Back to the Basics: Discussion

Citation:

James F. Engel (1979) ,"Back to the Basics: Discussion", in NA - Advances in Consumer Research Volume 06, eds. William L. Wilkie, Ann Abor, MI : Association for Consumer Research, Pages: 169-170.

Advances in Consumer Research Volume 6, 1979      Pages 169-170

BACK TO THE BASICS: DISCUSSION

James F. Engel, Wheaton Graduate School (Illinois)

INTRODUCTION

In a sense, the discussant in this session has a difficult assignment. The title, Back to the Basics, probably is an accurate one, but it is not altogether clear that this was a deliberate intent of those who presented papers. Rather, it reflects more on the difficulty faced by the program chairman in finding themes to use in organizing a multitude of sessions. Therefore, the common thread between the papers is not especially evident.

Perhaps one can stretch a bit and isolate these "basics" in the papers which were presented: (1) methodological validation (Arch and Kendall and Fenwick) and (2) replication (Mauser). In a limited sense, each has shed some light on consumer behavior research in these respects.

PDR MEASURES (ARCH)

About 1964, Herb Krugman, then with Marplan, burst on the scene with the latest great breakthrough--pupil dilation research. Glowing reports on this daring innovation soon were published by Krugman (1964 and 1965). The discussant also was with Marplan during that period and had opportunity to delve deeply into this methodology and found little in the way of conclusive validation. Later research has shown pretty clearly that the vast investment required for such equipment does little or nothing to augment present advertising research methodology (Goldwater, 1972 and Rich, 1974). In fact, it can lead to some misleading conclusions. Rice reports the example that designers of a frozen french fries ad were delighted with PDR test results until further analysis showed that people were looking at the steak in the picture rather than french fries (Rice, 1974).

David Arch has given us a competent review on this subject, and his paper is helpful for that reason alone. He has not, however, gone much beyond the contributions of Blackwell, Hensel, and Sternthal (1970). It was known at that time that PDR apparently measures load processing and that it says little about the importance of stimuli to the individuals in question. The pioneering research undertaken at The Ohio State University gave pretty strong indications that GSR (galvanic skin response) measures probably should be used in tandem with PDR on the likelihood that GSR measures arousal and importance. Arch unfortunately has not alluded directly to this research, because it is one way out of the dilemma that he has isolated.

The greatest contribution of this paper is that it clearly spells out the difficulties in using PDR methodology which must, by necessity, be confined to highly artificial laboratory settings. These considerations should be a red light to those who wax too enthusiastically over this type of physiological measure.

This reviewer is most puzzled by the author's conclusion that PDR measures have potential value in consumer research. As justification he cites three examples which are, without exception, confined to academic research. What practical applications exist, if any? It was Krugman's overly-enthusiastic claims on this dimension that ignited a controversy during the 1960s. PDR has been proposed and apparently used as a part of pretesting methodology in advertising. Is this valid or not? Arch must address such a pragmatic consideration; otherwise his paper is not much more than a competent review of what others have already said. It would seem from his review that PDR is pretty much of a dead end in all respects. If this is what he wanted to say, then it should be stated loud and clear in a pragmatic context of practical applications.

THE SHOPPER AND THE LABEL (KENDALL AND FENWICK)

Kendall and Fenwick have placed a heavy burden on the reader or listener as they wade through what seems to be an excessively difficult-to-decipher manuscript. But the effort is worth it for one significant reason--some really valuable new light is shed on the validity of self-reports in information processing.

These researchers wisely suspected that laboratory or self-report survey data may obscure the real situation with respect to usage of product labels by supermarket shoppers. Therefore, they did the only reasonable thing by designing a field study centering mostly on observation of actual behavior. In a very real sense, they have provided a validity check on the other methods and found them to come up severely wanting. Their data would have been even more conclusive had they compared actual observation and self-reports with the same people studied in their project. This probably is the next step to be taken by those interested in further research.

The other data reported are only of incidental interest. For example, it would have made much more sense to have probed immediately after observation to try and uncover which information was used. The label ranking experiment was artificial and of doubtful validity. Moreover, the attempt to classify shoppers by discriminant analysis made use of correlated variables and hence is questionable. For example, the type of shopper and the time of day of shopping are obviously interrelated.

Putting these considerations aside, Kendall and Fenwick have helped us by their careful attempt to remedy methodological deficiencies. In so doing, it appears that conventional wisdom about label use is defective. It is by such efforts as this that progress is made in arriving at defensible principles and propositions.

