Conceptual Underpinnings For the Use of Group Interviews in Consumer Research

Edward F. McQuarrie, Santa Clara University
Shelby H. McIntyre, Santa Clara University
ABSTRACT - The authors provide two justifications for the use of focus groups in consumer research. First, the kind of generalizable knowledge which qualitative research can offer is discussed. This is distinguished from the kind of generalization which survey research permits. Second, the advantages of group qualitative research over individual qualitative research are considered. It is shown that the biases created by interpersonal processes in groups may actually be advantageous to the marketer. The result of the paper is a sharper distinction between what focus groups can and cannot do for consumer researchers.
[ to cite ]:
Edward F. McQuarrie and Shelby H. McIntyre (1988) ,"Conceptual Underpinnings For the Use of Group Interviews in Consumer Research", in NA - Advances in Consumer Research Volume 15, eds. Micheal J. Houston, Provo, UT : Association for Consumer Research, Pages: 580-586.

Advances in Consumer Research Volume 15, 1988      Pages 580-586

CONCEPTUAL UNDERPINNINGS FOR THE USE OF GROUP INTERVIEWS IN CONSUMER RESEARCH

Edward F. McQuarrie, Santa Clara University

Shelby H. McIntyre, Santa Clara University

ABSTRACT -

The authors provide two justifications for the use of focus groups in consumer research. First, the kind of generalizable knowledge which qualitative research can offer is discussed. This is distinguished from the kind of generalization which survey research permits. Second, the advantages of group qualitative research over individual qualitative research are considered. It is shown that the biases created by interpersonal processes in groups may actually be advantageous to the marketer. The result of the paper is a sharper distinction between what focus groups can and cannot do for consumer researchers.

INTRODUCTION

It is our impression that a great deal of confusion continues to surround the topic of focus groups. Although the technique has long fascinated researchers, and is now very popular, the justifications typically offered for its use are not always persuasive. The same holds true for the criticisms commonly directed against the technique. What is lacking is a sound basis for judging when the group interview is the technique of choice, and when it ought to be avoided. While the seminal article by Calder (197?) made considerable strides in this direction, the subtleties of his treatment have not always been heeded. Too often today the choice to use or eschew focus groups remains based on tradition, fashion or opinion, rather than on a solid analysis of the technique's true strengths and weaknesses.

The purpose of this paper is to contribute a better theoretical perspective on what a focus group really does. Two questions have to be disentangled in order to address this issue. First it is necessary to contrast the possible contribution of qualitative as opposed to quantitative research. While the limits of qualitative research are well known, less attention has been paid to its unique strengths. Second, it is necessary to understand how group depth interviews differ from individual depth interviews. Because focus groups are not the only form of qualitative research, a justification of qualitative research alone would be insufficient. We will argue that interpersonal processes known to characterize the group setting may actually assist the marketer. Rather than contaminating the results, these group processes can strengthen the inquiry.

The supposed limitations of qualitative research are well-known. Most qualitative studies acknowledge these limits with caveats to the effect that the findings are not projectable, should be taken as suggestive only, and cannot stand on their own. Market research texts derive these limits in a rigorous, scientific fashion, pointing to the consequences of biased sampling procedures and inadequate sample sizes. But this rigor disappears when discussion turns to the strengths of qualitative research. One hears only that the technique yields "deeper" and "richer" information. This is not an adequate defense, particularly once it is realized that hundreds of millions of dollars are spent annually on qualitative research (Goldman and McDonald 1987). We think that the strengths of this technique should be conceptualized in the same rigorous manner as the weaknesses. Calder (1977) was among the first to attempt this task.

