Focus Groups: Theory and Method

Martin R. Lautman, ARBOR, Inc.
ABSTRACT - Theoretical issues and practical considerations in the conduct and analysis of focus groups are presented and discussed. Explanatory qualitative analysis information requirements are derived from an expectancy value model of attitudes. Exemplar questions illustrating how these information requirements can be incorporated into a moderator's focus group outline for a specific topic are also presented.
[ to cite ]:
Martin R. Lautman (1982) ,"Focus Groups: Theory and Method", in NA - Advances in Consumer Research Volume 09, eds. Andrew Mitchell, Ann Abor, MI : Association for Consumer Research, Pages: 52-56.

Advances in Consumer Research Volume 9, 1982      Pages 52-56


Martin R. Lautman, ARBOR, Inc.


Theoretical issues and practical considerations in the conduct and analysis of focus groups are presented and discussed. Explanatory qualitative analysis information requirements are derived from an expectancy value model of attitudes. Exemplar questions illustrating how these information requirements can be incorporated into a moderator's focus group outline for a specific topic are also presented.


Probably no market research methodology is employed as regularly by practitioners with as little understanding as focus groups. Like putty, the technique seems to be readily adaptable to whatever research problem is extant. It just seems so easy. Get a few people together and listen to them talk about whatever. However, when we take a closer look at focus groups we come up with some nagging questions.

Today I am going to address those questions and try to infuse some method into the madness in which focus groups currently operate. I'm going to divide my presentation into three main sections. In the first section I will discuss some of the philosophical questions and issues which underlie the use of focus groups.

In the second section will address six methodological and analytic issues on which focus group theoreticians and practitioners seem to disagree. In each case I will offer a practical resolution of each controversy which seems to have worked well for me. I will conclude this section with a brief discussion of four additional issues which, in my opinion, have not received sufficient attention from either theoreticians or practitioners.

Finally, in the third section I will conclude with a brief presentation of a focus group outline structure which we at ARBOR have used in guiding our exploratory qualitative work. This structure for a moderator's guide represents the synthesis of ARBOR's efforts to bring a systematic approach to one key aspect of groups -- the focus group outline generation process. We have found that the discipline introduced by this formalized outline structure has tended to reduce some of the reliability problems inherent in focus groups, such as those caused by different moderators not covering the same content areas in comparable ways.

Do Focus Groups Work?

I would like to start off our session today by asking the fundamental question, "Do focus groups work?" Reasonable arguments which dictate against the usefulness of focus groups have been made. For example, Yoell (1974) has written that he considers the whole focus group business a "grand charade." Among his many arguments is the highly compelling one that "group reactions may have no counterpart in the behavior of individuals." Lending empirical support to his contention, he reported a study where panelists were debriefed after a group session. The panelists admitted to saying things during the group session which were diametrically opposite to their (reported) actual behavior. While one might question whether respondents are more likely to tell the truth in a focus group or in individual debriefing sessions, the presence of contradictory reports is disquieting. The dynamic group process that we are so fond of citing as a reason for using focus groups as opposed to other methods of qualitative data collection may yield results that bear little resemblance to the individual decisions and behavior which they are oftentimes conducted to reveal.

A second problem pertaining to the question of whether focus groups work is that a description of behavior, in terms of attitudes and emotional states, in an environment different from that in which it took place can be expected to produce different results (Kanfer and Phillips, 1970). Although it is not a problem exclusive to focus groups, the fact that reported attitudes, emotional states and behaviors may not accurately reflect those which existed at the time of the activity may be exacerbated by the "foreign" focus group environment.

Numerous studies have noted the limited credibility one should place in verbal reports of behavior rather than the observation of behavior itself (for example, Azrin, Holz and Goldiamond, 1961). Conversely, one must rely on verbal reports to identify private thoughts and feelings which enter into a person's decisions and/or behavior since these cannot be observed. This would seem to suggest that greater insight into the determinants of a decision or action requires more than just outsider observation. Whether this insight can better be provided by the group environment of a focus session or in the relative privacy of an individual interview, however, is problematical. In any case, we would expect greater accuracy in self-reports which pertain to phenomenological experiences (those effects directly determined by a stimulus) than for cognitive experiences (those involving verbal inferences about experienced effects).

