Can Seller/Customer Interaction and Influence Be Studied in the Laboratory?

ABSTRACT - All methodologies used to study seller/customer interactions suffer from various important deficiencies. Although the recent literature in this area has included many new models conceptualizing the seller/customer interaction and influence processes, methodological development has failed to keep pace with these new theoretical models. This paper describes a laboratory procedure which combines the usual lab advantages of low cost and high control with reasonably good external validity and also allows measurement of a series of variables posited in some of the recent models. An exploratory experiment which varied the expertise of the salesperson is described.


Alan G. Sawyer, Terry Deutscher, and Carl Obermiller (1980) ,"Can Seller/Customer Interaction and Influence Be Studied in the Laboratory?", in NA - Advances in Consumer Research Volume 07, eds. Jerry C. Olson, Ann Abor, MI : Association for Consumer Research, Pages: 393-399.

Advances in Consumer Research Volume 7, 1980     Pages 393-399


Alan G. Sawyer, The Ohio State University

Terry Deutscher, University of Western Ontario

Carl Obermiller, The Ohio State University

[The authors want to acknowledge the advice of Clark Leavitt, The Ohio State University and the financial assistance of a Faculty Growth Grant, College of Administrative Science, The Ohio State University and the Fund for Excellence, School of Business Administration, University of Western Ontario.]


All methodologies used to study seller/customer interactions suffer from various important deficiencies. Although the recent literature in this area has included many new models conceptualizing the seller/customer interaction and influence processes, methodological development has failed to keep pace with these new theoretical models. This paper describes a laboratory procedure which combines the usual lab advantages of low cost and high control with reasonably good external validity and also allows measurement of a series of variables posited in some of the recent models. An exploratory experiment which varied the expertise of the salesperson is described.


Consumer researchers have demonstrated considerable interest in studying the communication and influence processes between a seller and a customer. This interest was very much in evidence at the 1975 Conference of the Association for Consumer Research in a session organized by Capon and Hulbert (1976). A search of the many reviews of the literature in this area (Capon, Holbrook, and Hulbert, 1977; Davis and Silk, 1972; Deutscher and Sawyer, 1978a; 1978b; Holbrook and O'Shaughnessy, 1976; Hulbert and Capon, 1972, Mathews, Robeson, and Bambic, 1977), yields several commonly agreed upon conclusions.

First, studies of the selling process should consider both parts of the sales dyad - the seller and the customer. Second, the responses of the participants in this dyad must be measured along a wide continuum ranging from cognitive to conative. Measures of much more than purchase are needed to begin to understand the influence process. Third, since personal selling is especially important in the marketing of higher-priced products or services, such items need to be studied and not merely low priced items. Ideally, customer-seller relationships longer than "one-shot" selling situations ought also to be studied. Fourth, theory ought to be used to generate the relevant variables of the seller, customer, sales message and buying situations that are manipulated and studied. Fifth, four primary approaches have been used to study seller/customer interactions or influence: correlational studies of buyer-seller characteristics and resulting sales (e.g., Evans, 1963; Gadel, 1964), field studies of interactions between sellers and customer (e.g., Olshavsky, 1973; Woodside and Sims, 1976), field experiments manipulating seller and sales message variables (e.g., Capon, 1975; Sternthal, Scott, and Dholakia, 1975), and laboratory experiments (e.g., Farley and Swinth, 1967; Reizenstein, 1971).

This paper describes an attempt to design a research procedure that has many of the advantages of each of the four types of past research and eliminates some of the respective disadvantages. This paradigm also permits most of the types of research topics mentioned in the conclusions above. After a description of the research methodology and its various advantages, an exploratory study manipulating salesperson expertise will be described.


The method described below was designed to meet several important objectives. First, we wanted to be able to experimentally manipulate seller and sales messages with the according advantages of internal validity.

Second, we wanted a laboratory-like setting which would offer the traditional advantages of enhanced control and lowered cost and time (cf. Ray, 1978). However, we also wanted to create a procedure that was higher than most past lab experiments in terms of what Carlsmith, Ellsworth and Aronson (1976) call experimental reality. At the very least, we felt it was vital to have a "live" salesperson and not a filmed presentation. Some past research (e.g., Busch and Wilson, 1976; Levitt, 1965) used filmed presentations presumably because of the advantage of control (i.e., the exact same message presented in exactly the same manner to all subjects in an experimental condition) and also the operational convenience and reduced cost of not needing a trained confederate in addition to the experimenter for every subject. However, the use of a filmed salespitch has major drawbacks. It results in lowered external validity. And it prevents the study of interaction within the sales dyad.