Calder's Defense of Qualitative Research

Calder's (1977) article did much to challenge accepted assumptions about focus groups. Although our treatment of the matter differs from his, a review of his contribution will provide a starting point. Calder argued that a focused group discussion is neither a proto- nor a pseudo-survey, but a different creature altogether, with different aims and assumptions. A key concept in his treatment is the distinction between everyday knowledge and- scientific knowledge. Everyday knowledge is represented by consumers' own understanding of, and explanations for, what it is they do when consuming. Calder observes that at the present time, the everyday knowledge held by consumers has a claim to validity that is at least equal to that of the scientific knowledge produced by the research community. The phenomenological type of group provides the means to capitalize on this insight. Unlike the clinical approach, where the analyst's command of depth psychology enables him to penetrate beyond consumer knowledge, or the exploratory approach, which uses the group discussion merely as a trigger to generate scientific hypotheses, in the phenomenological approach the purpose is simply to uncover the everyday knowledge that consumers possess.

In Calder's view, it makes little sense to talk about phenomenological focus groups as a preliminary research effort yielding tentative findings that must be confirmed by survey research. The two techniques seek different kinds of knowledge: the one the everyday knowledge of the consumer, the other the scientific knowledge of the researcher. Surveys cannot confirm focus groups in any strict sense because they do not and cannot uncover the same types of knowledge. However, Calder leaves unresolved the question of the generalizabilily of a focus group study, saying only that, 'The best way to generalize ... is to conduct additional groups" (p. 361).

Although indebted to Calder's treatment of these issues, (in particular his suggestion that focus groups uncover a different kind of knowledge), we believe that a more satisfactory answer can be given to the question of generalizability. And in fact, understanding the kinds of generalizability which it is possible for focus groups to attain is key to understanding the contribution of qualitative research in general. Moreover, such an analysis will also pinpoint what it is that focus groups cannot do, and the kinds of research objectives which are inappropriate for this tool.

Two Kinds of Generalization

Imagine a company with the following concerns:

The management of Nuphone Corporation wants to explore consumer reactions to an electronic telephone product currently at the concept stage.

They need to understand the nature of the appeal that electronic phones hold, and any fears or negative perceptions that the new technology involves.

In general, they want to know how electronic phones are perceived in comparison to more conventional phones.

These are the kinds of questions often addressed with qualitative research: questions concerning likes and dislikes, reasons and motives, needs and expectations. The typical justification for the approach is that qualitative research, in the form of focus groups, can provide "deeper and richer" insights into these matters. We think a much more convincing and specific justification can be given. Our argument is straightforward: 1) that a researcher in the situation described above could be interested in either of two distinct types of generalization; and 2) that one of these types can be achieved through qualitative research, while the other cannot. We shall refer to a Type 1 generalization (which qualitative research can obtain) as an "existence" generalization, and to a Type 2 generalization as an "incidence" generalization.

An examination of Exhibit 1 will help to clarify -these matters. One research objective could be to identify all the major reasons that motivate the acceptance or rejection of the electronic phone (a Type 1 generalization). These would include findings to the effect that the ring of an electronic telephone is more annoying to consumers than the ring of an ordinary phone, or that electronic phones are considered more prone to breakdown. The second objective might be to discover the incidence of any of these perceptions in the population (a Type 2 generalization). Here the answers would take the form of, "Two-thirds of the consumers in this research compared the beeping sound of electronic phones unfavorably to the ringing of an ordinary phone," or "Fewer than one in ten consumers were concerned about the durability of their electronic phones . "

Most of the standard critiques of the focus group discussion concern the inability of this technique to perform the second type of generalization with any adequacy (Advertising Research Foundation 1985; Seymour 1987; Wells 1974; Yoell 1979). Researchers comment on the small sample sizes in most group studies, the lack of random sampling, and the susceptibility of respondents to group pressure. We completely agree with this critique. We hold focus group studies to be incapable of providing accurate estimates of the incidence in the population of specific likes, reasons, or motives. And, we recognize that many market research studies are crucially concerned with the second type of generalization. An example would be segmentation analyses where one must estimate the relative size of several different target groups.