Kanfer and Phillips (1970), following in the Skinnerian tradition as discussed by Bem (1967), attempted to resolve the verbal report versus outsider observation controversy by suggesting that the same laws apply both to private and to observable behavior, stating that "... there is no reason to assume ... that private behaviors are essentially different from publicly observable behaviors or that their relationships to each other and to external events follow different laws." Moveover, whatever this hypothesis can be considered to apply equally well to a markedly different and new environment from that in which the activity originally occurred remains to be determined.

Why Do Focus Groups Seem to Work?

If one accepts the concept that focus groups do work, a reasonable question is "why?" I think there are at least three relevant theoretical threads which can help to suggest some of the mechanisms underlying the dynamics of focus groups. One is reinforcement theory. From the time of the introduction and warm-up, the moderator reinforces behaviors such as participation, openness, and speaking up. Panelists who exhibit these behaviors are positively reinforced. Those who don't are called upon and encouraged. Methods of administering the reinforcement can include body language, facial expressions, respondent selection and non-selection and verbal conditioning (through the systematic application of generalized or specific verbal reinforcers). A skilled moderator will recognize that these reinforcement contingencies may not only help to foster the group process but can also have the effect of subtly supporting his own biases (Rosenthal, 1961).

A second thread is related to an "assembly effect" where the presence of others creates a state of psycho-physiological arousal that intensifies attention, participation, competition, emotions and self-evaluations leading panelists to become highly involved in their task. As described by Zajonc (1965), this state tends to encourage and enhance "well-practiced" responses and suppress new or "weakly-practiced" responses. This might suggest that the focus group environment may necessitate additional effort on the part of a moderator to elicit new and creative responses from the group participants.

Finally, the diffusion of responsibility literature, as it relates to anonymity and non-individual accountability, would seem to provide a third theoretical thread. The nature of the relationships between "safety in numbers", the pressure to conform, tendencies toward risk (Wallach, Kogan and Burt, 1967) and other related issues as they pertain to focus groups, however, has attracted little empirical interest.

Arc Focus Groups any Good?

Are focus groups any good? It would seem to me that the first step one must take to answer that question is to develop a criterion for success. This criterion can he expected to vary depending on the type of focus group one is conducting. For example, in the case of what Calder (1977) calls a phenomenological focus group, defined as one designed to acquaint researchers with the nuances and vocabulary of consumer language, one could say the criterion for success is simply getting a transcript of the focus group,) session. Since this is the type of group run for new brand managers or copy writers, I would say that a group such as this can be considered a success as long as these individuals can now identify the (potential) importance of one heretofore unrecognized word or concept which will help them in communicating to the consumer.

However, in an absolute sense, is what was learned in the focus group correct or incorrect? Oftentimes, this can be determined only when that information is actually used in communicating back to the consumer. Therefore, a determination of the ultimate success of the group frequently must await a retranslation and communication process whose fidelity, in and of itself, requires confirmation.


Should Focus Groups be Dull or Exciting?

Should a moderator strive to make groups exciting for the participants? Bernstein (1978) says no, noting that "dull focus groups may reflect reality." On the other hand, Axelrod (1975) says they should be exciting, since one has to "keep up the enthusiasm of respondents so they don't say 'why bother?'" Langer (1973) talks about animating the group. The advantage of this excitement and enthusiasm would seem to be the encouragement of top-of-mind thinking and avoidance of rationalizations.

My resolution of this controversy is to recognize, like Bernstein, that some topics are not going to lead to particularly animated and excited respondents. One has to expect that a discussion of socks is not going to elicit much in the way of exciting psychosexual content. Even of greater importance is the recognition that there is a danger of trying to keep the group (and the moderator) from becoming bored. Dull groups often are informative groups and trying to make dull groups more exciting may "force" less truthful answers.

