Special Session Summary Consumer Research With Special Populations: Issues, Problems, and Solutions

Citation:

Kim Corfman and Debbie Roedder John (1998) ,"Special Session Summary Consumer Research With Special Populations: Issues, Problems, and Solutions", in AP - Asia Pacific Advances in Consumer Research Volume 3, eds. Kineta Hung and Kent B. Monroe, Provo, UT : Association for Consumer Research, Pages: 92-93.

Asia Pacific Advances in Consumer Research Volume 3, 1998      Pages 92-93

SPECIAL SESSION SUMMARY

CONSUMER RESEARCH WITH SPECIAL POPULATIONS: ISSUES, PROBLEMS, AND SOLUTIONS

Kim Corfman, New York University, U.S.A.

Debbie Roedder John, University of Minnesota, U.S.A.

Some of the most enduring issues in consumer behavior pertain to special populations. Special populations are defined here as those segments that lie outside the traditional focus for consumer research, namely individual adults aged 18-49 years. Children, elderly consumers, couples, organizations, and ethnic groups are examples of special populations.

Concerns about these special populationsBin terms of how to reach them, understand them, and influence their behaviorBhave become even more important of late. Firms can no longer focus all their attention on the mainstream mass market as our society becomes more fragmented and competition for new customers escalates. This is true for business customers as well as individual consumers.

One of the biggest problems in addressing these types of concerns about special populations is the lack of information about how to do research with these populations. Researchers are ordinarily trained quite well in survey research and experimental design for the typical 18-49 year-old subject. But, they are not exposed to issues and techniques related to how to study other types of subjects, whether it’s a 5 year-old child in an advertising experiment, a husband-wife duo in a family decision-making study, or organizational buyers in a field survey. Far too often, researchers have found, to their dismay, that they had no idea of the types of problems they would confront in working with these subjects and, even worse, that little help was available to solve whatever problems they confronted.

The purpose of this session was to bring together several experts, each with extensive experience in researching a special population, to discuss the types of problems and possible solutions that face someone wishing to work with a non-traditional subject population. In doing so, we focused our session on three different special populations and experts: children (Debbie Roedder John), groups (Kim Corfman), and organizational buyers (George John).

 

THE ABC’S OF DOING RESEARCH WITH CHILDREN

Deborah Roedder John, University of Minnesota

The purpose of this presentation was to outline some of the major issues and problems that arise when we work with children as research subjects: (1) recruiting samples, including gaining access to schools and huma subject requirements; (2) designing experimental procedures and manipulations, including problems with scheduling data collection and with self-administered surveys; (3) measurement problems, including concerns about open-ended questions requiring verbal responses and about rating scale problems; and (4) data analysis, focusing on problems in aggregating data across age groups.

Recruiting Subjects. Researchers face significant problems in obtaining subjects from public schools. There are typically layers of administrators and district officials that must sign off on research projects conducted in the schools, often requiring 6-9 months of lead time to secure a go-ahead to collect data. One solution, which I have used in most of my research projects, is to recruit subjects from private Catholic schools, which require approval from a minimum number of administrators (typically, the principal and sometimes teachers).

Designing Experimental Procedures. One of the biggest challenges in designing experiments is designing stimuli that are equally comprehended, equally meaningful, and equally liked by children from different age groups. One can, of course, pretest until the right stimulus comes along. A quicker solution, which I have used in most of my studies of children’s decision making, is to ask salespeople or managers in small toy and craft stores for their opinions. They often have good ideas about what appeals to different genders and different age groups. Teachers or day care personnel are good sources of information regarding what children of different ages might be able to understand or comprehend.

Measurement. There are a variety of concerns pertaining to measurement. One, related to the use of rating scales, is whether the responses of children as young as preschool and kindergarten reflect their actual feelings and knowledge of using the rating scale. One diagnostic measure, which I have used extensively with younger children, is to ask them to practice using the rating scale first working with stimuli that have convenient benchmarks. For example, if you are using a traditional 5-point smiley-face scale, you can ask children to use the scale to indicate how the feel about universally-loved objects (e.g., M&M’s or ice cream) and universally-hated items (e.g., icky medicine, liver).

Data Analysis. One of the most perplexing problems in comparing responses across age groups is assessing whether children of different ages have used a rating scale in approximately the same way. What often occurs is that young children tend to use the extreme points of the rating scale and not much in between, whereas older children do use more middle scale points. One solution is to standardize scale responses by individual or by age group.

