of aluminum cans placed in the recycling bin on a given floor. Measurements were verified by at least two research assistants.

Analysis and Results

Analysis was conducted on the dorms’ recycling behavior between weeks four and thirteen during the 15-week semester. The first three and last two weeks of the semester were not included in the analysis. Experimental manipulations were administered to participating dorms in the first three weeks of the term and break down of the recycling project commenced in week 14. The Figure presents weekly recycling data by condition.

Consistent with self-prophecy, the total number of cans recycled by dormitory floors making a prediction (M=30.53, s.d.=19.57) was higher than for those floors making no prediction (M=18.11, s.d.=9.98). This difference approached significance (t₍₁₂₎=1.50; p_1-tail=0.08) and, per Cohen (1988) approximated a medium to large effect size (r=0.40).

Follow-up analyses found additional differences between the two experimental conditions. Specifically, contact rate was marginally higher for the predict condition (M=0.83) than for the no predict condition (M=0.70; t₍₁₂₎=2.03; p=0.07). In addition, the aproximate floor size (i.e., number of residents per floor) was slightly smaller for those floors in the predict condition (M=21.14) than for those floors in the no predict condition (M=25.43; t₍₁₂₎=1.51; p=0.16). To ascertain the effects of these factors, the variables were included in separate ANCOVA models. While contact rate had no effect (t₍₁₁₎=0.99; p=0.35), the floor size covariate was significant (t₍₁₁₎=2.15; p=0.05). With floor size as a covariate, the self-prophecy effect was stronger (F_(1,11)=5.89; p=0.03), such that the number of recycled cans was higher in the predict condition (M_Adjusted=33.94) than in the no predict condition (M_Adjusted=14.71).

Discussion

The results of Study 1 are encouraging in that they show a moderate- to large-sized effect of self-prophecy on recycling of aluminum cans. When examined in light of other attempts to manipulate environmentally conscious behavior, self-prophecy has some benefits. When compared to other programs (as previously reviewed) the self-prophecy manipulation is simple and inexpensive to administer, does not require reinforcement nor public recognition, and is harmless to the subject, both financially and socially.

Once the self-prophecy effect has been demonstrated in a given domain, future implementation of the technique could be as simple as asking a prediction question of a target group of individuals. Unfortunately, difficulty arises if one must assess the actual affects of self-prophecy in a substantive domain (such as when a follow-up assessment of the technique is needed or when self-prophecy is demonstrated in a new domain as in Study 1). The difficulty lies in the measurement of overt behavior; consider, for example, the time and effort associated with Study 1. This study required ten undergraduate research assistants working approximately five hours per week, the coordination of five dormitories for an entire semester and approximately eighty hours of up-front time by the primary researchers. Of course, many experiments in natural settings have similar concerns; nonetheless, this study illustrates the issue of whether the application and, more importantly, the assessment of this technique is viable in a large scale effort like recycling. Study 2 addresses this issue by examining whether a self-report measure of recyclingCa less labor- and time-consuming dependent variableCwould provide evidence of a self-prophecy effect.

STUDY 2

All prior studies have employed some form of behavioral measurement to assess the experimental effects of self-prophecy. Such measures include health club attendance from business records (Spangenberg 1997), voting via official polling records (Greenwald et al. 1987), and cheating assessed by the grading of a take-home assignment (Spangenberg and Obermiller 1997). Study 2 questions whether such measures of behavior are necessary to find a self-prophecy effect: Would a self-report measure of behavior show a self-prophecy effect similar to that found with measures of actual behavior? If self-report measures work in a similar manner, then researchers would have a more tractable approach for assessing the effects of self-prediction. In addition to being useful in a natural setting, self-report measures would also provide a more efficient manner by which to conduct laboratory research. Because the theoretical mechanism underlying the effect has yet to be definitively determined, it is unclear whether one can predict manifestation of a self-prophecy effect using a self-report measure as the criterion variable.

FIGURE

STUDY 1: CANS RECYCLED BY TREATMENT

Method

Experimental Design, Sample, and Self-Prophecy Manipulation. One-hundred twent six undergraduate business students at Washington State University (56% female) were randomly assigned in equal proportions to prediction and no prediction conditions. The self-prophecy manipulation was similar to the one used in Study 1. In both conditions, subjects were asked to complete a one-page survey with five forced-choice decisions based on written scenarios. The scenarios in the no-prediction condition were the same as those used in Study 1, but also included a scenario asking subjects to make a prediction about a request for a refund after being overcharged at a busy coffee stand.