EXHIBIT 1

TWO TYPES OF GENERALIZATION STEMMING FROM A FOCUS GROUP STUDY

Where we part company with existing commentary is in our interpretation of this limitation. We see it not as a characteristic flaw of the technique, but simply as a warning about how not to use the technique. As any practicing moderator will attest, it is a warning that bears repeating. Most management education is biased toward quantitative techniques, and it is a rare client, new to focus groups, who can resist the temptation to state qualitative findings in numerical terms: "Five of the six groups were solidly in favor of the concept," or "Six out of ten group members agreed that the product was worth more than $100," or "Less than 10% of the respondents were indifferent to the key feature developed by our engineers." Any such conclusion from a qualitative study is unwarranted. This is not so much because it belies what-actually happened in the group, but rather, because it suggests a kind of generalizability that cannot be obtained from qualitative research.

The danger in this critique is that it obscures the contribution that qualitative research can make, which rests on the kind of generalization which is possible with focus groups. The proper task for qualitative research concerns the first type of generalization: the identification of those responses that do exist among consumers. Here the research objective is very different; it is not to determine how many people dislike the ring of an electronic phone, but whether or not there are any consumers who dislike this ring, and the kind of reasons they give in explanation of this reaction. The outcome of the research project is not a judgement of how frequent some particular response is,-but a listing and discussion of those responses that did occur in the groups. In other words, groups yield testimony to the effect that, "These are the kinds of things that consumers care about when they make purchase decisions for electronic phones." In evaluating the capabilities of qualitative research, the crucial question becomes: How well does the set of consumer responses generated in the group reflect the set of responses that actually characterizes consumers in general? The task now is to generalize from the set of ideas occurring in the group to the set of ideas that exist in the population, and not from the number of people holding a specific idea in the group to the number holding that idea in the population.

Given four assumptions, this task appears to be well within the capabilities of a typical focus group study. The first assumption is that with regards to most consumer issues, there are a relatively small number of characteristic responses, on the order of a dozen. In other words, there are only so many reasons why one might like or dislike an electronic phone product. (Here one recalls Fishbein's and Ajzen's (1975) contention that most actions will have five to nine salient consequences.) The second assumption is that most group members will be aware of, or capable of, more than one response to a product issue under study. The third assumption is that respondents in the group are in fact members of the population of consumers that one wishes to study. (Note that the group members do not have to be randomly or systematically sampled from that population to fulfill this requirement.) The final assumption is that respondents are recruited independently of one another. Although this does not guarantee a heterogenous group that will reflect the diversity in the population, it at least makes an excessively homogenous group unlikely.

Granting these assumptions, the question about Type 1 generalizability can be rephrased as, "How likely is it that four groups of eight people each, talking for two hours, will touch upon all dozen motives for purchasing an electronic phoner Although this question has not been addressed by any empirical work, intuitively the answer would seem to be, "Very likely." True, the respondents have not been sampled randomly; but even so the odds remain very much against bringing together thirty-two people, everyone of whom is unaware of several characteristic reasons for liking or disliking an electronic phone. Yes, some group members may dominate interaction and sway others to their views, but it is another matter altogether to claim: 1) that all four groups will fall prey to such individuals; 2) that these dominant members will actively suppress responses, rather than simply producing the appearance that a majority favors their views; 3) that the dominant members will suppress the same responses in every group; and, 4) that all of this will occur despite the efforts of a skilled moderator working to prevent such an outcome. Similarly, it is a fact that conformity pressures do operate in groups, that respondents may attempt to please the moderator, and that the moderator may bias the group. And, all of the above factors destroy any hope of estimating the incidence of any of the dozen reasons in the population at large. But, none of these biases and limitations would seem potent enough to cause the complete suppression of any reason that is a common occurrence among consumers. ("Common' might be defined as held by 20% or more of the population.) In which case, one must conclude that focus group studies do permit Type 1 generalizations.

