Discussion of &Quot;Back to the Basics&Quot; Research Papers

Joseph W. Newman, University of Arizona
[ to cite ]:
Joseph W. Newman (1979) ,"Discussion of &Quot;Back to the Basics&Quot; Research Papers", in NA - Advances in Consumer Research Volume 06, eds. William L. Wilkie, Ann Abor, MI : Association for Consumer Research, Pages: 171-173.

Advances in Consumer Research Volume 6, 1979      Pages 171-173

DISCUSSION OF "BACK TO THE BASICS" RESEARCH PAPERS

Joseph W. Newman, University of Arizona

At the outset, I should like to comment that all three papers represent conscientious efforts to do careful, thorough work and the two empirical studies attempted to employ improved methodology. This is important, of course, because such efforts are essential to what we hope is continuing progress in understanding consumer behavior.

THE KENDALL-FENWICK PAPER

I shall start with the paper by Kendall and Fenwick and deal with it more extensively than with the others because it is more closely related to work I have done in the past. The study is interesting for several reasons. It is one of the few efforts to observe what real consumers do in the market place vs. the laboratory. It also is one of the few attempts to compare findings obtained by observations vs. self-reports. The findings are provocative and help highlight a need for more comprehensive research.

The extent to which the authors really found out much about the prevalence of label reading depends, of course, on the validity of their measure of the dependent variable. They made the key assumption that their clockings of the time shoppers spent examining the product alternatives were a function of the extent of reading, processing, and use of label information. That may not be the case. Viewing might represent no information acquisition at all and even if acquired, information may not be used. Among factors which might make for different viewing times among shoppers are skill in label reading which probably is related to experience and education; whether the shopping is done leisurely or in haste; the decisiveness of the shopper; how easy or difficult it is to comprehend information from the label format; and the extent to which there are environmental distractions such as those provided by young children parents may have in tow while shopping.

While such considerations raise questions about the dependent variable, they do not mean that nothing useful can come out of observing behavior. On the contrary, it is important to observe what is happening in real life. No direct measure of reading, processing or use of information is possible from in-store observations alone. Instead one must settle for what hopefully is a useful proxy and the authors were thoughtful in their consideration and use of the time variable.

Perhaps the study's most important contribution is that it provides evidence that measures of search behavior based on direct observation and measures based on self-reports of the same behavior do not agree. Only a few researchers have done this (Jacoby et al., 1976; Newman and Lockeman, 1975). At least two other studies (Olshavsky, 1973; Newman and Lockeman, 1975) found considerably more apparent information seeking activity by using direct observation than one would expect from reported survey research findings. Yet most of what is accepted today as knowledge of amount of consumer information seeking is based on survey data.

The authors stated that the "grabbers" in their self-reports tended to claim more label reading than did the "lookers." It would have been helpful had more of the self-report findings been presented in the paper.

The authors observed a large number of shoppers but there is the usual question about the extent to which the findings can be generalized. No descriptive information was given of the stores' clienteles and the products involved in this study may not be representative of most food products. If the stores' customers were upscale on education and income, the reported observation of considerable label reading would not be as surprising as the lesser amount of reading indicated by the self-reports.

As I am sure the authors are aware, there is danger in generalizing from their limited data on canned meat/fish vs. that on dehydrated soup as to the amount of label reading for established vs. new products because of differences in characteristics of the products and possible differences in their importance to the consumer.

As for the question of who looks at labels, the six explanatory variables used in the first analyses of the data were very limited in nature and really did not add much to understanding, as the authors recognized. Actually, it does not appear that the six variables discriminated better than could be expected by chance. The authors reported that use of the variables resulted in correctly classifying as either "grabbers" or "lookers" 75 per cent of the shoppers. But the ratio of "grabbers" to "lookers" in the sample was one to three. And the test was made using the original data rather than a hold-out sample. Limitations of the variables also are indicated by the finding that the discriminant function for the first six variables showed some sign changes when applied to the Aisle 2 vs. the Aisle 1 products. The other three variables (satisfaction with current label information, extent of support for mandatory federal label regulations, and degree of menu planning) seemed to be more relevant.