Of course, gaining the greater impact and realism of a live salesperson instead of a filmed presentation implies a potential loss of control. A live presentation that permits or even encourages the customer to ask questions and express opinions will not be the same standard stimulus to each subject in a given experimental condition. This is acceptable as long as the lack of standardization leads to randomized error and not systematic error or bias across different treatments. Furthermore, we would argue, similar to the somewhat controversial position of Carlsmith, Ellsworth and Aronson (1976) about allowing flexible experimental instructions and treatments, that the proper goal of a controlled treatment is not to say the same thing to every subject but to bring each subject to the same level of effect. Similarly, most principles of personal selling do not seek to standardize the presentation; in fact, the opposite is true. Instead, the goal is to standardize the effect and maximize the chances of a successful sell. Standardization of the effect and not the treatment is the goal.

Third, it is important that the customer-subject perceive that he or she is confronted with an actual purchase influence attempt and must make some sort of purchase decision such as buy now, wait until later to decide, get more information, or refuse to buy. Requiring subjects to role-play or assume themselves to be in the market (Levitt, 1965; Busch and Wilson, 1976; Reizenstein, 1971) is too low in experimental realism to fully engage subjects' sincere consideration.

Fourth, we wanted a laboratory situation that, after the conclusion of an interaction session which included the customer's behavioral response(s) to at least one trial close, would permit the administration of a series of measures to permit the exploration or even the testing of one or more of the recent models of the interpersonal communication process. This might include a content analysis of verbal and perhaps even non-verbal (cf. Hulbert and Capon, 1972) interactions. Some clever researchers have combined the mundane reality of a field setting with reasonably good control and high treatment impact (Brock, 1965; Varela, 1971). However, these studies have had to confine their measurement to the behavioral criterion of whether or not a purchase was made. In addition to the fact that a behavioral criterion is not always a valid measure of success, especially for higher priced items where developmental selling is nearly as important as closing the sale (cf. Wilson, 1975), the behavioral criterion is insufficient to explore and test many important behavioral science theory hypotheses.

Two final objectives were also sought and quite easily met. It was desired to have a one-on-one selling situation and not a group of customers (see Jolson, 1975, 1976; Reed, 1976). Finally, because of the likelihood that idiosyncratic characteristics of a seller --especially a "live" one --might dominate any manipulated characteristic or message variable, it was deemed important to have several different salespersons. Capon (1975), one of the few researchers who systematically varied the actor-confederate in the salesperson role, found a significant salesman/treatment interaction but no main effects; the use of only one salesperson would lead to the erroneous conclusion that one of the six sales message treatments was superior in all cases. "Replicating over irrelevancies" in the Brunswick (1956) tradition will greatly add to the external validity of any results and, at the expense of statistical conclusion validity due to the greater "noise" created by the different salespersons and their unique enactments of assigned treatments, will add confidence in the construct validity of the cause of any observed effects (see Cook and Campbell, 1976).


Our goal in this research was to apply laboratory experiment methodology to the investigation of the buyer-seller dyad in order to expand our understanding of that relationship according to the objectives mentioned above. Since the methodology was our primary concern, we were content to replicate what we believed to be a well-established relationship, that of expertise and effectiveness. Because of previous research findings, particularly Brock (1965), Woodside and Davenport (1974), and Busch and Wilson (1976), but dating back to Hovland, Janis and Kelley (1953), we felt that, if the manipulation of expertise could be made, we would get a difference in salesperson effectiveness. Our concerns were chiefly with our ability to make the manipulation with a live salesperson and an interactive sales pitch, creating realism, measuring the intervening variables of attitude change and cognitive responses, and applying a content analysis to the interaction.


Subjects were Ohio State University college students enrolled in the introductory marketing management course. Course credit was given and participation was completely voluntary. When asked to volunteer, subjects were told that the research involved "interpersonal communication in non-profit marketing organizations." Roughly equal amounts of males (52.3%) and females (47.7%) participated.

Upon arriving at the experiment, a subject was randomly assigned to one of four experimental conditions: male high-expertise, female high-expertise, male low-expertise, and female low-expertise salesperson. The subject was admitted to a room with three chairs and a table with a tape recorder. An experimenter introduced herself (or himself) and told the subject he or she would be introduced to an administrator of the Continuing Development Program, that the administrator was sincerely interested in "selling" the student on a new course, and that we were studying the interaction because of our interest in interpersonal communications. After explaining that the sales discussion was going to be tape recorded so that the conversation could be replayed later, the experimenter called in and introduced the confederate salesperson, turned on the tape recorder, and left the room.