At this point someone might object that we have "burned the village in order to save it." If qualitative group research can truly say nothing about incidence; if all such research can provide is the assurance that the reasons discussed in the group do operate among consumers at large; then, how valuable is the technique? Here it helps to remember that although there may be only a small number of reasons for liking or disliking electronic phones, there can be no assurance that the product planners for Nuphone Corporation are aware of all these reasons. Moreover, there can be no guarantee that the reasons the Nuphone product development people do have in mind bear much resemblance to what consumers actually feel. Although one may be limited to discovery of the occurrence or absence of a consumer response, such knowledge can be of great benefit.

It is also the case that group qualitative research is not so limited in these matters as individual interviews. As the next section will show, some rudimentary grasp of the relative incidence of certain ideas can be gained through group interviews. Group discussions can at least separate common from idiosyncratic consumer responses. Several distinctive properties of the group environment will be seen to confer this advantage.

GROUP VERSUS INDIVIDUAL INTERVIEWS

The goal of qualitative research is thus to identify consumer reactions that do exist, and not to determine the incidence of these reactions in any larger population. However, the focus group is not the only qualitative technique that can achieve these goals. Why bother to interview consumers in a group when one could just as well talk to them individually? Wouldn't it be better to avoid altogether the potential for social influence, conformity, and undue dominance by certain individuals? If nothing else, consider the degrees of freedom that are sacrificed when consumers are combined into groups. Thirty-two consumers interviewed individually would allow for thirty-two different reactions to emerge, uninhibited by the presence of other consumers with their own agendas. The same people interviewed in groups of eight would yield only four degrees of freedom. As stated earlier, a Justification for qualitative research is not sufficient to be a justification for focus groups.

Only if groups convey some unique advantage can the risks of social influence and conformity pressures, and the loss of degrees of freedom, be justified. Two arguments in favor of group over individual interviews can be identified. The gist of each argument is that group processes do bias the outcome in terms of the information gained from the research; but, that this bias is actually useful because it works in a direction that enhances the quality of this output. In brief: 1) groups will spend more time discussing ideas that are common to many consumers; and 2) group interviews, in comparison to individual interviews, minimize the role of the researcher/interviewer. These points will be developed to show the unique benefits that accrue from group interviews.

To begin, assume that there are only three types of thoughts about any given element of a group discussion topic: 1) widely shared thoughts (those that just about everyone harbors); 2) segment thoughts (those that are specific to a given segment); and 3) idiosyncratic thoughts (those that are unique to a given individual). These thoughts are depicted in Exhibit 2 in terms of the idea structure that would exist in an idealized focused group. However, it should be realized that the idiosyncratic ideas are certainly not restricted in any real setting to one per individual, as we have shown in the idealized case. Rather, it is probable that the idiosyncratic ideas, for any one individual, actually outnumber the common ideas.

EXHIBIT 2

THOUGHT STRUCTURE IN AN IDEALIZED FOCUSED GROUP

Thoughts Are Not Expressed at Random

We hypothesize that in a focus group discussion, participants will not bring up their ideas at random. Rather, they will tend to bring up fat those ideas that they feel are common to other members of the group. As discussion continues and the widely shared ideas are exhausted, some of the segment level ideas will tend to emerge, while individuals continue to suppress what they feel are more idiosyncratic thoughts. We hypothesize, on the other hand, that during an individual in-depth interview (particularly one characterized by probes such as, "Anything elser'), respondents will be much more likely to bring up their idiosyncratic ideas. In fact, the idiosyncratic ideas may tend to come forth early and cause the common ideas to be missed by respondents.

Another group dynamic at work is the "tag along" comment (or nod of agreement). When an idea is brought up in a group, another individual will often encourage this behavior (either explicitly or implicitly). Thus, no matter what ideas are mentioned, the ones that become discussion points tend to be the shared ones. The idiosyncratic points again get washed out of the process. Contrast this to the individual in-depth interview where interviewer bias must be guarded against. In that case, any type of idea can come forward with equal probability.

