Can Role Playing Be Substituted For Actual Consumption?

ABSTRACT - A review of the issues involved in using role playing methodologies is provided, together with some of the requirements for appropriate usage. The results of an experiment using both role playing and actual consumption are presented to assess, empirically, the impact of role playing on the data collected.

Citation:

Carol Surprenant and Gilbert A. Churchill, Jr. (1984) ,"Can Role Playing Be Substituted For Actual Consumption?", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 122-126.

Advances in Consumer Research Volume 11, 1984      Pages 122-126

CAN ROLE PLAYING BE SUBSTITUTED FOR ACTUAL CONSUMPTION?

Carol Surprenant, New York University [Assistant Professor of Marketing]

Gilbert A. Churchill, Jr., University of Wisconsin-Madison [Donald C. Slichter Professor in Business Research]

ABSTRACT -

A review of the issues involved in using role playing methodologies is provided, together with some of the requirements for appropriate usage. The results of an experiment using both role playing and actual consumption are presented to assess, empirically, the impact of role playing on the data collected.

INTRODUCTION

Role playing or role enactment is a research technique in which the researcher asks a subject to behave as if he or ,he were in some situation. The subject may be asked to project his/her own behavior or the behavior of an "other." The key characteristic of the role playing methodology is that the situation or combination of conditions does not, in fact, exist.

The effect of using role playing as a methodology has received little attention in the marketing literature. Although the method is used by marketing researchers, little attention has been given to the appropriate use of the methodology or to the potential effects of its use on data quality and results, with the exception of Sawyer (1977). By contrast, the psychology literature contains numerous reviews and empirical studies (cf Kelman 1967; Aronson and Carlsmith 1968; Miller 1972; and Geller 1978) which assess the limitations of the methodology.

There are several potential explanations for this apparent neglect in the marketing literature. First, as members of in applied discipline, many marketing researchers may feel :he use of role playing is indefensible for any reason. rhus, many would argue that field studies are the only appropriate research technique leaving laboratory and simuLation studies to psychologists. Second, given the existence of a psychological literature on the subject, additional research by marketing researchers may be unnecessary. finally, because role playing is widely viewed as producing -relatively low levels of subject involvement, researchers who employ the technique may prefer to downplay the issue, fearing to call the results of their research into question.

This predominantly negative view has had major effects. Role playing is often used without consideration of the potential impact or limitations of the methodology. Researchers may avoid the technique, substituting more expensive and/or less internally valid methods, not because :he techniques are more appropriate but because they are more accepted. Moreover, since marketing researchers have neglected this area, we often have little information on low the use of this method influences results.

Uses of Role Playing

A major use of role playing in psychological research is as an alternative to deception (Cooper 1976; Forward, Canter, and Kirsch 1976; Greenberg 1967; Hamilton 1976; Holmes and Bennett 1974; Miller 1972; among others). While concern about deception in marketing studies has been voiced (Tybout and Zaltman 1974), this has not been a critical issue. In most marketing studies the level of deception is low and seldom involves significant physical or psychological harm to subjects. Brand names may be falsified, sponsorship disguised or cover stories to hide experimental hypotheses may be invented. However, to the extent that deception is incomplete and/or generates subject reactivity, its use may provide biased data (Sawyer 1977). Data from role playing, on the other hand, may be less biased and ethically superior.

Role playing has more often been used in experimental settings to test hypothesis regarding constructs difficult to measure in "real" settings. Jacob , Speller and Berning (1974), for example, asked subjects to behave as if they were actually shopping for frequently purchased goods, in a study of information used in making brand choices. Calder and Burnkrant (1977) asked subjects to take the role of observers in a study of attribution processes. Similarly, Hansen (1972) used role playing to investigate consumer choice processes. The justification in Hansen's study was to be able to manipulate multiple variables which otherwise would have been difficult or impossible to examine.

Role playing methodologies are also used when testing hypotheses about relatively inaccessible groups. Sawyer cites Rosenberg and Stern's (1971) study in which students played the role of salesmen and purchasing agents in a study of bargaining. Similarly Whitney and Smith (1983) used students to simulate work groups in a study of organizational decision making.