 

ALL TOGETHER NOW... THE CHALLENGES OF CONDUCTING RESEARCH ON GROUPS

Kim Corfman, New York University

Wives, husbands, and purchasing groups are not usually considered "special" populations in the sense that, as individuals, they pose unique problems requiring creative approaches to recruiting, measurement, and analysis. However, once you put them together into groups, they become very "special" indeed.

Recruiting Subjects. One of the simplest approaches to conducting research on groups is to create ad hoc groups in the laboratory and run experiments on them. While there are aspects of this methodology that make it difficult, this approach has the advantage that subjects can be recruited from anywhere, and mixed and matched at the experimenter’s convenience. While this is an excellent place to begin the process of isolating group phenomena, it is extremely limited in its generalizability to the groups of greatest interest for consumer behaviorBe.g., families, buying centers, committees, clubsBwhich have many characteristics that cannot be created or approximated in the laboratory.

When we move to the realm of "real" groups, the recruiting task increases n complexity. In some cases, it is even a challenge to identify all relevant group members. When the method is experimental, the researcher has no choice but to go to the family or buying center, for example, or try to get them to come to the research site, which requires that the incentive to participate be substantial.

Methodological Logistics. Scheduling multiple subjects to appear in the same place at the same time, and coping with "no shows" (who make it impossible to use subjects who did show up), are large administrative headaches. It is even harder when it matters which subjects are in which groups (as with "real" groups) and when one wishes to engineer the participant mix. Further, because group interactions involve discussion, it is often necessary to run only a single group (i.e., observation) at a time. Laboratories with multiple break-out rooms increase the number of groups that can be handled simultaneously, but it remains small.

Measurement. Even when a survey methodology can be employed, it has been amply demonstrated that taking the far easier route of asking one member of a group to report on other members’ preferences, behavior, attitudes, influence, etc. is subject to substantial bias. One alternative is to survey each member, multiplying the data collection task by the number of members in each group. For some kinds of investigations, it has been shown that group members do a very poor job of reporting even on their own role in the group process.

Even when the group is "real," with an experimental method the researcher must still tackle the problem of making the task realistic. Approximating naturally occurring tasks and eliciting natural responses from groups under observation are goals that are rarely, if ever, achieved.

Analysis. Another problem associated with running choice and judgment experiments with groups is that each decision takes more time than a comparable decision made by an individual. Thus, fewer decisions can be made by a group before fatigue or time constraints kick in, making it hard to gather enough data from each group to do group-level analysis.

 

CONDUCTING RESEARCH IN ORGANIZATIONAL BUYING PRACTICES

George John, University of Minnesota

Interest in commercial buyer-supplier relationships, purchasing alliances, and the like has intensified among academics and practitioners. This presentation explored issues and problems with existing methodological approaches in the field.

Subject Recruitment. Intact firms and dyads are commonly employed for tests of theories in this arena. Although scarce, the available data are humbling with respect to the lack of convergent validity of constructs specified at the firm and dyadic levels. This dilemma is often brushed aside or excused with an "exploratory stage" comment. I scrutinize this issue, and proffer advice on theory building and testing. Some retrenchment of complex models and a close scrutiny of the level(s) of analysis implicated by the theory at hand may be in order.

Data Collection. The research techniques commonly utilized by researchers and market research firms cluster around (a) survey questionnaires filled out by "key informants," and (b) case studies. Typically, a relatively small number of business customers account for a significant fraction of a supplier’s business. Furthermore, these ties between suppliers and their "important" buyers (lead users, etc.) are close, and face-to-face in nature. In these circumstances, both survey questionnaires and face-to-face interviews are problematic since responses are clouded by strategic attempts by buyers to influence the supplier (and vice versa). A significant buyer knows that a poor response (e.g., on a satisfaction item on a questionnaire) will evoke supplier action. Common practices such as sponsorship of research by industry associations to gain access to "subject pools" only exacerbates this problem.

I offer advice on designing surveys and case studies to alleviate this problem. For instance, surveys by third parties may be preferred over direct contact. Likewise, the key informant method may be over-used to the detriment of respondent methods. Cross-validation using experimental methods is also indicated.

With respect to case studies, they are normally relegated to "exploratory" stages, or else apologetically written up as pre-formal investigations. Campbell’s (1975) little-cited paper on using case studies for formal theory-testing is developed into a framework for making appropriate design choices in these settings.

Data Analysis. The most vexing problem is that of analyzing data dealing with multi-level models. For instance, a multi-level theory about team productivity would hold variance within teams as more than just noise. Most published work relies on single-level data analysis, which ignores all but one of the multiple levels. Recently developed approaches using random effects "hierarchical" models are explored for their utility in addressing such data.

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