The confederate salesperson proceeded to informally present a salespitch designed to get the customer subject to sign up for a tennis course sponsored by a fictitious non-credit course program which was nearly identical to an actual popular program sponsored by the university. The salespitch involved a description of the program, questions about the subject's past experience and knowledge about non-credit courses, and a description of the demand for tennis instruction by students dissatisfied with the large classes offered by the university. The pitch went on to tell how an experienced, ex-varsity player had been secured as an instructor, the advantages of knowing about tennis, questions about the subject's tennis experience and preferred modes of expertise, and an outline of the course. The course was described as open to any "member of the OSU community," costing $30 for 18 classes, with classes held for one hour at a campus facility at 7:00 a.m. twice a week or at a more convenient time if it could be arranged. The class was to include 16 students grouped by ability.

After answering any questions, the salesperson asked 1) if the subject was interested in the course, 2) if they wished to reserve a place in the course by paying $5.00 to be applied to the $30.00 course fee (to be returned if too few students signed up), 3) if the subject wished to sign up for a $1.00 introductory lesson to meet the instructor and other potential students, and 4) if they would give names of other students who might (also) be interested in the course. If a subject wanted to pay the $5.00 sign-up fee but did not have the funds at the time, he was told that he could sign up as long as he delivered his $5.00 fee to a campus address within two weeks.

The confederate salespersons had memorized a sales talk of about 1500 words but were encouraged to use that in an informal manner, departing in wording as they saw fit. At several specific points in the pitch, they were told to solicit questions. (Subsequent monitoring indicated that the salespersons carried out their assignments as prescribed.) [A copy of the salespitch is available from the authors.] The result of the salespitch was an interactive situation in which the customer subjects reacted to questions of the salesperson and asked questions of their own. The salesperson talked about eighty percent of the time. The length of these talks ranged from fifteen to twenty-five minutes.

At the conclusion of the sales session, the salesperson thanked the subject for his time and left the room. The experimenter returned to the room and asked the subject to fill out a questionnaire. The questionnaire asked if the subjects thought that the sales presentation was clear and the speaker was effective and asked about the subject's experience with tennis and tennis lessons. The subject then rated the tennis course on six-point bipolar scales as good, well-planned, good value for the money, convenient to me, expensive, covers all aspects of tennis, adapted to individual needs, and good instructor. The speaker (salesperson) was similarly rated in terms of persuasive, likeable, interesting, trustworthy, like me, expert, attractive, aggressive, knowledgeable, and dynamic. "Tennis, in general" was also rated on points covered in the salespitch (expensive, easy to learn, exciting, good social activity, can be played all through life, good exercise, and enjoyable). Finally, the subject rated how likely he was to take the course, (after indicating his preferred time) the likelihood at his preferred time, the likelihood if the price were lowered to $10.00, and his estimate of the likelihood for the "average OSU student."

When the subject had completed the questionnaire, the experimenter described the process of gathering cognitive responses.

Subjects listened to the taped interaction and listed their thoughts. After conclusion of the taped replay, the subject answered questions about any guesses they might have had about the experimenter's hypotheses and what the subject thought the research was about.

The experiment lasted about seven weeks. At the conclusion of the experiment, subjects were debriefed by a typed description of the experiment and the results, and questions were answered in the introductory marketing class. All money received was returned.

Independent Variables and Design

The prime independent variable was the expertise of the salesperson. Expertise was randomly assigned to a subject after the experimenter left the room. About one-third through the presentation, the salesperson began to tell about the course content. At this point, the salesperson manipulated expertise in the following manner:

(High) I can give you a very good idea of what the course will be about since I worked with the instructor in the course. I played varsity tennis as an undergraduate at Ohio University and I've taught tennis for several years.

(Low) I'll try to give you some idea about the course. I don't know much about tennis myself; I've only played a few times. However the instructor gave me an outline of the course.

Expertise was further augmented when the salesperson went over a copy of the course outline. To four of the nine topics, the expert salesperson added a few words of "tennis lingo" to the topic heading. For "discussion of basics," the words "grip, stance, ready position, basic strokes" were added. Similarly, "forehand, backhand, volley, serve, overhead" were added to "basic strokes;" "serve and volley, defensive baseline, aggressive baseline" were added to "singles strategies;" and "court positions, poaching, lobbing" were added to "doubles and mixed doubles strategies." There were no other differences in the two expertise conditions. Confederate salespersons were cautioned to act the same in every other way; random monitoring of the sales pitches by means of watching through one-way mirrors and listening to tapes indicated that the confederates were successful. Although there were differences in the number of sales presentations made by each of the seven confederates (three men and four women), each confederate presented a nearly equal number in each expertise condition. The confederate salespersons, all MBA or Ph.D. students, were "trained" by a tennis teaching pro. They were familiarized with typical tennis courses and answers to possible questions. Pretests indicated the believability of all confederates in the high expert condition.