Importantly, the scenarios for the prediction condition in the second study differed from those of Study 1. Whereas Study 1 examined only recycling, Study 2 explored five behaviors, several of which have manifested a self-prophecy effect in prior studies using behavioral dependent measures: drinking and driving, recycling (this paper, Study 1), eating a healthy snack (Spangenberg et al. 1998), cheating in class (Spangenberg and Obermiller 1997), and exercising (Spangenberg 1997). For the prediction condition, the focal scenarioCidentical to the one presented in Study 1Cregarded recycling behavior and appeared in the second position. The two predictions (i.e., will or will not recycle) were again counterbalanced to preclude order effects.

Procedures and Dependent Measure. During the first week of Spring 1997 classes (the same semester as Study 1), a research assistant administered the self-prophecy survey to the members of two senior marketing classes. Throughout the semester (for six times), the same research assistant administered the dependent measures for focal behaviors. Dependent measures included two items (illustrated here for recycling behavior). Specifically, subjects were asked to "Please indicate how often in the past week you have participated in recycling behavior." This prompt was followed by two scales: (1) a fixed-choice alternative with "Not Once," "1 Time," and so on up to "6 Times," and "More than 6 Times" and (2) a 7-item scale anchored with "Not at All" and "Very Often." Scale items were summed across the six measurement periods, the average of the two items was then used as the primary dependent variable in the analysis (r=0.82).

Analysis and Results

Self-reported recycling behavior for subjects making a prediction (M=8.99, s.d.=7.50) was not significantly different from those making no prediction (M=9.47, s.d.=6.54; t₍₁₂₄₎=0.38, p_1-tail=0.35). Individual t-tests were conducted for each time period, none of these tests were significant (all t’s<0.74). It should be noted that Study 2 had sufficient power; a powerful 1-tail test (i.e., power=0.90) for a medium-sized effect (as assessed by d) requires a total sample of 85 subjects, a number we greatly exceeded. [There were no self-prophecy effects for eating a healthy snack or exercising (both t's<1). Analyses conducted on cheating and drinking and driving behaviors were called into question due to the restricted nature of the dependent variable (e.g., many subjects reported not performing such behaviors at any time during the data collection).]

Discussion

Results of Study 2 provide no evidence of a self-prophecy effect for any of the behaviors using self-reported measures. Recall that prior research has found self-prophecy effects for several of these when the behavior was unobtrusively measured.

More than one explanation exits for the null result of Study 2. There is the potential for a demand artifact; it is possible that the instrument itselfCadministered over several time periodsCelicited a socially normative response wherein subjects reported "good" behavior (whether honest or not). A socially normative response would likely be elicited for any self-report of these behaviors in order to present one’s self in the most positive light. It alternately may be that subjects deceived themselves at the time of self-report to avoid cognitive dissonance (Festinger 1957). Alternatively, student responses in Study 2 may accurately reflect behavior. Recurrent recycling self-reports may have operated as a socially sensitive piming mechanism that increased all subjects’ recycling (over others in the student population not involved with the study) with self-reports accurately capturing these behavioral changes. This latter explanation suggests that the measurement process itself elevated behavior across all conditions, thus creating a ceiling effect under which there was little or no room for a self-prophecy effect to manifest.

Inasmuch as the existing empirical evidence for self-prophecy effects is substantial, especially regarding the focal behaviors of Study 2, it would be hard to argue that the prediction manipulation had no effect upon the actual behaviors. Our doubt regarding the inefficacy of the manipulation leads us to believe then that there must be something about self-reporting behavior that "masks" the effect of the prediction manipulation. Future research could investigate this issue more fully with a design examining correspondence between self-report measures and performance of the behavior.

GENERAL DISCUSSION

Prior research has shown the self-prophecy effect to be robust and reasonably predictable for socially normative behaviors. The present research contributes to this ever-growing body of literature by establishing a new domain for the effect (i.e., recycling in Study 1) and an apparent limitation on how self-prophecy effects can be measured (Study 2).

Study 1’s findings are consistent with previous self-prophecy research. The results, as with others in this stream, are interesting when one considers that a simple prediction is able to influence overt behavior. The fact that this effect was found in a field setting, where uncontrollable events and noise are the rule, attests to the strength of the self-prophecy technique. While Study 1 provides additional support for the efficacy of self-prophecy, concern remains regarding the costs associated with determining the existence of a self-prophecy effect in the field. Unfortunately, the failure to find significant effects in Study 2 calls into question the use of self-report measures of behavior. If these null findings are in fact due to some effect of the self-report measure in the no prediction condition (such as dissonance reduction), it may suggest processes underlying the effect. A question raised by these results relates to timing of the effect: Does the mechanism of the self-prophecy effect manifest at prediction request or at the behavioral opportunity? If the results of Study 2 are not accounted for by any other methodological or theoretical explanation (which have not yet been ruled out) they suggest that the mechanism underlying the self-prophecy effect may occur at behavior. This is in contrast with new data suggesting that the self-prophecy effect occurs at the time of the predictionCnot at the time the behavioral opportunity occurs (Spangenberg et al. 1998).