A final use of role playing in marketing research is to examine consumer's behavior toward expensive or infrequently purchased goods. The great majority of marketing studies have focused on "non-durables" or frequently purchased items which tend to be inexpensive. High product costs, long consumption times and the difficulty of identifying consumers actually in the process of purchasing such products have Led researchers to avoid durable purchases. The tendency has been to generalize what is known from research on nondurable products to all products. While several have noted the likelihood or substantially different processes between.durable and non-durable purchases (e.g., Day 1977), the expense and difficulty of doing traditional research with these products has constrained investigation.

Pros and Cons or Role Playing

Sawyer (1977) has provided a comprehensive review of both the advantages and disadvantages of role playing, thus only a summary will be offered here. A primary advantage of role playing is its ethical superiority over deception. Additionally, as an alternative to deception, role playing may provide less biased data, while as an alternative to more traditional methods role playing is flexible. Manipulations which otherwise would be expensive or extremely difficult (such as timing of choice) are relatively easy to accomplish using role playing. Moreover, role playing permits the researcher to control variables that are possible causes of behavior.

On the negative side, role playing may be more subject to demand characteristics. Subjects may adopt subject roles (e.g. the "good" subject) which will affect the dependent variables. Subjects may be unable (or unwilling) to predict their behavior in the situation of interest. Finally, role playing may not produce the subject involvement generated to more traditional techniques.

While it is clear from the brief discussion above that role playing is not without limitations, the same can be said o any methodology currently in use. Both laboratory and field experiments have strengths and weaknesses as do survey methods. The tact that a methodology has limitations should not argue against its use; nor should researchers gloss over the limitations of a methodology in discussing results. Despite its disadvantages, role playing will continue to be used by marketing researchers. It is, therefore, necessary to compare role playing and naive subject results to asses the comparability of the data generated.

STUDY

The data for this analysis were part of a larger experiment reported elsewhere (Churchill and Surprenant, 1982). The primary purpose or the experiment was to examine the relationship of satisfaction to expectation, product performance, and disconfirmation. Expectation and performance levels were manipulated in a 3 x 3 factorial design, while disconfirmation was measured after exposure. Two products were tested in the full experiment; an expensive, durable product, (video disc player) and an inexpensive, non-durable product, (a plant). Two products were used to see if the relationship of satisfaction to the independent variables differed by product class. Previous experimental studies had used non-durable products (e.g., Anderson 1973; Cardozo, 1965). A major consideration in the choice of the specific products used was the ability to manipulate performance.

The experiment described below used both role playing and traditional experimental procedures. Since satisfaction i a post-purchase/consumption variable, it should be measure after use (although how long after remains open to question For the durable product this would have been expensive and time consuming without using a role playing methodology. Thus, subjects were asked to respond as if they had purchased the product in question. In order to assess the potential impact or this role playing methodology on the data collected, for the plant condition the sample was split. Half the subjects responded to a role playing scenario, while for the remainder of the subjects, dependent measures were collected after actual consumption of the product.

Subjects

One hundred eighty subjects between the ages of 19 and 65 were recruited at a shopping mall and offered three dollar to participate in new product testing. Ninety subjects were assigned to the role condition, ninety to the actual consumption condition. Because of the four week lag for t actual consumption condition, five subjects could not be reached to complete the experiment. Thus complete data wa available for 175 subjects.

Procedure

Subjects were randomly assigned to treatment conditions. The first part of the experimental situation involved a video disc player. Subjects were exposed to the expectation manipulation then given a manipulation check. Following this they were given a five minute demonstration or the product and its capabilities. At the end of the demonstration, subjects were given the opportunity to examine the product and ask questions, then returned to the first room where they were provided with a scenario to facilitate role playing. They were asked to consider that they had purchased the product, have it in their home, and it has been operating for one month, just as the one they have seen. Post-exposure expectation measures and satisfaction questionnaires were then administered. Postcards requesting more information on the product were included in the packet or post-exposure measures. Return of these postcards constituted an indirect behavioral measure of commitment to the product.

Following this treatment, one-half of the subjects were given a similar task with a second new product, a plant. Information about the product was again given to establish initial expectations (Appendix A). Subjects were shown an example of the product, along with a comparison product, a similar plant four weeks old. Post exposure measures were again taken using a role playing scenario.

