Comments on Papers on Variety Seeking

Flemming Hansen, Professor, ekon.dr., The Copenhagen School of Business Administration & Economics
[ to cite ]:
Flemming Hansen (1980) ,"Comments on Papers on Variety Seeking", in NA - Advances in Consumer Research Volume 07, eds. Jerry C. Olson, Ann Abor, MI : Association for Consumer Research, Pages: 270-271.

Advances in Consumer Research Volume 7, 1980     Pages 270-271


Flemming Hansen, Professor, ekon.dr., The Copenhagen School of Business Administration & Economics

The paper by P. S. Raju & M. Venkatesan provides a very nice and brief review of some of the major theories of exploratory behaviour and compares these in an intelligent fashion. I shall not argue with this review, but rather depart in the concluding statements emphasizing that there are several major areas of study which have been neglected, namely:

Impact of stimulus characteristics on exploratory behaviour, and informa-search, and the effect of stimulus repetition, as well as individual differences in exploratory behaviour.

Let me hasten to agree with the authors about the importance of these questions. Let me also rush to suggest that we might more fruitfully approach these questions directly, not having to force them into a framework of a model, which I personally am beginning to question the usefulness of in the area of consumer behaviour. It is so fascinating to apply these theories of curve linear relationships between stimulation and cognitive activities to consumer behaviour, because with curve linear models there is no single set of data, which cannot be interpreted in terms of the model. We only have to refer the data to the proper part of the curve linear function.

One might even claim that every relationship in this world is curve linear if we stretch the independent variable far enough - but in practice, we rarely vary stimuli very widely. Advertisements for instance are rarely very complex or extremely dull.

Given this, the question is, how sensitive is the model, i.e., how flat is the curve around the optimal point?

Because if it is flat, and if we never vary stimulation very much, the whole thing does not matter very much.

Here, by the way, I find Raju & Venkatesan's distinction between novelty effects and conflict effects very useful. It attempts to decompose the complex curve linear relationship. Should I make a recommendation here, it would be to put more emphasize on the role of expectation with regard to novelty and conflict.

Turning to the paper by Stephen A. Goodwin, I like very much his presentation of the concepts of specific and diverse exploration as well as his discussion of collative stimuli properties. I think it is a good thing to be explicit about what aspects one is studying. This is particularly true, when one is working in such a complicated area as variety seeing and arousal controlled behaviour.

I am somewhat more hesitant regarding the experimental manipulation of collative stimulus properties, however. Actually, what is being done experimentally. I remember having seen somewhat similarly back in the fifties in the early days of the study of attention value.

That is, for the naive observer it is not obvious that the experimental manipulation used, really results in variations in what Goodwin defines as collative stimuli. Really what I am saying is, that to be able to carry out experiments of the kind reported here, we are badly in need of a theory based upon the contents of mass communication, which we more certainly can classify.

In line with my uncertainty as to what is really being manipulated, I am also uncertain as to the conclusions which are "tentatively" drawn by the author. Departing in figure 1, I could with as much vigour as the author does argue for almost the opposite conclusions and nobody could really prove me wrong. What I am saying is that I think Goodwin has been somewhat selective in looking upon figure 1 to find a curve linear relationship between the low medium and high levels of manipulations and the dependent variable: Transformed looking time.

All in all, the findings puzzle me. I simply do not understand them. Apart from what has been said so far a possible explanation could be the choice of dependent variable. A number of choice studies were reported in the psychological journals in the late sixties. Studies looking into various aspects of choice behaviour, including choice time. In all of these studies, however, it was very hard to make sense out of the time measures.

Time-use, it seems, is always very difficult to interpret, since it may be influenced by a large number of other factors than those controlled by the experimenter. Actually, I think this may just have happened in the present case.

I do not know whether it will help, but I could suggest to make a combined "low", "medium", and "high" stimuli complexity level variable. This could be done by combining the manipulation levels on all three dimensions. I realize a number of methodological objections which could be raised against this, but on the other hand, such a crude total index of complexity just might work better than the separate measures studied by the author. I do not guarantee for the results, but it might be worth trying.

Finally, a very general observation. We are here talking about a study which uses very large resources. Some of these resources may not be accounted for very easily. I am thinking of the time of the author, the time of the people sitting in this audience, the people organizing this session, the actual carrying out of the experiment, the time of the subject, the data analyses, etc. Having realized how many of society researchers which are involved, it is sad to read about sampling, that what is done is to "draw a convenience sample from the female population of a large community church in a major metropolitan area". I do not criticize the author for the procedure. I know he does not really have much choice. But I criticize an educational and scientific community which has organized itself in such a fashion, that since some resources are accounted for and others are not, we end up, wasting tremendous resources on collected data from respondents in situations from which we can hardly generalize anything.

Edmond W. J. Faison in his paper makes the observation that people differ in the amount of variety they prefer in different areas. Obvious as this observation is, it has often been neglected and it has rarely been treated systematically. This variety seeking behaviour phenomenon, however, could be studied more directly. That is, without having to relate it to arousal theory.

Faison's paper has some problems associated with it, which I shall comment on. Not so much as a critic raised against the paper. Rather I take this opportunity to point at a couple of methodological problems, which one always encounters in studying exploratory or variety seeking behaviour.

First, problems arise because what is being measured -frequency of most requested dinner, of musical selection, and of brand of toothpaste - are, at best, measured at interval scales. This means, that we cannot really make comparisons between the different areas. Let me elaborate a little about this. Faison concludes that more variety is sought in musical selections, than in dinner choices. He concludes so because, on the average, people repeat their musical selections less frequently in 7 choices than they repeat their meat selection within a week. However, how can we know that something measured in terms of "last 7 musical selection" in any objective sense is comparable with the number of different meat types, which has been selected within the last week? Is one and a half musical selection more or less than two dinner meat choices?

The second methodological problem in Faison's paper has to do with the comparison between Japanese and American respondents. Some tentative conclusions are made suggesting that with regard to musical selection Americans have less preference for variety whereas in the area of menu-choices, Japanese have less preference for variety. However, these observations are based upon measurements which - like others which can be carried out in connection with variety seeking - are subject to a number of situational influences. Influences, all of which, it is important to control. In this particular case, one could speculate that if the experiments had been carried out with Americans as tourist in Japan compared with Japanese in their home country, we may have gotten completely reversed findings. This suggests that the situational influence stewing from the fact that the Japanese are away from home, may have more impact upon their variety seeking behaviour than their cultural or any personality based characteristics.

Finally, let me make one comment in connection with Faison's introduction. In this opening discussion he tries to relate the variety seeking behaviour to more fundamental questions regarding men's ability to make free choices.

I do not really see a need for relating the choice of variety regarding dinner meats to questions of existentialism, etc. More importantly, however, I do not really see the connection. To me it is perfectly feasible to form a deterministic model of human behaviour incorporating variety seeking. Actually, as I see it, this is exactly what Berlyne has been attempting with his concept of optimal arousal level controlling variety seeking behaviour.

In conclusion I am sorry having had to be negative both to those involved in theoretical speculations in the area, and to those spending resources on practical research. Honestly, however, I do not believe we have the methodology needed for measuring what we need to measure, if we want to relate concrete consumer behaviour to the abstract theories of arousal level.

I think it is necessary, now and again, to stop up and ask what one is doing; and if my provocative statements have made some of you so angry that you go home and solve the problems, which I so far see unsolved, it has been worth-while to be negative as I have been.