Children's Consumer Research: a Call For Rigor

John R. Rossiter, University of Pennsylvania
ABSTRACT - Children's consumer research is clearly in a growth phase headed toward maturity as an important subfield of consumer behavior. However, children's consumer research will attain mature status only if greater attention is given to theoretical contributions, practical value, measurement considerations, and generalizability.
[ to cite ]:
John R. Rossiter (1979) ,"Children's Consumer Research: a Call For Rigor", in NA - Advances in Consumer Research Volume 06, eds. William L. Wilkie, Ann Abor, MI : Association for Consumer Research, Pages: 424-426.

Advances in Consumer Research Volume 6, 1979      Pages 424-426

CHILDREN'S CONSUMER RESEARCH: A CALL FOR RIGOR

John R. Rossiter, University of Pennsylvania

ABSTRACT -

Children's consumer research is clearly in a growth phase headed toward maturity as an important subfield of consumer behavior. However, children's consumer research will attain mature status only if greater attention is given to theoretical contributions, practical value, measurement considerations, and generalizability.

INTRODUCTION

Over 20% of the nation's consumers are children, and although most consumer behavior textbooks continue to neglect children's consumer behavior, more and more studies are appearing at conferences and in the journals. So it seems just a matter of time before this important subfield of consumer behavior becomes fully established. But it is just a matter of time? The point of this discussion paper is to suggest that it is not just more studies that are needed, but better studies.

The four preceding studies are part of the growth trend in children's consumer research. The studies are used to illustrate the need for better studies of a more rigorous nature. Researchers are urged to plan and execute their studies more rigorously by considering four standard criteria for good research:

1. Theoretical contributions. In order to advance our understanding of children's consumer behavior, studies should proceed from a theoretical background and attempt to contribute to this body of theory. Several theoretical approaches are possible and desirable. Major theoretical perspectives are reviewed in Robertson and Feldman (1975) and Ward, Wackman and Wartella (1977).

2. Practical value. Theoretical contributions have their own virtue. But there is a great need, as well, for children's research which has practical value. Policymakers in particular look to academic researchers to provide research findings of practical relevance to the difficult policy decisions they have to make. Marketers, too, seek practical studies but doubtless conduct their own studies to meet their needs.

3. Measurement considerations. One reason why our understanding of children's consumer behavior has not advanced as quickly as it should is due to widespread inattention to measurement considerations. Different measures with varying degrees of validity (usually no more than face validity) are used to represent supposedly identical constructs. Reliability is hardly ever examined (Rossiter, 1977) with a consequent lack of concern about the accuracy of findings. Internal rigor is thus a major problem in research to date.

4. Generalizability. If children's consumer research is to mature we have to give more attention to generalizability across studies, in the external sense. This applies not only to sample selection, but also to the sampling of content areas within studies. Family decision areas, children's products, and children's advertising should be sampled with just as much care as the children themselves.

These criteria are discussed in relation to the studies comprising this paper session.

TODDLER'S TOY SATISFACTION

Bjorklund and Bjorklund (1978) present an interesting study of toddlers' (age 1 to 22) satisfaction with number and type of toys.

Theoretical contribution. This study represents a test of Piaget's substages of sensorimotor development. The test is a fairly reasonable one and it fails to support the substage notion. In an area where almost everyone who includes age as a variable in children's consumer research believes that the ubiquitous main effect for age supports Piaget's theory, this is a refreshing outcome. It should contribute to the search for other theories to account for these types of results.

Practical value. The Bjorklunds' study has some practical value for policymakers and marketers in that it implies that you do not have to portray a lot of toys, for example in a commercial, in order to attract or satisfy a toddler. It would be interesting to extend the study to older, less egocentric children to see whether the common belief that older, more peer-oriented children are admired or envied for the number of toys they have holds true. Parents may dispute the practicality of the "satisfied with a few toys" finding, however, as they typically have to keep toddlers amused for longer than the 10-minute periods used in the study.

Measurement considerations. The use of behavioral measures of children's satisfaction shows appropriate recognition of the validity problem with prearticulate children, where verbal measures cannot be used. Even in studies with older, more articulate children there is a need for greater use of behavioral measures, especially to supplement or cross-validate verbal reports and verbally based questionnaire ratings.

Reliability is also acknowledged in this study. However, the use of coders who are familiar with the study's objectives reduces the faith that can be placed in the reliability estimate. But all too often principal investigators code their own data, a bad practice to begin with but unforgivable without inter-coder (parallel forms) reliability checks. At least an attempt was made to avoid this problem here.

Generalizability. Given the small sample size, particularly for each cell of the design, it is amazing that the ANOVA produced treatment effects which emerged through the subject variance. The study could be criticized, however, for using 12 to 20-month-old children when the population of interest comprises the 12 to 30-month-old age group. Also, the very strong upward SES bias limits generalizability and this should have been discussed; for example, in terms of I.Q. and attention span for toy play activities.