The remaining half of the subjects were exposed to a somewhat different situation with the plant. They were told that as part of their compensation for participating in the experiment, [All subjects were given a plant as part of their compensation for participating in the experiment. However, only those in the actual consumption condition were interviewed at the end of four weeks.] they would receive an experimental new product. Subjects in this condition were also given information about the product and shown the comparison product. They were then dismissed after obtaining names and addresses so that the rest of their payment could be mailed to them. Four weeks after this portion of the experiment was concluded, [The average blooming time for the plant chosen, a chrysanthemum, is four to six weeks. At the end of this time, most people dispose of the plant although some transplant it in a garden for the following year.] questionnaires were delivered to those subjects who received the plants. Following completion of the questionnaires, subjects were given the opportunity to exchange their plant for another. This constituted a behavioral measure of satisfaction with the product.

Measures

For both products, multiple measures were taken. However, since data for both role playing and non-role playing methodologies are only available for the non-durable product, the plant, only that data will be discussed. Post-exposure measures for both groups included beliefs, affect, disconfirmation, satisfaction, and purchase intention (Table 1). All were seven point scales with the exception of performance, a line scale. Internal reliability for all scales was high, .80 and above.

RESULTS

Analysis of variance results given in Table 2 using expectation, performance, and methodology as independent variables and satisfaction as the dependent variable showed there was a significant main effect for methodology. In addition to the main effect, method interacted strongly with performance. Table 3 presents the cell means for each of the conditions. It is clear from Table 3 that the effect of role playing was complex.

Within expectation conditions, the direction of the results was the same under both conditions. Role playing primarily served to compress the results. That is, in general role playing subjects were more satisfied with low performance and less satisfied with high performance than were non-role playing subjects.

The major effect of role playing, however, is seen by looking across expectation conditions in the low performance cells. As is clear from Figure 1, for the non-role playing subjects, when expectations exceeded performance, satisfaction decreased. However, for the role playing subjects, increasing expectations increased satisfaction even though performance was low. While only the difference between the low expectation/low performance and high expectation/low performance cells in the non-role condition is significant, a comparison of pane A and panel B in Figure 1 clearly demonstrates this trend.

TABLE 1

SCALES USED

TABLE 2

MEAN SATISFACTION WITH PLANT OUTCOMES AS A FUNCTION OF EXPECTATION, PERFORMANCE, AND EXPERIMENTAL METHODOLOGY

TABLE 3

CALL MEANS a: PLANT SATISFACTION AS A FUNCTION OF EXPECTATION, PERFORMANCE, AND METHODOLOGY

When belief and affect measures were used as dependent variables in an analysis of variance, there were no significant effects. Neither main effects nor interaction effects were significant as can be seen in Table 4. Role playing subjects produced results equivalent to those obtained from subjects who actually consumed the product. An examination of the cell means (Table 5), however, indicates that for the affective measure, a pattern of responses exist similar to that observed when satisfaction was the dependent variable. Subjects who actually consumed the product showed a slight decline in affect scores when expectations increased in the low performance condition. Role playing subjects did not. Moreover, role playing subjects results were, again, compressed. While the affect results are similar to those obtained using satisfaction as the dependent variable, in this case they are not strong enough to achieve significance. Belief scores for the two groups are very similar with a minor compression of scores evident. Again, however, the effect of methodology is non-significant.

DISCUSSION

In the study reported here it is clear that role playing subjects did not produce exactly the same results as subjects who actually consumed the product. Main effects for the two groups were similar, however there were subtle differences. Particularly when the product performed poorly, e.g. was smaller than anticipated or had fewer flowers, the role playing subjects were unable to adequately simulate what their reactions would be.

FIGURE 1

MEAN SATISFACTINO BY PLANT EXPECTATION

Willis and Willis (1970) found that role playing subjects in a conformity study could simulate main effects but not subtle interactions. In a study which is perhaps more relevant here, Holmes and Bennett (1974) found that role playing subjects couLd simulate self report data but not physiological responses, i.e. processes not directly linked to cognitive control. Similarly, in this study. differences between the two groups were far more evident on variables of an affective nature (satisfaction and affect). These were not simulated as well as more cognitive (belief) variables. To some extent, inability to simulate affective responses may be idiosyncratic to the product used. Subjects may develop more affective responses to plants only after they have owned and cared for them for a time. In order to determine the generalizability of these results, additional research with other types of products is needed.