From a substantive perspective, if the self-prophecy effect is as robust as it appears, programs attempting to influence socially conscious behavior should benefit from this technique when situations arise that are congruent with the boundary conditions of the effect. Groups could invoke the self-prophecy effect when gathering information, or soliciting involvement in a project that has socially relevant outcomes.

REFERENCES

Aronson, Elliot (1992), "The Return of the Repressed: Dissonance Theory Makes a Comeback.," Psychological Inquiry, 3 (4), 303-311.

Bandura, Albert (1969), Principles of Behavior Modification, New York: Holt, Rinehart & Winston.

Bryce, Wendy J., Rachel Day, and Thomas J. Olney (1997), "Commitment Approach to Motivating Community Recycling: New Zealand Curbside Trial," The Journal of Consumer Affairs, 31 (1), 27-52.

Cohen, Jacob (1988), Statistical Power Analysis For the Behavioral Sciences, 2d ed., Hillsdale, NJ: Erlbaum.

Cook, Stuart W. and Joy L. Berrenberg (1981), "Approaches to Encouraging Conservation Behavior: A Review and Conceptual Framework," Journal of Social Issues, 37 (2), 73-107.

Festinger, Leon (1957), A Theory of Cognitive Dissonance, Stanford, CA: Stanford University Press.

Greenwald, Anthony G., Catherine G. Carnot, Rebecca Beach, and Barbara Young (1987), "Increasing Voting Behavior by Asking People If They Expect To Vote," Journal of Applied Psychology, 72 (2), 315-318.

Greenwald, Anthony G., Mark R. Klinger, Mark E. Vande Kamp, and Katherine L. Kerr (1988), "Self-Prophecy Election Studies," unpublished manuscript.

Higbee, Kenneth L. (1969), "Fifteen Years of Fear Arousal: Research on Threat Appeals, 1953-1968," Psychological Bulletin, 72 (6), 426-444.

Katzev, Richard D. and Anton U. Pardini (1987-1988), "The Comparative Effectiveness of Reward and Commitment Approaches in Motivating Community Recycling," Journal of Environmental Systems, 17 (2), 93-113.

Maloney, Michael P. and Michael P. Ward (1973), "Ecology: Let’s Hear from the People; an Objective Scale for the Measurement of Ecological Attitudes and Knowledge," American Psychologist, 7, 583-586.

Obermiller, Carl, Eric R. Spangenberg, and April Atwood (1992), "Getting People to Give More: A Telephone Funds-Soliciting Strategy Based on the Self-Erasing Nature of Errors of Prediction," in Marketing Theory and Applications, Vol. 3, Chris T. Allen et al., eds. Chicago, IL: American Marketing Association, 339-345.

Rogers, Ronald W. (1975), "A Protection Motivation Theory of Fear Appeals and Attitude Change," Journal of Psychology, 91, 93-114.

Seaver, W. Burleigh and Arthur H. Patterson (1976), "Decreasing Fuel-Oil Consumption through Feedback and Social Commendation," Journal of Applied Behavior Analysis, 9, 147-152.

Sherman, Steven J. (1980), "On the Self-Erasing Nature of Errors of Prediction," Journal of Personality and Social Psychology, 39 (2), 211-221.

Spangenberg, Eric R. (1997), "Increasing Health Club Attendance Through Self-Prophecy," Marketing Letters, 8 (1), 23-32.

Spangenberg, Eric R. and Anthony G. Greenwald (1999), "Social Influence by Requesting Self-Prophecy," Journal of Consumer Psychology, 8 (1), 61-89.

Spangenberg, Eric R. and Carl Obermiller (1997), "To Cheat or Not To Cheat: Reducing Cheating by Requesting Self-Prophecy," Marketing Education Review, 6 (3), 95-103.

Spangenberg, Eric R., David E. Sprott, Robert J. Fisher, and Anthony G. Greenwald (1998), "The Self-Prophecy Effect: Theoretical Explanations and Limiting Conditions," working paper.

Stone, Jeff, Elliot Aronson, A. Lauren Crain, Matthew P. Winslow, and Carrie B. Fried (1994), "Inducing Hypocrisy as a Means of Encouraging Young Adults To Use Condoms," Personality and Social Psychology Bulletin, 20 (1), 116-128.

Wang, Theodore H. and Richard D. Katzev (1990), "Group Commitment and Resource Conservation: Two Field Experiments on Promoting Recycling," Journal of Applied Social Psychology, 20 (4), 265-275.

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