CHILDREN'S MEDIA ATTITUDES

Lindquist (1978) gratifyingly follows a suggestion made by Rossiter (1977) that children's attitudes toward television advertising should be compared with their attitudes toward advertising in other media, in this case radio, children's magazines and comic books. Lindquist's study improves on an earlier comparison by James (1971) by using a multi-item attitude measure with demonstrated reliability.

Theoretical contribution. This study provides a good example of how not to link theory with empirical data. Instead of thinking about the nature of each medium and developing some hypotheses about how children might perceive advertising in the respective media, the author offers the hypotheses after the data have been presented! Also, no attempt is made to relate the findings to those of James (1971) or Rossiter (1977) and thereby to consolidate knowledge in this area.

Practical value. Children's attitudes toward various media as advertising vehicles may be of some use to advertisers (though I doubt it based on the validity issue discussed below). On the other hand, Lindquist's study does have a certain relevance for public policy, if only to remind policymakers that television commercials are not the only source of advertising to which children are exposed.

Measurement considerations. The 7-item attitude measures used by Lindquist were either identical to (in the case of TV advertising) or directly adapted from (in the case of the other three media) the measure developed by Rossiter (1977). The measure was shown to have acceptable internal consistency reliability and test-retest reliability, the two appropriate types of reliability in this instance, for attitudes toward TV advertising measured on a sample of 4th through 6th grade children. Unfortunately, Lindquist did not have time to compute comparable reliability estimates for the present data. I, for one, anxiously await these computations. Others should too, for this is the only self-administered attitude measure in the children's consumer research literature for which reliability estimates are available. The sooner we begin to use standardized measures of known reliability, the better our results are going to get, simply because of reduced error variance.

The crucial issue, however, concerns the validity of children's attitude measures. While there is little doubt that the attitude measure used in this study taps the construct of attitude adequately (construct validity) and samples the traditional cognitive, affective and motivational components of attitude (content validity), there is considerable doubt as to its predictive validity. There is no evidence that increasingly negative attitudes toward television commercials as children grow older bear any relationship to actual behavior, such as requests for advertised products (Rossiter, in press). It is unlikely that the picture would be any different for other media. Children simply seem to learn more cynical verbally expressed attitudes but these attitudes do not seem to affect their behavior. Advertisers, therefore, would be ill advised to place stock in these types of findings.

Generalizability. As noted, it would have been interesting to see to what extent Rossiter's (1977) results generalized to a somewhat broader sample (3rd through 6th grade versus 4th through 6th, and a wider SES spread) and to other media. Also, given the Federal Trade Commission's focus on age 7 as a cutoff for "total" protection from TV advertising--a focus based, incidentally, largely on attitude data--extension of the study to 1st and 2nd grades would be useful. This can be accomplished by using a reader to assist younger children, as Lindquist did at 3rd grade, or by administering the test as a series of open-ended questions and using multiple coders, as recommended in Rossiter (1977).

FAMILY DECISION MAKING

The studies by Jenkins (1978) and Nelson (1978) examine the role of children in family decision making. As such, they can be considered together.

Theoretical contribution. Both studies are essentially atheoretical. This is a pity because a little forethought as to what children's "influence" (Jenkins) or "involvement" (Nelson) in family decision making entails might have prevented the largely uninterpretable findings that resulted in both instances. Surely any study of children's roles in family decision making must acknowledge the difference between what Haley, Overhol-ser and Associates (1975) termed direct influence and indirect influence. The former represents an active role based directly on the decision maker's own needs whereas the latter represents a passive role in which the decision maker takes another family member's needs indirectly into account. This distinction was developed for husband-wife dyads, but it would seem to be crucial when children are included in the decision process (cf. Well's 1965 concept of "passive dictation"). In short, neither study contributes theoretically to our understanding of the role (or roles) of children in family decision making.

Practical value. As straight empirical studies Jenkins' and Nelson's efforts may have had some practical value to marketers were it not for definitional ambiguities, particularly in Jenkins' case, and sampling problems, particularly in Nelson's case. These problems are discussed below.

We should not, however, lose sight of the importance of more research on family decision making in the true sense--that is, including children, when appropriate. As Jenkins and Nelson note, far too many studies of family decision making focus exclusively on adults.

Measurement considerations. It is impossible to obtain valid measures of constructs such as "influence" or "involvement" unless these constructs are clearly defined at the outset. Jenkins freely admits in his pretest that interpretation of the word "influence" varied from one person to another, yet the study proceeded with this fatal ambiguity unresolved. Nelson apparently decided to avoid this complication for his concept of "involvement" by not pretesting at all.

Unreliability is likely to have entered the measures (quite apart from the basic validity problem) in a number of ways. The data in both studies rely on parental reports; and while it has been shown that, on an aggregate basis, husbands can fairly reliably report for wives and vice versa, this has by no means been demonstrated for parents reporting for children (Rossiter and Robertson, 1975). Other problems include the use of single item measures and reliance on recall, the latter severe in Jenkins' study but not so much in Nelson's, although Nelson is at fault for trying to use monadic measures rather than ratio measures, essential for the notion of joint decisions. Finally, both studies introduce "noise" by including teenagers as potential role-takers when they purport to be studies of children.