The experiment was a two (expert) by two (male or female subject) by seven (salespersons) between-subjects design. Some analyses also examined the effect of salesperson sex and covaried on subject's interest in various types of sports and exercise which had been measured in a separate pretest that was presented to the subjects as an in-class questionnaire exercise the first week of class. [This pretest questionnaire did not refer to the subsequent experiment.]


A prime objective was to measure the subjects' reactions to the experimental setting and the sales pitches for the tennis course. Subjects' reactions were assessed by informal observation by the salespersons and by the authors (through one-way windows) and by content analysis of the cognitive responses. The former assessment, which may not have been objective, indicated that nearly all subjects appeared involved and attentive to the sales pitch. In addition, nearly thirty percent of the subjects signed up for the tennis course, fifteen (ten percent) paid their five dollar deposit by the stated deadline, and fifty-nine percent signed up for a one dollar introductory lesson. Some of the subjects may have signed up to get away from the pressure of the salespitch, just as in "real" selling situations, but there was no evidence that subjects considered themselves to be in other than a real selling situation. Moreover, the final questions of the lab session, which asked for subjects' reactions and hypothesis-guessing, indicated over eighty-five percent of the subjects found nothing unusual or artificial in the situation. Six percent felt uneasy being taped, nine percent felt it was a peculiar way to sell a tennis course.

Cognitive Responses

As the tape of the sales interaction was replayed, subjects wrote down whatever thoughts they recalled occurring at various times during the sales pitch. The cognitive responses were coded as Cacioppo and Petty (1979) have advocated. Each response was coded according to its valence, origin, and target. Valence was the degree to which a statement was in favor of or opposed to what was advocated by the salesperson and was coded as either positive, neutral, negative, or curiosity. Curiosity was further coded into favorable, neutral, or negative curiosity. Origin referred to the primary source of the thought--either message, message-modified, or the recipient (subject). Target referred to what the thought was directed to and was coded as either the message or topic of the message or the source of the communicator or his communication style. In addition to the three factor codes suggested by Cacioppo and Petty (1979), the content of the cognitive response was coded to reflect what particular aspect of the target was the topic of the response.

Two judges independently rated each of the responses; disagreements were resolved by a third judge. Coder reliability was over ninety percent; most of the few disagreements involved the origin. The results showed that the mean number of expressed thoughts was 16.01 with a standard deviation of 8.7.

Manipulation Checks

As a precursor to questions about whether salesperson expertise was effective, the question of whether expertise was successfully manipulated must be answered. Several manipulation checks on perceived expertise were included in the post-test measures. Subjects rated the salespersons on a six-point scale between "Expert" and "Not Expert" and again on a scale from "Knowledgeable" and "Not Knowledgeable." Perceptions of "expert" were significantly increased in the expert condition (F = 13.61, 1 and 121 df, means: high = 4.0, low = 3.3, p < .0001, w2 = .084) as were rating of "knowledgeable" (F = 16.59, 1 and 121 df, p < .0001, w2 = .098). Also, through content analysis of the cognitive responses, it was possible to ascertain respondents' impressions of the salespersons' tennis expertise. One content category of the cognitive responses grouped responses mentioning the knowledge or ability of the salesperson. Fourteen responses favorably rated salesperson's expertness and twenty-four responses referred to a lack of knowledge or ability. In the expert condition, positive mentions of expertness were significantly higher (F = 9.70, 1 and 121 df., p < .002, w2 = .058) and negative mentions were significantly lower (F = 15.38, 1 and 121 df., p < .0001, w2 = .079). Clearly, expertise was successfully manipulated. Almost as important was the lack of significant differences in the low and high expert conditions for other ratings of the salesperson. None of the other dimensions were significantly different; the only near exception was "dynamic" which was higher in the expert condition (F = 3.53, 1 and 121 df., p = .063).

The Effect of Expertise

Table 1 presents the raw results on the relationship between expertise of the salesperson and the critical dependent variable, whether or not the subject signed up for the course. Expertise had essentially no effect at all! A chi-square analysis indicated the lack of any significant effect (c2 = .054).



As portrayed in Tables 2 and 3, the results were much the same for the two other major dependent variables, whether or not the subject subsequently paid the five dollar deposit on the course, and whether or not the S at least signed up for the introductory lesson. The respective chi-square values (.060 and .852) were nonsignificant.