Should Clients Observe the Groups?

The range of opinions on this question covers the entire spectrum of possibilities. Langer (1978) states that she wants clients watching since it gets "her adrenalin going" and adds "... to the intensity of the experience." Deuterman (1977), on the other hand, emphatically notes that he wants "no young brand managers watching" his groups. Following in the experimenter bias (Rosenthal, 1963; 1966) and demand effects research traditions (Orne, 1962), Kennedy (1976) worries that with clients observing there may be more of a tendency (conscious or unconscious) to manipulate the discussion to support the clients' prejudices and preconceptions.

From a methodological perspective, both Checkman (1980) and Quiriconi (1976) report modifying their moderating techniques to be sure observers understood a point which they had already recognized. 'lay (1978) notes that when those behind the mirror are "creatives", in contrast to the respondents, they tend to be "young and trendy" and, he tries to do some things to help keep these clients interested and awake (see discussion above). Relatedly, there is an increase in the pressure on the moderator to find out something new and exciting (Kennedy, 1976).

My own feeling is that there is a definite tendency for a moderator to "play to" the observers. This tends to be especially prevalent among inexperienced moderators and/or when there are others besides highly experienced observer researchers behind the one-way mirror. This situation tends to foster a carnival atmosphere to the group session making the moderator feel that to do well he has to be a graduate of the Stanislavsky School of Moderating.

However, having said all that, if I was the client, I would want to observe the groups; and, that is the crux of the problem. Recognizing that it is politically necessary in most cases to have observers, in lieu of a resolution I offer the following. Both the moderator and the observers should recognize that the bias exists. Furthermore, it must be explicitly stated that a moderator must be allowed to do his job without having to make sure everyone behind the mirror gets all of the key points. Similarly, it must be recognized that some groups will flop and/or be uninteresting and, in neither case, is the moderator necessarily at fault. This attitude should help reduce the on-stage anxiety problem.

Finally, since no one would hire a moderator who is anything less than an expert, a client should defer to the moderator's preference regarding the presence of observers. This decision implies recognizing that if the moderator chooses not to have observers, this may reflect his judgment that better findings will emerge without the increased pressure of the bright stage lights of client observation -- not that he is afraid of being watched.

How Many Respondents ShouLd There be in a Group?

Once the decision has been made to conduct a focus group, the inevitable second question (The first question is "what will it cost?") is "how many respondents should be present at the session?" Client recommendations will typically include every possible number up to the accountant model of the more the better at a given price. In practice, due to the inevitability of no-shows, the actual number of respondents often is less than the number invited; and, the moderator is put into the unenviable position of convincing the client that one can do an adequate job with less than the promised number of panelists.

In recommending the number of participants, one should recognize that there are about 80 useful minutes in a focus group session and about 25 percent of that time is usurped by the moderator. With ten participants, this leaves an average of only about six minutes per individual.

Payne (1975) recommends up to eight respondents in a group. Levy (undated) says there should be eight. Wells' (1974) recommendation brackets those of Payne and Levy suggesting that there should be six to ten.

I feel that there is no single correct size for a group with at least five issues relevant to the preferred number of respondents. There are the (1) topic, (2) type of respondents, (3) length of the agenda, (4) the number of groups in the study, and (5) type of group. With respect to the topic, the more interesting ant/or the more technical the subject matter, the fewer the respondents needed. Similarly, the more articulate and the more professional the respondents and/or the shorter the agenda, the fewer that are required. With everything else equal, the more groups which are authorized the fewer the number of respondents needed. Lastly, the purpose of the group will affect the number recruited. Following Calder's (1977) terminology, more respondents should be recruited for a phenomenological group than for a clinical group.

In the way of some absolute guidelines which I would rec = end, I tend to recruit four to eight respondents for clinical and professional groups and eight to eleven for consumer and phenomenological groups.

Should One Use a Conference Table or Living Room Layout?