Generalizability. Earlier in this discussion it was remarked that content generalizability is just as important as sample generalizability- Each study provides an illustration.

Jenkins selected eight content areas to represent family decision making: major appliances, automobiles, furniture, groceries, savings, life insurance, vacation decisions and a miscellaneous category. How and why were these selected? Only one content area (vacations) turned out to have substantial children's influence. This is not too surprising a result unless one were to have distinguished direct influence from indirect influence. It is hard to imagine parents deciding on automobiles, furniture, or life insurance, for example, without taking children indirectly into account. Neglected are content areas where children a priori are known to have a greater role in purchase decisions (Ward and Wackman, 1972). Without adequate sampling of content we cannot hope to make meaningful generalizations about family decision making.

Nelson's study contains the more traditionally recognized sampling problem: subject selection. Participation in his study was confined not only to people who frequently ate pizza out but also to those who reported eating their last pizza out as a nuclear family. Let me suggest that there may be a conspiracy among children to veto pizza dining in favor of McDonald's. The sampling problem should be obvious.

CONCLUSIONS

Children's consumer research has reached a stage of development in which increased rigor in planning and execution is of paramount importance. In the rush to fill the knowledge void concerning children's consumer behavior we have all been guilty of lack of rigor in particular areas. Four such areas are suggested for increased attention in future research:

1. Theoretical contributions. Lack of theory usually stems from lack of thought. Studies should be planned as contributions to an overall body of knowledge rather than as singular empirical ventures.

2. Practical value. At the present time the overriding need is for studies that have practical relevance for policymakers. The alternative is policies based largely on speculation accompanied by the allegation that we, the experts, have failed to deliver.

3. Measurement considerations. Validity and reliability are the hallmarks of sound research and accepted principles of scientific measurement cannot continue to be neglected, either in adult consumer research (Jacoby, 1978) or in children's.

4. Generalizability. Generalizability is the crux of science (see, for example, Nunnally, 1978). Sampling of content as well as sampling of subjects must be given greater attention in the interest of generalizable research results.

If children's consumer research should pass through a fad cycle, as so many of our research interests do, let us be sure that we leave a substantial and sophisticated contribution. Only by constructive efforts along the lines suggested above, at the inevitable price of a little destructive activity among colleagues, can we achieve this.

REFERENCES

G. Bjorklund and R. Bjorklund, "An Exploratory Study of Toddlers' Satisfaction With Their Toy Environments," paper presented at the Ninth Annual Conference of the Association for Consumer Research (Miami, Florida, 1978).

R. I. Haley, C. E. Overholser and Associates, Purchase Influence: Measures of Husband/Wife Influence on Buying Decisions. Summary Report (New Canaan, Connecticut: Haley, Overholser and Associates, 1975).

J. Jacoby, "Consumer Research: A State of the Art Review," Journal of Marketing, 42 (April 1978), 87-96.

D. L. James, Youth, Media and Advertising (Austin, Texas: Bureau of Business Research, University of Texas at Austin, 1971).

R. L. Jenkins, "The Influence of Children in Family Decision Making: Parents' Perceptions," paper presented at the Ninth Annual Conference of the Association for Consumer Research (Miami, Florida, 1978).

J. D. Lindquist, "Children's Attitudes Toward Advertising on Television and Radio and in Children's Magazines and Comic Books," paper presented at the Ninth Annual Conference of the Association for Consumer Research (Miami, Florida, 1978).

J. E. Nelson, "Children as Information Sources in the Family Decision to Eat Out," paper presented at the Ninth Annual Conference of the Association for Consumer Research (Miami, Florida, 1978).

J.C. Nunnally, Psychometric Theory (New York, New York: McGraw-Hill, 2nd edition, 1978).

T. S. Robertson and S. Feldman, "Children as Consumers: The Need for Multitheoretical Perspectives," in B. B. Anderson (Ed.), Advances in Consumer Research: Volume III (Cincinnati, Ohio: Association for Consumer Research, 1975), 508-512.

J. R. Rossiter, "Reliability of a Short Test Measuring Children's Attitudes Toward TV Commercials," Journal of Consumer Research, 3 (March 1977), 179-184.

J. R. Rossiter, "Does TV Advertising Affect Children?," Journal of Advertising Research, (in press).

J. R. Rossiter and T. S. Robertson, "Children's Television Viewing: An Examination of Parent-Child Consensus," Sociometry, 39 (September 1975), 308-326..

G. J. Szybillo and A. Sosanie, "Family Decision Making: Husband, Wife and Children," in W. D. Perrault, Jr. (Ed.), Advances in Consumer Research: Volume IV (Atlanta, Georgia: Association for Consumer Research, 1977), 46-49.

S. Ward and D. B. Wackman, "Children's Purchase Influence Attempts and Parental Yielding," Journal of Marketing Research, 9 (August 1972), 316-319.

S. Ward, D. B. Wackman and E. Wartella, How Children Learn to Buy (Beverly Hills, California: Sage: 1977).

W. D. Wells, "Communicating with Children," Journal of Advertising Research, 5 (June 1965), 2-14.

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