An issue is involved in aggregating data. The procedure the authors used assumes that in their evaluations all subjects were trying to do the same thing in the same way. It assumes use of an additive model without interaction, and common subjective utilities for label information. If these assumptions are not valid, aggregating the data can lead to erroneous results. In the study by Best and McCullough (1977) to which the authors referred, subjects' responses to labels first were viewed individually. This led to the identification of three different groups of respondents in terms of subjective utilities for label formats.

In summary, the Kendall-Fenwick study, along with a few others, suggests that we may know less than we have assumed about the amount and character of label reading or, for that matter, consumer information search and use in general. Different methods have yielded different findings. Kendall and Fenwick suggested that for public policy purposes any studies based on self-reported label reading should be treated with caution. The same warning, however, is appropriate for studies employing direct observation. This should not lead us to abandon advocacy of empirical research as an important input to public policy decision making. But right now we must recognize a need for better measures and more complete information on label reading and other information seeking activities. Better data would seem to require more complete data gathering approaches than have been employed thus far.

For data on label reading to be meaningful in terms of consumer welfare, we need to know what the consumer tried to find out and whether the effort succeeded and, if it did, whether it was with ease or difficulty. In this connection, there is a need to distinguish among consumers on the basis of knowledge of the product and brands and other characteristics that make for differences in interest in and ability to acquire and use information. Kendall and Fenwick suggested that it is not as much the type of person but the situational variables which determine label reading. Their data give little support to that conclusion, however, because of limitations of their explanatory variables. Some of the variables may well be proxies for unidentified consumer characteristics we need to know more about. It is quite reasonable to believe that both consumer characteristics and situational variables may be important.

The suggestion that different information formats may be appropriate for "grabbers" and "lookers" is not very practical in that both shop for the same merchandise in the same stores. More important, the suggestion implies thinking in terms of a fixed dichotomy rather than recognizing that today's "grabbers" may be yesterday's "lookers" who have learned about product and brand alternatives and how to get what they want in the supermarket with a minimum expenditure of time.

The assumption often has been made that more search, more reading and more deliberation mean better informed consumers and, consequently, more satisfactory purchases. Evidence in support of this assumption, however, has been lacking because few, if any, studies have attempted to cover that much ground.

THE MAUSER PAPER

The Mauser paper reflects serious concern with employing appropriate methodology in a taste test and care in controlling for such things as order effect and sample presentation. Even so, I must confess to having some difficulty with the procedure employed and with accepting some of the conclusions.

The study is not exactly "Allison and Uhl Revisited" as its title states. There are several differences in addition to methodology. Allison and Uhl confined their test to beer, using six brands, apparently all lager beers. Mauser included three categories of malt beverage: ale, lager beer and light lager beer. This being the case, one would expect that he would find more evidence of consumer ability to tell differences. Allison and Uhl tested to learn whether subjects would rate the brand they said they drank most often as superior in both labeled and unlabeled test conditions. Mauser did not attempt to do that. Another difference worth noting is that Allison and Uhl left a six-pack containing bottles of three unidentified brands with their subjects for one week, then left another six-pack of bottles with labels for another week before picking up rating forms. Mauser used a one-time test of one-ounce samples. The effect of such differences in exposure is a subject for future research.

It would have been helpful had Mauser described more fully the characteristics of the brands and his rationale for brand selection and presented hypotheses or statements as to what he expected to find. If beer consumers can discriminate on the basis of taste and aroma, one might expect that they would perceive brands in each of the three categories (ale, lager and light lager) as being similar. Yet inspection of his Figure 1 reveals that this did not clearly happen. The two ales, for example, were spaced far apart in both the labeled and unlabeled configurations. And Schlitz appeared removed from Old Style and Blue even though all three presumably are lager beers (although the paper did not clearly identify brands by type).