Limitations on airtime have a similar effect. During a two hour focus group with eight people, if each person gets a fair share of speaking time, the total allocated to each person would be fifteen minutes. The rest of the time, that person is evaluating what others are saying and contemplating what he/she should add to the conversation. In the individual interview, the respondent is just left to roll along, with one idea leading to another, save for general redirection by the interviewer to the topic under discuss on.

Differential Sampling of Ideas

Putting these notions together, we arrive at the diagram in Figure 1. This relaxes the strict dichotomies of Exhibit 2 and grants the fact that ideas exist along a continuum of frequency in the population. The vertical axis in Figure 1 indicates the proportion of the total interview time devoted to a specific reaction, motive or other consumer response. This vertical axis describes an observed variable, and refers to all the interviews conducted within a given study. Airtime is assumed to reflect the depth and detail of the information gathered on any particular response. The horizontal axis orders ideas by their true (unobservable) incidence in the population of consumers, with ideas common to many people arranged to the left, and ideas held by few people, or idiosyncratic to one person, arranged to the right. The lines should be imagined as connecting a finite number of points, each representing a distinct consumer response to the product under discussion (i.e., a specific reaction to the sound made by electronic phones in comparison to regular phones). As shown by the graph, we argue that groups devote much less airtime to ideas that are relatively idiosyncratic. By contrast, while individual interviews will unearth a greater number of distinct consumer responses than will groups, a higher - proportion of these responses will be idiosyncratic. The assumption bears repeating that airtime is monotonically related to the depth and detail of the discussion of the idea.

It should be apparent that the graph supports not an argument that groups are ipso facto better, but rather, the position that groups are best for some things while individual interviews are best for other purposes. The graph is consistent with the findings of Fern (1982), one of the very few studies to conduct an empirical comparison of group and individual interviews. Fern found that consumers working individually produced more ideas in an idea generation task (ways to recruit more women into the armed forces). Previously psychologists had come to the same conclusion in research on brainstorming: a given number of individuals will produce more ideas than the same number of people combined into groups. This finding makes intuitive sense if one considers the degrees of freedom argument (thirty-two people working independently versus four groups of eight), and the limits on airtime available to any one person in a group.

FIGURE 1

HYPOTHETICAL ALLOCATION OF DISCUSSION TIME AMONG IDEAS IN GROUP VERSUS INDIVIDUAL INTERVIEWS

There are obviously a number of circumstances where marketers would desire the greatest possible diversity of consumer response, and the greatest number of distinct ideas. One thinks of the idea generation stage of the new product development process. Individual interviews are recommended here (McQuarrie and McIntyre 1986). But in many other cases, what marketers desire is not the one-of-a-kind reaction, or the unique perspectives of individual consumers, but the representative response: the reason or motive that is characteristic of many consumers. If it is true that group discussions tend to concentrate on ideas shared by most of the members present, then this structural bias could prove very valuable. Group discussions would serve as a filter that naturally works in the direction the marketer desires. Note that we do not argue that individual interviews will miss or totally overlook common perceptions or reactions; in fact, at the global level, we would presume that both techniques would be capable of identifying the same set of three or four very prevalent ideas.

The case for group interviews thus rests on the fact that marketers are most often concerned not with individuals but with market segments: homogenous groupings of individuals. Under these circumstances the idiosyncratic reactions of an individual consumer are of little value. In fact, the tendency of individual interviews to unearth reactions which are unique to the respondent can be seen as a negative feature. When qualitative research is done, the research sponsor may have no prior knowledge of what is and is not a common idea among consumers. The possible deleterious consequences of relying on individual interviews in this situation can be developed with reference to Figure 1. In terms of the horizontal axis, in one-on-one interviews the proportion of unique ideas to the total number of ideas identified will be high. The danger is that the researcher will be misled into thinking that certain of these ideas are representative or typical when in fact they are not. With reference to the vertical axis, note that the slope of the frequency counts is much flatter for individual interviews than for groups. The difference in frequency Counts between ideas which are truly common and those which are relatively idiosyncratic is much less. The danger again is that the results of the research will not clearly distinguish between ideas which represent basic consumer reactions to the product, and ideas which are relatively unusuaL specific, and of limited relevance to marketing strategies.