Capon (1975) in his research on telephone selling, found major differences among his three salespeople in effectiveness of different message conditions. Individual differences would be a plausible explanation for the aggregate results of no effect in this study. Perhaps some salespeople were more effective in the high expertise condition and others in the inexpert situation.

Table 4 presents the results on an individual basis for two dependent variables - fraction of subjects who signed up for the course, and fraction who at least signed up for the introductory lesson.



It is noteworthy that in all cases but one (Carl, in selling signups for the course), percentage of successes for high and low expertise conditions were not significantly different. In fact, in most situations there was a remarkable similarity between the two. Even then, it is interesting that all four female salespersons and one of the three males were more effective at selling the entire course in the low expertise condition. Of the other two males, one was far less successful in either treatment than any other experimenter (he produced only two course signups as an expert salesman and one in the low expertise condition). The remaining salesperson, Carl, was the only one who seemed to be significantly more successful in the high expertise condition. Although the others were evidently convincing in their portrayals of tennis experts (as indicated by the manipulation check [None of the expert-salesperson interactions were significant for any of the manipulation checks.]), Carl was the only one of the seven who really was a tennis expert, and this factor may have been related to his success in the expert role. However, a comparison of ratings of the individual salespersons showed that Carl was not perceived to be significantly more expert than the other confederates. It is possible that an unconscious desire to demonstrate his expertise may have resulted in Carl's being somewhat more aggressive in closing the sale when in the role of expert. Except for this one case, however, the salespersons tended not to be more effective as experts.


Explanations of a failure to support predicted hypotheses are seldom difficult to generate and not always meaningful or interesting, if only because they tend to be self-serving. However, when a new procedure is applied to fairly well-researched phenomenon and predicted results are not obtained, some investigation of the reasons is warranted before accepting one of two facile conclusions: that the procedure is inappropriate or that the previous research findings regarding the phenomenon were in error. Therefore, some hypotheses have been offered along with a discussion of which appear to be likely explanations and their implications.

At least two hypotheses assume failures to adequately operationalize experimental constructs. (1) The most obvious explanation for any null finding is that the manipulation was unsuccessful. This hypothesis was easily tested because several manipulation checks on perceived expertise were included in the post-test measures. In all cases there was a significant (p < .01) effect for perceived expertise.

(2) The experimental setting was designed to be high on experimental realism. If this goal were not attained, the respondents would not have been involved in the sales attempt and would have reacted indifferently to any manipulated variables. However, in addition to the successful manipulation checks of expertise, cognitive responses indicated high involvement for both high and low expertise respondents. Subjects believed the actors were real salespersons and the offers were legitimate. Post-experimental interviews revealed that nearly all subjects became immersed in the sales situation. They either forgot about the research aspect during the sales presentation, or they considered the research explanation as a pretext for a real sales pitch. A few respondents indicated that it was an unusual way to sell a tennis course, but even these few accepted the cover story. Mundane realism was reduced by the laboratory setting and the tape recording, but no evidence was obtained to indicate that these distractions reduced the believability of the sales pitch. Given the success of the sales pitch in generating actual sales and high interest, we can only conclude that subjects were involved and convinced of the legitimacy of the offer.

Actually, such high overall involvement may itself be an alternative explanation for the null effect of expertise. Previous research has found that sources of high and low credibility produce differential persuasion under low but not high involvement conditions (Johnson and Scileppi, 1969; Rhine and Severance, 1970). Petty and Cacioppo's (in press) research suggests that the effect of non-message cues such as expertise should be maximal in low involvement conditions when information processing (measured by cognitive responses such as in this study) is relatively low. In high involvement conditions, however, message characteristics should have maximal impact. Support for this proposition was offered by Chaiken (1978) who found that subjects were more influenced by the number of arguments within a message under high-than low involvement conditions, but more affected by source attractiveness under low than high involvement.

In other words, it may be that low involvement produces "automatic processing" (Schneider and Shiffrin, 1977) in which expertise is processed without assessment of its relevance or the convincingness of the message, whereas high involvement leads to controlled processing where the message is more scrutinized and evaluated. Thus, it could be possible that achieving high subject involvement in this experimental setting reduced the impact of expertise. If the high involvement of this study meant that message characteristics were likely to dominate any source effects, then one should expect that the results would not differ among conditions since the message was always the same.

Assuming that the theoretical constructs were properly operationalized, there remains the possibility that the laboratory procedure was somehow inappropriate for the particular buyer-seller situation that was studied. Two possibilities are apparent: The expertise effect may be product-specific, and the tennis course may be a product for which salesperson expertise has no effect. Alternatively, the effect of expertise may be masked by other powerful factors introduced with the use of live salespersons and interactive sales pitches.