Since consumers tend to have living rooms and not conference tables in their homes, Axelrod (1975) recommends using a living room arrangement to "keep the consumer in a consumer role." Quiriconi (1976) cautions against using a couch in a living room set-up since people who sit on the couch, especially the person in the middle, can't face anyone directly and seems to "sink in and get lost." Payne (1975) prefers a conference table.

I see this as probably a non-issue with consumers (homemakers) and, in practice, is simply a matter of moderator preference. With professionals and men, however, I tend to prefer a conference table format since it seems to give the respondents both the physical support and psychological distance which helps the flow of the conversation.

Should A Focus Group Report Include Respondent Quotes?

Hussey (1975) recommends completely deleting quotes. Calder (1977) suggests that they are appropriate only in phenomenological groups where one wants actual consumer language.

I disagree. The reporting of verbatims seems to me as much a question of methodology as the audience for whom the report is being written. As a practical matter, focus group reports are read by clients with all types of backgrounds; ant, the immediate goal of the research is to get the findings read. Many creative types in particular make a religion out of proclaiming that they don't understand numbers but they will read and recall quotes. Furthermore, I think that visual imagery and memory research would tend to support the notion that those points concretely illustrated and supported by quotes would tend to be recalled better. Therefore, my recommendation is, whenever possible include verbatims and make then work in the report to support analytic observations and conclusions.

Should A Moderator Provide Instant Analysis?

After a particularly thought-provoking group, few clients want to grab their coats and briefcases and dash out. Almost everyone wants to sit around and discuss the implications of what was heard. And, inevitably, the moderator is asked for his opinion " as an expert who has done many of these sessions."

Langer (1978) considers a moderator debriefing an "absolute necessity." Kennedy (1976) is totally opposed to it. I ask somewhat ambivalent and feel that the decision depends on the experience of both the moderator and the observers .

The arguments against such a debriefing include (l) biasing future analysis on the part of the moderator, (2) "hip-shooting commentary" without leaving time for reflecting on what transpired, (3) recency, selective recall and other factors associated with limited memory capabilities, and (4) not being able to hear all that was said in a less than highly involved and anxious state.

Arguments for employing instant analysis include (1) hearing the observations of several highly intelligent and informed individuals on how the groups sounded and appeared from the outside, (2) getting an initial hearing of and reaction to the moderator's top-of-mind perceptions, and (3) using the heightened awareness and excitement of the moment to generate new ideas and implications in a brain-storming type of environment.

My own preference has been to engage in "instant analysis" when clients are experienced focus group observers and are familiar with the tenuousness of "instant marketing strategy." In any case, whether instant analysis is or is not employed, the moderator should always explicitly reserve the right to change his opinion after reviewing the tapes.

I'd like to complete this section of my presentation by mentioning four additional issues related to focus groups which have not attracted a significant amount of critical comment in the marketing literature ant, for reasons of time, will only be briefly discussed here. These include (1) group composition, (2) timing of moderator interventions, (3) quantitative training for qualitative researchers and (4) moderator skills overload.

Group Composition

There has not been much discussion as to the type of personalities and/or backgrounds (homogeneous or heterogeneous) of panelists which tend to produce the best focus group results. In the case of synectic groups, an attempt is made to bring in a wide diversity of backgrounds. In the case of focus groups, the guideline I use is to recruit a relatively homogeneous group of respondents but with a sufficient diversity of experience to completely cover the range of the topic area. An unambiguous definition of homogeneity, however, is sometimes difficult to come by.

Timing of Interventions

During a focus group a moderator can be expected to make dozens of interventions, including deciding when to cut someone off, whom to call on and when to change topics. When making these interventions, the moderator is continually making intuitive judgments about the relative merits of continuing on as opposed to changing direction. Yet, there is no way of knowing whether each intervention will lead to more/better information than continuing on uninterrupted.

No simple resolution of this issue is readily apparent, but second-guessing the timing of an intervention by a skilled moderator rarely appears to be justifiable.