In his review of role playing as a methodological tool, Sawyer (1977) has suggested five conditions when role playing is likely to be less biased. When subjects are forecasting their own behavior, when there is no embarrassment circumstances are familiar, the research situation is simple, and hypotheses limited to main effects, role playing is likely to provide unbiased results. Those conditions would seem to obtain in this experiment unless, of course, the experience of having a product which performs poorly was unfamiliar to the subjects. If, in fact, the conditions stated above have all been met, it would appear that role playing data does seem to do an adequate job of simulating major results. For more subtle results, however, and where intensity of feeling is of concern, the methodology seems inadequate based on the results reported here.

There is, however, one additional factor which must be considered before concluding that role playing is inadequate. Geller (1978) has argued that role playing may not produce good results when the experimental situation is not powerful or involving enough. Zimbardo's famous prison study (1973) is one well known e:;ample of a role playing methodology which did produce very high levels of involvement on the part of the subjects. Such high levels of involvement however, quite obviously did not exist in the experiment r ported above. The product would generally be described as a low involvement one. Subjects may have volunteered to participate primarily because of the durable product, the video disc player, and failed to give their attention to the other product. If involvement with the product was low, it is not surprising that role playing subjects would not reproduce the more emotional responses measured.

A final point to be considered relates not so much to the use of role playing as a methodology as to the issue of when to measure satisfaction. Satisfaction is a post-purchase variable. When consumption occurs over a period of time, the issue of when to measure satisfaction becomes critical (Day 1977). The useful life of the product used in this study was approximately four weeks. Thus, subjects who actually used the product were measured at the end of that period. Role playing subjects were measured immediately after the simulated "purchase" of the product, and asked t predict their future level of satisfaction in the presence of another four week old plan;. It is entirely possible that the underlying constructs measured, especially affect and satisfaction, might change over this period. Unfortunately, this issue, while interesting, cannot be answered here

CONCLUSIONS

Role playing's use by marketing researchers is unlikely to disappear nor, perhaps, should it. Some of the uses of the methodology discussed elsewhere (Sawyer 1977; Hansen 1972) are worthwhile and useful. It day, in fact, play an important part in an overall research system under certain conditions.

The study reported here is one attempt to assess the comparability of data generated using role and non-role methodologies. Based on these results, role playing seems to do a reasonable job of producing comparable results, at least in so far as main effects are concerned. It is also clear that affective responses are not as well simulated as cognitive responses. This, of course, is only one study and. therefore, generalizability is limited.

The indications here are that role playing may be useful in certain well defined research situations. In other words, the answer to the question posed in the title of this paper must remain . . . it depends. If marketing researchers are to continue to use the methodology it is necessary to clearly determine the optimal conditions for its use. It must be subjected to the same type of methodological scrutiny given other research methods.

APPENDIX A

HIGH EXPECTATION MANIPULATION

Product Description: Hybrid 261

The morifolium chrysanthemum is a new hybrid recently developed for the home market. It is larger than other mums averaging from 8 to 9 inches. The average size of most mum varieties is 6 to 7 inches. The flowers exhibit richer, deeper colors than those of ordinary mums. In combination with the large number of flowering stems, 6 or mores the full foliage produces a plant that has a luxuriant appearance, full and hardy.

A further advantage of this plant is its exceptionally long blooming time. Unlike most potted chrysanthemums, which bloom, on the average for 2 to 3 weeks, this hybrid remains in bloom for a full 4 to 5 weeks. It can tolerate a wide range of temperatures and performs well under most ordinary household conditions. Best results are obtained when the plant is placed in bright indirect or curtain filtered light and is kept uniformly moist.

In comparison with other flowering houseplants, this chrysanthemum has been rated EXCELLENT by both commercial growers and horticulturalists. Growers believe that this variety will become the most popular mum on the market, soon almost replacing other varieties. Laboratory tests confirm that it is a superior hybrid with desirable characteristics in size, flower production and lifespan. Test results for the plant are given below with comparisons to average results on other mum varieties. Qualitative statements represent the general opinion of horticulturalists.