By contrast, the slope of the group line is much steeper. Hence, the sponsor has some assurance that ideas which dominate the group discussion will in fact prove representative of widely shared reactions. Thus, although incidence generalization from groups is not advisable, groups should enable one to make gross distinctions between common and idiosyncratic consumer perceptions. Of course, it remains the case that a notion which occupies a large amount of airtime in the group may prove to characterize only 20-25% of the consumer population; and similarly, that a notion which receives a moderate share of airtime in the group is actually the single most common consumer response in the population at large. As argued earlier, no precision is possible in the estimate of incidence when a qualitative technique is used. But when group qualitative research is done, one can be assured that the ideas which dominate the group discussion are all of common occurrence. In our judgement, this is the signal advantage of group over individual qualitative interviews.

The Polarization Phenomenon

In the preceding we have presented a conceptual analysis in support of group interviews. The propositions are graphically represented by the two lines drawn in Figure 1. While it should be possible to test these propositions, we know of no empirical studies that directly address the matter. However, there is a long tradition of research on group dynamics in social psychology which indirectly supports our formulation. The key phenomenon which led us to hypothesize the relations shown in Figure 1 is known as polarization (Burnstein and Schul 1983). In earlier research polarization in groups was studied under the heading of the "risky-shift." It had been found that a group would tend to make riskier decisions than the average level of .risk acceptable to individuals within that group, as determined by measures taken prior to the group discussion. In later research, it was also found that groups would sometimes make more conservative decisions than the average of the individuals in the group. The key in both cases is that the group tended to exaggerate whatever tendency had preexisted among the set of individuals comprising the group. If risk takers predominated, the group made an even more risky decision than could have been predicted by knowing the number of risk taking individuals; and if risk-averse types predominated, then the group made a more cautious than predicted decision. In other words, group discussions tend to exaggerate whatever pre-existing tendencies characterized the majority of people in the group. At the same time, minority views tend to be suppressed.

Polarization, in this first sense of the word, thus operates to protect the research sponsor from being distracted by ideas that are idiosyncratic to individual consumers. The ideas most likely to dominate a group discussion are those shared by most of the people in the room. (This factor, of course, is what makes groups an inferior technique for idea generation.) Ideas which characterize only one individual tend not to receive much airtime, either because people self-censor these contributions, or because the group does not pick them up.

But there is also a second sense of the word polarization which is relevant here. This is polarization within the group, as originally discussed by Bales and Cohen (1979), and developed for this context by Goldman and McDonald (1987). A structural property of groups, in Bales view, is that coalitions will tend to form among like-minded individuals, creating sub-groups within the group. A unification dynamic draws together any two or three or four members who are somewhat similar to one another, and, a polarization dynamic forces the coalitions so formed apart from one another. In other words, in the typical focus group we can expect that respondents will divide into opposing camps. These camps will then dispute with one another about what the key benefits of the product are, the nature of its most desirable features, and so forth.

The relevance of this phenomenon to the study of segments within the consumer population is striking. After all, the ideal segmentation scheme is one that maximizes similarity within segments, while also maximizing dissimilarity between segments. What better way to grasp the segmentation possibilities within a market than to watch two coalitions within a group dispute the merits of their positions relative to one another? Again, one will learn nothing about which segment is the larger one; but one will learn a great deal about the differences that divide any two segments, and about the factors that unite consumers within each segment.