It is possible that salesperson expertise has a product-specific effect. Three such explanations are available: (1) A common, widely-experienced product, like tennis, may be perceived as sufficiently simple to reduce salesperson expertise to relative unimportance. (2) Because of prior experience or confidence in Ohio State, respondents may have felt so little doubt in the quality of the product as to eliminate the effect of salesperson's expertise. (3) Because the product was standardized and the offer a take-it-or-leave-it one, salespersons may have had no meaningful way to apply their expertise. The first of these explanations seems to be contradicted by the manipulation effect. Since expertise is at least partially a relative phenomenon, the manipulation check indicates that subjects regarded the salespersons as expert vis-a-vis themselves. We can offer no strong refutation for the possibility that expertise was irrelevant because respondents had confidence in the quality of the tennis course. However, a content analysis of the interactions revealed no direct support for such an explanation. The hypothesis does raise interesting implications for management (Levitt, 1965). For instance, if a company has the benefit of a strong reputation does that reduce the effectiveness of or need for expertise of salespeople? On the other hand, if a company is new, does that imply a greater significance of salesperson expertise? Finally, the explanation of standardized product/selling situation may also offer interesting implications. If this explanation were valid, it would suggest that expertise should be most important when customers are confronted with a complex choice among alternatives and salespersons perform a service in matching products with needs. This would be the case with products such as computers or insurance but may not be in many other situations where expertise is nonetheless typically stressed by sales managers.

It might also be asserted that the employed measures were not sufficiently sensitive to detect an effect of expertise. Measures of purchase, purchase intention, or information seeking (the dollar trial lesson) may not have been sensitive, reliable, or valid. Although these problems may have existed, one of the major goals of this study was to measure a series of variables that presumably intervene between a sales presentation and a sale. A series of ratings of the salesperson, the product (the tennis course), and tennis were included. Finally, an extensively coded set of cognitive responses was measured which included several indices such as the total number of responses, the number of positive and negative responses, the number of responses that referred to different targets or originated from different sources, or mentioned various topics in a positive, neutral negative, or curious manner. However, other than the described measures of the effectiveness of the expertise manipulations, none of these measures were significantly influenced by expertise. Although there was no assessment of the reliabilities and validities, it seems unlikely that all of these measures were insufficiently sensitive to measure an effect of expertise if such an effect were actually present. [Most of the analyses of cognitive responses and interaction content analyses are not included because of their lack of positive results, but they are available from the authors.]

The most cogent explanation for the failure to support the predicted relationship between expertise and effectiveness seems to be that the relationship probably was there but was masked by overriding factors that prevented the difference between groups from showing. These overriding factors were the result of a live salesperson and an interactive presentation. Both actors/salespersons and respondents indicated that persons were aggressive and actively attempted to sell every prospect that seemed to be interested. Under such conditions, respondents may have recognized that any sign of interest would be interpreted as yielding and only lead to further efforts by the salesperson. Such a display would manifest itself both during the sales pitch and on subsequent paper-pencil measures. Thus, the single factor that would explain the largest amount of variance in sales situations might be the buyer's inclination to buy. Since the respondents in our sample were in no way predisposed toward purchase, they were apt to be defensive. Moreover, because of the interactive nature of the presentation, respondents were forced to process a great deal of information about the salesperson--his appearance, his personality, his association with the experiment, as well as his message and the questions he asked.

Given this combination of aggressive salespersons, interactive presentations and complete lack of predisposition to buy, it is not unlikely that any effect of expertise, however real, would be overwhelmed by "noise". This noise would be absent in the sales situations of previous research findings: Jolson (1975) found canned presentations more effective than live presentations, when expertise was manipulated. The canned presentation eliminated the noise of interaction and reduced the pressure to defend against persuasion. Both these factors would tend to make the effect of expertise more salient. Busch and Wilson (1976) found a significant effect of expertise only when respondents were instructed to assume they were in the market for life insurance. It is likely that this role-playing tended to reduce the effect of the major explanatory variable--predisposition to buy. This noise reduction made visible the effect of expertise. Brock (1965) and Woodside and Davenport (1974) both conducted field experiments in which the participants were actually purchasers of the product or related products. This self-selection bias assured that respondents were "in the market," thus, again, increasing the salience of the expertise manipulation. However, even when such extraneous variables such as tennis expertise of the subject and individual constraints presented by the product were controlled in correlational analyses, no effect of expertise was found.


