New Approaches to Purchasing Behavior: a Discussion

Oded Gur-Arie, The University of Michigan
[ to cite ]:
Oded Gur-Arie (1983) ,"New Approaches to Purchasing Behavior: a Discussion", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 243.

Advances in Consumer Research Volume 10, 1983      Page 243


Oded Gur-Arie, The University of Michigan

Although the three papers presented in this session all deal with some aspects of purchasing behavior they have very little in common and therefore I will comment on each separately.

It is refreshing to see an economist divorcing traditional economic thought and proceed to analyze consumer behavior from a marketing perspective. This is essentially what happens in Richard Thaler's paper on transaction utility.

The author illustrates with several succulent examples how consumers' behavior in the marketplace can and does deviate from traditional economic explanations. The author proceeds to develop a theory that attempts to explain such phenomena. The major contribution of this paper, I think, lies in the attempt to formalize and to lay some ground rules for further research.

The phenomena which the author uses to illustrate and develop his theory are mostly examples of various market situations well recognized in marketing, yet about which no formal body of research has emerged. Various labels have been used to explain particular aspects of consumer behavior (e.g. cognitive dissonance, perceived piece quality relationships, etc.). The transaction utility theory presented in the paper can apparently be used to integrate and explain several such concepts.

The weakness or the paper is in the fact that other than the examples which were used there is no empirical validation to the proposes theory. It would appear that to some of the examples used, other alternative explanations which are not contradictory to traditional economic thought can be found. The explanation to the scalping example, for instance, does not necessarily lie with consumer transaction utility but with sellers' long-term profit maximization objectives. It must be said that the theory is elegant and should stimulate further research and formal hypotheses testing.

The paper by LeBlanc is an attempt to validate the so called "buyclass taxonomy." The author is essentially correct in stating that it is not at all clear how valid was the method used by Robinson and Faris to reach their conclusions. I do disagree, however, with Professor LeBlanc's contention that his findings "contradict" the findings of Robinson and Faris. His findings do reveal, however, more characteristics of the buying tasks and by doing so do extend our knowledge.

Comparisons between the studies could, however, be misleading since the studies are not equivalent. In this study, different measures were used on a convenience sample which unfortunately is too heterogeneous. Seventy-eight national, medium size and small local firms as well as governmental agencies were grouped together. It is very tempting to speculate that meaningful differences exist in the buying processes and decision making across the various groups. One should expect that size, power, financial strength, risk perception as well as other variables should all, to some degree, be functions of organization size and nature, and should affect both the relevance and the importance values-of the attributes studied. Subgroup analysis could reveal such differences.

In the third paper "The Application of Venture Analysis to Consumer Research Problems," the authors extend traditional venture analysis to a more practical level which allows the estimation of errors, error costs, and the selection of data collection methods as well as the required sample size.

While conceptually elegant the proposed method is somewhat naive from a practical viewpoint. The reliance on sample size determination via tolerable error, variance and significance level, results in a number which is only an approximation (not to mention that tolerable error is determined based on a single mean and hence implicitly assumes a one question survey). Neither does the proposed method take into account other practical aspects of sampling problems such as the different non-response rates associated with the various sampling methods which are being evaluated. Hence, given such inaccuracies and the fact that the researcher must be able to estimate the shape of population distribution, one should question the overall precision level of the estimated probabilities of Type I and Type II errors.

The method however can be used for sensitivity analysis to allow researchers to at least approximate when does a particular survey procedure become more efficient than others (assuming the various methods generate data or equal quality). One way by which the paper can be improved is by generating a power curve for the relationship between sample size and error for the various survey procedures.