Here is a case where groups do not simply possess an advantage over individual interviews, but actually exhibit a feature which cannot be duplicated in one-on-one interviews. An interviewer speaking to a single respondent cannot argue with that respondent; cannot deny the merits of his views; cannot reject the value of his perspective: in a word, cannot oppose the respondent. By contrast, all of these behaviors are possible between any two respondents in a group discussion. When channeled by a skilled moderator, this adversarial energy between coalitions can lead to a much sharper differentiation of respondent needs and wants. As can be seen in Figure l, the group and individual lines diverge most sharply in the case of ideas which are moderately Common. These, of course, are ideas held by a sub-group of consumers, what we have termed segmented ideas. The tendency of coalitions to polarize is what leads to the relatively large fraction of airtime devoted to these kinds of ideas within group discussions.

CONCLUSION

Our purpose in this paper was to place the use of group discussions on firmer ground. By advancing specific propositions concerning what groups can and cannot do, we hope to have clarified the role of qualitative techniques in consumer research. Building on the foundation provided by Calder (1977), our defense of qualitative research has taken a different form than those commonly offered. Rather than extol its benefits with vague and difficult to measure terms such as "richer" and "deeper," we sought to portray the kinds of generalizable knowledge that can be gathered with the aid of these techniques.

Our arguments in favor of group over individual qualitative research similarly depart from conventional positions. We made a virtue out of what is typically seen as the Achilles heel of focus groups: the operation of interpersonal influence processes. We hypothesized that the biases created by such processes are favorable to the purpose that guides most market research studies. Group discussions are biased in favor of ideas that most consumers hold, and group discussions also tend to polarize among opposing sub-groups. These interpersonal processes serve as a useful filter, and can also act to enhance contrasts of interest.

It hardly needs stating that the positions we have taken are hypothetical and as yet unsupported by experimental investigations. We have sought to develop our views in sufficient detail to make such empirical work possible. The propositions advanced in this paper are, we believe, eminently testable, and they do build on earlier theoretical work. A study such as Fern's (1982) provides a model for the kinds of experiment that need to be done. In the meantime, the ideas presented here may stimulate further work on what remains a poorly understood consumer research technique: the focused group discussion.

REFERENCES

Advertising Research Foundation (1985), Focus Groups: Issues and Approaches. New York: Advertising Research Foundation.

Bales, Robert F. and Stephen P. Cohen (1979), SYMLOG: A System for the Multiple Level Observation of Groups, New York: Free Press.

Burnstein, Eugene, and Yaacov Schul (1983), "Group Polarization," in Small Groups and Social Interaction, Vol. 2, eds. H. Blumberg ct. al., New York: Wiley.

Calder, Bobby J. (1977), "Focus Groups and the Nature of Qualitative Marketing Research," Journal of Marketing Research, 14 (August), 353-64.

Fern, Edward F. (1982), "The Use of Focus Groups for Idea Generation: The Effects of Group Size, Acquaintanceship, and Moderator on Response Quantity and Quality," Journal of Marketing Research, 19 (February) 1-13.

Fishbein, Martin and Icek Ajzen (1975), Belief, Attitude, Intentions and Behavior, Boston: Addison-Wesley.

Goldman, Alfred E. and Susan Schwartz McDonald (1987), The Group Depth Interview, Englewood Cliffs, NJ: Prentice Hall.

McQuarrie, Edward F. and Shelby H. McIntyre (1986), "Focus Groups and the Development of New Products by Technologically Driven Companies: Some Guidelines," Journal of Product Innovation Management, 3 (March), 40-47.

Seymour, Daniel T. (1987), "Focus Groups and the Development of New Products by Technologically Driven Companies: A Comment," Journal of Product Innovation Management, 4 (March), 50-54.

Wells, William D. (1974), "Group Interviews," in Handbook of Marketing Research, ed. Robert Ferber, New York McGraw-Hill.

Yoell, William A. (1979), "How Useful Is Focus Group Interviewing?" in Focus Group Interviews, eds. Keith K. Cox and James B. Higginbotham, Chicago: American Marketing Association.

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