TABLE

Our research indicates that this product, in general, is EXCELLENT. It outperforms other hybrids on almost all characteristics. Consumer testing is now being conducted to obtain buyers' reactions to this hybrid.

REFERENCES

Anderson, Rolph E., "Consumer Dissatisfaction: The Effect of Disconfirmed Expectancy on Perceived Product Performance," Journal of Marketing Research 10 (February 1973), 30-44.

Aronson, Elliot and Carlsmith, J.M. (1968), "Experimentation in Social Psychology" in The Handbook of Social Psychology Vol. II, G. Lindzey and E. Aronson, eds., Reading, Massachusetts: Addison-Wesley.

Calder, Bobby 3. and Burnkrant, Robert E. (1977), "Interpersonal Influence on Consumer Behavior: An Attribution Theory Approach," Journal of Consumer Research, 4 (June), 29-38.

Cardozo, Richard N., "An Experimental Study of Consumer Effort, Expectation, and Satisfaction," Journal of Marketing Research, 2 (August, 1965), pp. 244-249.

Churchill, Gilbert A. Jr. and Surprenant, Carol F. (1¦82), "An Investigation into the Determinants of Customer Satisfaction," Journal of Marketing Research, 19 (November), 491-504.

Cooper, Joel (1976). "Deception and Role Playing: On Telling the Good Guys from the Bad Guys," American Psychologist, (August), 605-10.

Day, Ralph (1977), "Toward a Process Model of Consumer Satisfaction," in H. Keith Hunt, ed. Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction, Marketing Science Institute. 153-186.

Forward, John, Canter, Rachelle, and Kirsch, Ned (1976), "Role-Enactment and Deception Methodologies: Alternative Paradigms?," American Psychologist, (August), 595-604.

Geller, Daniel M. (1978), "Involvement in Role-Playing Simulations: A Demonstration With Studies on Obedience," Journal of Personality and Psychology, 36 (March), 219-35.

Greenberg, Martin S. (1967), "Role Playing: An Alternative to Deception?," Journal of Personality and Social Psychology, 7, 152-7.

Hamilton, V. Lee (1976), "Role Play and Deception: A ReExamination of the Controversy," Journal for the Theory of Social Behavior, 6, 231-50.

Hansen, Flemming (1972), Consumer Choice Behavior: A Cognitive Theory, New York: The Free Press.

Holmes, David S. and Bennett, David H. (1974), "Experiments to Answer Questions Raised by the Use of Deception in Psychological Research: I. Role Playing as an Alternative to Deception: II. Effectiveness of Debriefing After a Deception: III. Effect of Informed Consent on Deception," Journal of Personality and Social Psychology, 29 (March), 358

Jacoby, Jacob, Speller, Donald E., Berning, Carol Kohn (1974), "Brand Choice Behavior as a Function of Information Load: Replication and Extension," Journal of Consumer Research, I (June), 33-42.

Kelman, Herbert C. (1967), "Human Use of Human Subjects: The Problem of Deception in Social Psychological Experiments," Psychological Bulletin, 67 (January), 1-11.

Miller, Arthur G. (1972), "Role Playing: An Alternative to Deception? A Review of the Evidence," American Psychologist, 27 (July), 623-34.

Rosenberg, L.J. and Stern, L.W., "Conflict Management in the Distribution Channel," Journal of Marketing Research 8 (November 1971). 437-42.

Sawyer, Alan G. (1977), "The Role of Role Playing in Experiments About Consumer Behavior," in Contemporary Marketing Thought, Barnett A. Greenberg and Danny N. Bellenger, eds., American Marketing Association, 191-194.

Tybout, A.M. and Zaltman, G., "Ethics in Marketing Research: Their Practical Relevance," Journal of Marketing Research 11 (November 1974), 357-68.

Whitney, John C. and Smith, Ruth A. (1983), "Effects of Group Cohesiveness on Attitude Polarization and the Acquisition of Knowledge in a Strategic Planning Context," Journal of Marketing Research 20 (May), 167-176.

Willis, R.H. and Willis, Y.A., "Role Playing vs. Deception: An Experimental Comparison," Journal of Personality and Social Psychology, 16 (November 1970), 472-77.

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