One further related issue can be raised. The introduction of these interventions can justly be claimed to provide the flexibility which is the strength of the focus group technique. However, it precludes application to the analysis of the extensive literature on group problem solving and the types of interaction processes which would be expected (for example, Bales and Strodtbeck, 1951) in groups without formal leadership of the type provided by a moderator.

Quantitative Training

In practice, the distinction between qualitative and quantitative research has been a sharp one. We talk about qualitative research as an art and quantitative research as a science. Moreover, practitioners of one are rarely practitioners of the other. In market research suppliers, advertising agencies and companies alike, qualitative researchers and quantitative researchers have tented to lock into their selected roles and define the research process as sequential, with quantitative research (and one group of researchers) taking over after qualitative research (and a second group of researchers) is finished.

In my opinion, this distinction is at best artifactual and at worst counterproductive. Quantitative research requires qualitative explanations; and, should subsequent quantification be required, one cannot adequately conduct qualitative research unless one has in mind the type of quantitative model and/or instrument one will later seek to employ.

Furthermore, knowledge of the procedures used in some highly technical analyses is necessary to recognize the data requirements involving qualitative definition. For example, unless a qualitative researcher is familiar with the literature on hierarchical decision making models (to date reported only in the "quantitative literature"), it will be difficult to conduct focus groups designed to reveal the types of decision making rules used by respondents.

Moderator Skills Overload

My concern here is with all of the skills which a focus group moderator is asked to master. Generally, these include question/problem/hypothesis formulation, focus group outline development, moderating, analysis, oral presentation, and a written report. it is not inconceivable that good moderators may not be good writers and/or good presenters, etc. In any case, all of these skills require some form of systematic training. I an particularly uncomfortable with the "see one, do one, teach one" model which many of us seem to have adopted for focus group moderator training.


Eighteen information requirements derived from current expectancy value attitude theory serve as the basis for ARBOR's systematic Methodology for exploratory qualitative analysis. They form a system for developing a comprehensive focus group moderator's guide from which the outlines for individual studies can be constructed consistent with their specific goals. In presenting this system, I will provide an example of one of the several specific questions used in each area to help clarify the issues being investigated within a specific information requirement.

1. Definition of significant classes of the attitude object

* "What kinds of cookware are there?"

2. Brand awareness

* "What brands of cookware are you familiar with?"

3. Evaluation of attitude objects

* "Which brand is best, worst, and why?"

4. Situational contexts/relevant others

* "How, when and where do you use cookware?"

5. Weights of situational contexts/relevant others

* "When giving cookware as a gift, what is important? "

6. Evaluation of each attitude object-in each situational context/relevant others

* "Which brands do you prefer as a gift, and why?"

7. Attributes of the attitude object for each situational context

a. Physical attributes

* "When you think about cooking with aluminum pans, what features of the cookware cone to mind?"

b. Interpersonal

* "Does anyone in your family care what type of cookware you use?"

c. Affective

* "Do you have any special feelings towards particular pots and pans?"

8. Associations among attributes

* "If a pot is heavy, will it be more or less likely to have even heat distribution?"

9. Dimensions, levels and range of attributes

* "When you say you want a heavy pot, what do you mean by 'heavy'?"

10. Threshold of satisfaction

* "How long does a pot have to last for you to consider it durable?"

11. Beliefs and opinions of brands on attributes dimensions and threshold of satisfaction

* "Are Mirro aluminum pans durable enough for you to consider buying them?"

12. Latitude of acceptance of beliefs and opinions

* "Would you believe it if I said that the Teflon TM coating on a pan will last longer than the pan itself?"

13. Evaluation of attributes (salience)

* "For which of these things which you say you want in your next set of cookware would you be willing to pay more?

14. Determination of values

* "How would you characterize someone who is a good cook?"

15. Hierarchy of values

* "Would you rather be a good cook or have a successful business outside of the hone?"

16. Saliency of relationships between attributes and values

* "You say you want a pot with even heat distribution. What does that affect, your health, your reputation as a cook, or what?"