The title of this paper asks whether seller-customer interaction and influence can be studied in the laboratory. Although we definitely believe that this answer to this question is affirmative, we must admit that we have not yet demonstrated the sensitivity, let alone the validity, of the method described in this paper. Perhaps this method might be able to detect the effects of other variables such as message or presentation characteristics which might be expected to have a larger effect than salesperson expertise. Despite the lack of success in this exploratory study, we believe that the extended measurement system is worth the added effort. Perhaps a different product is needed. Also, it may be that, in order to study a product that involves a price of more than a few dollars, subjects other than students will be worth the extra research expense.

The research reported in this paper represented an attempt to improve and extend personal selling research. Although the product of an initial effort has been disappointing, we encourage other researchers to join us in either improving our method or designing a completely new one to achieve our goal.


Brock, Timothy C. (1965), "Communicator-Recipient Similarity and Decision Change," Journal of Personality and Social Psychology, 1, June, 650-654.

Brunswick, Egon (1956), Perception and the Representative Design of Psychological Experiments, Berkeley, California: University of California Press.

Busch, Paul and Wilson, David T. (1976), "An Experimental Analysis of a Salesman's Expert and Referent Bases of Social Power in the Buyer/Seller Dyad," Journal of Marketing Research, February, 3-11.

Cacioppo, John T. and Petty, Richard E. (in press), " "in Cognitive Responses in Persuasion, Richard E. Petty, Tom M. Ostrom and Timothy E. Buck (eds.), Hillsdale, New Jersey: Erlbaum Associates.

Capon, Noel (1975), "Persuasive Effects of Sales Messages Developed for Interaction Process Analysis," Journal of Applied Psychology, 60, 238-244.

Capon, Noel, Holbrook, Morris B and Hulbert, James M. (1977), "Selling Processes and Buyer Behavior: Theoretical Implications of Recent Research," Consumer and Industrial Buying Behavior, Arch G. Woodside, Jagdish N. Sheth and Peter D. Bennett (eds.), Elsevier, North-Holland, 323-332.

Capon, Noel and Hulbert, James M. (1976), "Interpersonal Interaction and Persuasion Processes: An Overview," Advances in Consumer Research, Vol. 3, B. B. Anderson (ed.), Cincinnati: Association for Consumer Research, 364-369.

Carlsmith, J. Merrill, Ellsworth, Phoebe and Aronson, Elliott (1976), Methods of Research in Social Psychology, Reading, Massachusetts: Addison-Wesley Publishing Co.

Chaiken, S. (1978), Use of Source and Message Cues in Persuasion. Paper presented at the 86th Annual Meeting of the American Psychological Association, Toronto, Canada, August.

Cook Thomas and Campbell, D. T. (1976), "The Design and Conduct of Quasi-Experiments and True Experiments in Field Settings," in Marvin D. Dunnette (ed.), Handbook of Industrial and Organizational Psychology, Chicago: Rand McNally, 223-246.

Davis, Harry L. and Silk, Alvin J. (1972), "Interaction and Influence Processes in Personal Selling," Sloan Management Review, 80, 97-125.

Deutscher, Terry and Sawyer, Alan G. (1978), "Dependent Measures in Studies of Persuasion in Buyer/Seller Interaction," College of Administrative Science, The Ohio State University, April.

Deutscher, Terry and Sawyer, Alan G. (1978), "Dependent Measures in Buyer/Seller Interactions," C. Leavitt (ed.), Proceedings of Consumer Psychology Division of American Psychological Association Convention, 40-42.

Evans, Franklin B. (1963), "Selling as a Dyadic Relationship: A New Approach," American Behavioral Scientist, 6, 76-79.

Farley, John V. and Swinth, Robert L. (1967), "Effects of Choice and Sales Message on Customer-Salesman Interaction," Journal of Applied Psychology, 51, April, 107-110.

Gadel, M. S. (1964), "Concentration by Salesmen and Congenial Prospects," Journal of Marketing, 28, 64-66.

Holbrook, Morris and O'Shaughnessy, John (1976), "Influence Processes in Interpersonal Persuasion," Advances in Consumer Research, Vol. 3, B. B. Anderson (ed.), Cincinnati: Association for Consumer Research, 364-369.

Hovland, Carl, Janis, I. L. and Kelley, H. (1953), Communication and Persuasion, New Haven: Yale Press.

Hulbert, James and Capon, Noel (1972), "Interpersonal Communication in Marketing: An Overview," Journal of Marketing Research, Vol. IX, February, 27-34.

Johnson, H. H. and Scileppi, J. A. (1969), "Effects of Ego-Involvement Conditions on Attitude Change to High and Low Credibility Communicative," Journal of Personality and Social Psychology, 13, 31-36.