Inspection of Figure 1 as it appeared in his paper also makes it difficult to accept the conclusion of reasonably tight clusters of brands. Instead, one is left wondering what is going on here. The author bases his conclusion that beer drinkers can distinguish among brands on a space diagram which does not seem to make sense even though the stress statistic of .126 for the unlabeled condition is statistically significant at the .05 level for nine stimuli and two dimensions (Klahr, 1969). The stress statistic of .185 for the labeled condition, however, has a .5 probability of being obtained by chance.

Observations such as I have made cultivate a distrust of the data and a suspicion of an unidentified source of error.

A test such as this one might be more informative if an effort were made to distinguish subjects who can discriminate reasonably well on the basis of taste from those who can not. This could be done by means of a triangle test. It would be interesting to know the proportion of discriminators. A taste test such as Mauser conducted then could use only discriminators as subjects which should eliminate some of the noise which may be clouding the picture presented in Mauser's Figure i and the results should be a better guide for product planning.

It appears from Figure 1 that knowing the brand made a difference in how the brand was "tasted" in four of the eight cases (Schlitz, Olympia, 50 Ale and Export Ale). In this connection, it is interesting to note that Allison and Uhl found that the gains in brand ratings which resulted when brand was identified were not uniform among the brands tested. Neither study attempted an explanation. Examination of advertising budgets and messages might provide some insight here.

The author's conclusion that familiarity with the product class did not significantly affect taste judgments is open to challenge because familiarity was not measured. Subjects were categorized as having either high or low familiarity depending on whether they drank six or more bottles of beer a week. So the measure really was of consumption which does not necessarily mean familiarity with the various brands in the product class. A heavy, brand loyal beer drinker may be less familiar with alternative brands than a more occasional beer drinker who engages in considerable brand switching.

As for the finding of a label effect in four taste characteristics, a factor analysis of all eight characteristics used in the ratings might reveal that only one, or perhaps two, dimensions were represented.

In summary, the Mauser findings leave me puzzled and unclear about what contribution the paper makes.

THE ARCH PAPER

The paper by Arch is an excellent review of the whole area of pupil dilation. It cites the relevant literature and goes on to succinctly identify potential benefits and limitations of the method, related design problems, guidelines for use and potential applications.

It is clear that much remains to be learned about what pupil dilation actually measures, the relative importance of the problems associated with its use and the sensitivity of dilation to the possible distorting influences. Validity, of course, is the key question here as it was with observations of viewing time in the Kendall-Fenwick study of label reading. Both papers help highlight a need for more research which employs more than one method at a time in order that more may be learned about validity.

Thorough reviews of other research methods like the one Arch has done on pupil dilation would be welcome contributions to our consumer research literature.

REFERENCES

Ralph I. Allison and Kenneth P. Uhl, "Influence of Beer Brand Identification on Taste Perception," Journal of Marketing Research, 1 (August 1964), 36-39.

Roger Best and James McCullough, "Evaluation of Food Labeling Policies through Measurement of Consumer utility," in Keith Hunt (Ed.), Advances in Consumer Research, 1978, 5, 213-219.

D. Klahr, "A Monte Carlo Investigation of the Statistical Significance of Kruskal's Nonmetric Scaling Procedure," Psychometrika, 34, 319-20.

Jacob Jacoby, Robert W. Chestnut, Karl C. Weigl and William Fisher, "Prepurchase Information Acquisition: Description of a Process Methodology, Research Paradigm and Pilot Investigation," in B. B. Anderson (Ed.), Advances in Consumer Research, 1976, 3, 306-314.

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

Joseph W. Newman and Bradley D. Lockeman, "Measuring Prepurchase Information Seeking," Journal of Consumer Research, December 1975, 2, 216-222.

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