17. Attribute salience and latitude of acceptance as related to values

* "How much do you think easy cleaning cookware can really affect your lifestyle?"

18. Category importance as related to value system

* "How much time in an average day do you spend with cookware?"


Axelrod, Myril D. (1975), "The Dynamics of the Group Interview," (ed.) H. Keith Hunt, in Advances in Consumer Research, Vol. 3, pp. 437-441.

Azrin, Nathan H., Holz, W. C., and Goldiamond, Israel (1961), "Response Bias in Questionnaire Reports," Journal of Consulting Psychology, 25, pp. 324-326.

Bales, Robert F., and Strodtbeck, Fred L. (1951), "Phases in Group Problem Solving," Journal of Abnormal and Social Psychology, 46, pp. 485-495.

Bem, Daryl J. (1967), "Self Perception: An Alternative Interpretation of Cognitive Dissonance Phenomena," Psychological Review, 74, pp. 183-200.

Bernstein, Donald (July 1978), "Focus Group Rapport Can Mislead," Advertising Age, 50.

Calder, Bobby J. (1977), "Focus Groups and the Nature of Qualitative Marketing Research," Journal of Marketing Research, 14, pp. 353-364.

Checkman, David (January 1980), "For Focus Group Success, Moderator Must Make Discussion Seem Natural," Marketing News, 16.

Deuterman, John L. (June 1977), "Approach Focus Group Use Cautiously, Meet Told," Advertising Age, 23.

Hussey, John (November 1975), "Focus Group Wrongs Include Construction, Analysis and Price," Marketing News, 8.

Kanfer, Frederick H., and Phillips, Jeanne Se (1970), Learning Foundations of Behavior Therapy, (New York: John Wiley and Sons, Inc.).

Kennedy, Frank (1976), "The Focused Group Interview and Moderator Bias," Marketing Review, 31, pp. 19-21.

Langer, Judith (September 1978), "Clients: Check Qualitative Researcher's Personal Traits to Get More; Qualitative Researchers: Enter Entire Marketing Process to Give More," Marketing News, 10.

Levy, Sid (unpublished manuscript), "Focus Group Interviewing."

May, J. P. (1978), "Qualitative Advertising Research-A Review of the Role of the Researcher," Journal of the Market Research Society, 20, pp. 203-218.

Orne, Martin T. (1962), "On the Social Psychology of the Psychological Experiment: With Particular Reference to Demand Characteristics end Their Implications," American Psychologist, 17, pp. 776-783.

Payne, Melanie S. (1975), "Preparing for Group Interviews," (ed.) R. Keith Hunt, in Advances in Consumer Research, Vol. 3, pp. 434-436.

Quiriconi, Roy S. (September 1976), "Moderators Focus on Groups: Session Yields 7 Hypotheses Covering Technology Trend, Professionalism, Training, Techniques, Reports, etc.," (ed.) Thomas E. Caruso, in Marketing News, 12.

Rosenthal, Robert (1963), "On the Social Psychology of the Psychological Experiment: The Experimenter's Hypothesis as Unintended Determinant of Experimental Results," American Scientist, (ed.) Thomas E. Caruso, 51, pp. 268-283.

Rothenthal, Robert (1966), Experimenter Effects in Behavioral Research, (Nev York: Appleton-Century-Crofts).

Wallach, Michael A., Kogan, Nathan, and Burt, R. B. (1967), "Group Risk Taking and Field Dependence-Independence of Group Members," Sociometry, 30, pp. 323-338.

Wells, William D (1974), Group Interviewing," (ed.) Robert Ferber, in Handbook of Marketing Research, (New York: McGraw-Hill), 2, pp. 133-147.

Yoell, William A. (1974), "How Useful Is Focus Group Interviewing? Not Very...Post Interviews Reveals" Marketing Review. 29. pp. 15-19.

Zajonc, Robert B. (July 1965), "Social Facilitation," Science, pp. 269-274.