Jolson, Marvin A. (1975), "The Underestimated Potential of the Canned Sales Presentation," Journal of Marketing, 39, January, 75-78.

Jolson, Marvin A. (1976), "The Underestimated Potential of the Canned Sales Presentation: A Reply to Reed," Journal of Marketing, Vol. 40, January, 68-70.

Levitt, Theodore (1965), Industrial Purchasing Behavior: A Study of Communications Effects, Division of Research, Graduate School of Business Administration, Harvard University, Boston.

Mathews, H. Lee, Robeson, James and Bambic, Peter J. (1977), "Achieving Seller Acceptability in Industrial Markets: Development of the Communications Mix," Consumer and Industrial Buying Behavior, Arch G. Woodside, Jagdish N. Sheth and Peter D. Bennett (eds.), Elsevier, North-Holland, 221-228.

Nisbett, Richard E. and Wilson, Timothy DeCamp (1977), "Telling More Than We Can Know: Verbal Reports on Mental Processes," Psychological Review, Vol. 84, No. 3, May, 231-263.

Olshavsky, Richard W. (1973), "Customer-Salesman Interaction in Appliance Retailing," Journal of Marketing Research, 10, May, 208-212.

Petty, Richard E. and Cacioppo, John T. (in press), "Issue Involvement Can Increase or Decrease Persuasion by Enhancing Message-Relevant Cognitive Responses," Journal of Personality in Social Psychology.

Ray, Michael L. (1978), "The Present and Potential Linkages Between the Micro-theoretical Notions of Behavioral Science and the Problems of Advertising: A Proposal for a Research System," in Davis, H. and Silk, A. J., Management and Behavioral Science in Marketing, New York: Wiley.

Reed, Jim D. (1976), "Comments on 'The Underestimated Potential of Canned Sales Presentations'," Journal of Marketing, Vol. 40, January, 67.

Reizenstein, Richard C. (1971), "A Dissonance Approach to Measuring the Effectiveness of Two Personal Selling Techniques Through Decision Reversal," Proceedings, American Marketing Association Fall Conference, 176-180.

Rhine, R. L. and Severance (1970), "Ego-Involvement, Discrepancy, Source Credibility and Attitude Change," Journal of Personality and Social Psychology, 16, 175-190.

Schneider, W. and Shiffrin, R. M. (1977), "Controlled and Automatic Human Information Processing: I. Detection, Search, and Attention," Psychological Review, 84, 1-66.

Sternthal, Brian, Scott, Carol A. and Dholakia, Ruby Roy, "Self-Perception as a Means of Personal Influence," Advances in Consumer Research, Vol. 3, B. B. Anderson (ed.), Cincinnati: Association for Consumer Research, 387-393.

Varela, Jacobo A. (1971), Psychological Solutions to Social Problems, New York: Academic Press.

Wilson, David T. (1975), "Organizational Buying: A Man and Machine Information Processing Approach," Communication Research, 2, 279-287.

Wilson, David T. (1977), "Dyadic Interaction," Consumer and Industrial Buying Behavior, Arch G. Woodside, Jagdish N. Sheth and Peter D. Bennett (eds.), Elsevier, North-Holland, 355-365.

Woodside, Arch G. and Davenport, J. William Jr. (1974), "The Effect of Salesman Similarity and Expertise on Consumer Purchasing Behavior," Journal of Marketing Research, 11, May, 198-202.

Woodside, Arch G. and Sims, J. Taylor (1976), "Retail Sales Transactions and Customer 'Purchase Pal' Effects on Buying Behavior," Journal of Retailing, 52, Fall, 57-64,95.



Alan G. Sawyer, The Ohio State University
Terry Deutscher, University of Western Ontario
Carl Obermiller, The Ohio State University


NA - Advances in Consumer Research Volume 07 | 1980

Share Proceeding

Featured papers

See More


Q13. Liquid Consumption From Another Perspective: The Case of “Investomers”

Carina Hoffmann, Heinrich-Heine-University
Lasse Meißner, Heinrich-Heine-University
Peter Kenning, Heinrich-Heine-University

Read More


Brands as Complex Social Processes

Andrea Hemetsberger, University of Innsbruck, Austria
Hans Mühlbacher, International University of Monaco
Eric J Arnould, Aalto University, Finland

Read More


Divorcing the Market

Deniz Atik, University of Texas Rio Grande Valley, USA
A. Fuat Fırat, University of Texas Rio Grande Valley, USA
Ebru Uzunoğlu, Izmir University of Economics

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.