ALLISON & UHL REVISITED (MAUSER)

Any field within the behavioral sciences progresses toward maturity by developing a tradition of rigorous replication. Blackwell, Kollat, and the discussant have long called for replication in their books and at previous ACR Conferences. It is encouraging to see in very recent years that replication is underway, and this paper should be viewed in that context.

There is no question that Mauser improved on Allison and Uhl's design by making use of paired comparisons. Allison and Uhl, however, did have an advantage of allowing people to sample various brands of beer, both labeled and unlabeled, in a more natural setting. That fact in itself, of course, removed some of the opportunities for control that Mauser instituted. His bottles consumed per week is arbitrary and certainly is no improvement on Allison and Uhl's criterion (beer consumed at least three times per week). Other than that, the controls on the actual experimental conditions, the analysis of order bias, and the use of perceived similarity measures are genuine improvements.

There is one place in which Mauser appears to be measuring something different from Allison and Uhl who rated beers along various attributes in terms of "too much," "just enough," and "not enough." Mauser, on the other hand, merely focused on whether one of the samples in each pair was judged as having more of the particular characteristic. While he was following standard paired comparison procedures, his data are not really comparable to Allison and Uhl's, thus making direct comparison difficult.

Contrary to Mauser's own statement, it can hardly be said that his findings are in "sharp contrast" to those of Allison and Uhl. Drinkers were able to distinguish between the eight brands on only four of the eight attributes in both experimental conditions. As he indicated, this is quite likely due to the wider range of tastes in the beers used. If that is the case, then how is this a replication? Furthermore, the perceived similarity ratings were only roughly comparable between the two experimental conditions. The author did not do justice to this latter analysis, because one cannot help but wondering whether other factors also were shaping these perceptual judgments. In short, the case has not been proved definitively.

But suppose there had been really major differences between the two experiments? Many will ask, so what? Was the author merely replicating an interesting experiment, or was he testing a proposition of consumer behavior? Nothing is said about the implications in terms of consumer information processing. As it stands, one could argue that a cannon was used to kill a flea. Allison and Uhl were saying something important in terms of a conceptual framework, and that seems to have become lost here. Replication is of value only when these conceptual issues are given primary focus. Because this focus is absent here, at least insofar as the write-up, an otherwise useful paper probably is destined to fall into the "interesting but so what?" category.

IMPLICATIONS

Of these three papers, Kendall and Fenwick made the greatest contribution through their clear-cut challenge to existing methodology. As was mentioned, their case would have been even stronger had there been direct methodological comparison, but no matter. They have issued a real challenge both to existing methodology and to the conclusions which have appeared in the literature. The door which has been opened should be used by others, keeping in mind that we always should strive toward "real world" behavioral conditions while endeavoring to maintain experimental control. Kendall and Fenwick showed that this is by no means impossible.

That paper by David Arch is most notable for its careful literature review, although he did neglect some important substantive considerations as noted. Nevertheless, he has demonstrated the value of assessing findings from fields beyond the borders of consumer research. Such scholarly cautions could lead us to avoid the promotional overkill which always seems to accompany a tool such as PDR. It is worth mentioning once again that the inflated claims during the 1960s never had any real substance in fact. Yet those without methodological sophistication piled on the bandwagon. Now we find that Krugman is at it again, this time claiming that analysis of brainwaves is the great new frontier again (Krugman, 1978). While he could be correct, we are in danger of making the same mistake made in the 1960s unless some of us react with healthy skepticism and put on the caution light as Arch has done with PDR. The questions we should ask are these: (1) "what has been learned elsewhere?" and (2) "how does technique represent any improvement over existing methodology?" This requires both extensive literature review and validation research.

Finally, replication is a necessity, and Mauser should be commended in this sense. But let's get some sense of priority and avoid replication just for the sake of replication. The emphasis should be on replication to verify or disprove important propositions. This, in turn, demands a grasp of those areas in which this type of verification is most needed.

REFERENCES

Roger D. Blackwell, James S. Hensel and Brian Sternthal (1970), "Pupil Dilation: What Does It Measure?" Journal of Advertising Research, 10, 15-18.

Bram C. Goldwater (1972), "Psychological Significance of Pupillary Movements," Psychological Bulletin, 77, 340-55.

Herbert E. Krugman (1964), "Some Applications of Pupil Measurement," Journal of Marketing Research, 1, 15-18.

Herbert E. Krugman (1965), "A Comparison of Physical and Verbal Responses to Television Commercials," Public Opinion Quarterly, 29, 323-24.

Berkeley Rice (1974), "Rattlesnakes, French Fries, and Pupilometric Oversell," Psychology Today, 55-59.

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