The techniques of multidimensional scaling have been used in marketing research for the analysis of buyer perceptual and preference judgements. As part of the usual solution, the apparent bases for buyer judgements are extracted, and a multidimensional geometric portrayal of varying relationships between products and buyers is the end result.
Marketing applications of the techniques to date have involved sets of products about which buyers could be reasonably expected to have prior knowledge or opinions. Thus, in the tests to date, derived bases of judgement have not been greatly at variance with what one might have intuitively expected. They have invariably related to some perceived product qualities, be these physical qualities or otherwise.
This thesis tests multidimensional scaling techniques under conditions where buyers do not h3.ve prior knowledge or opinions about some of the products in question. Average perceptual and preference judgements about eight branded food products are collected and analysed. Because of the uneven prior knowledge about the products, the solution to the analysis of perceptual judgements is of little informational value. It appeal's that when faced with a, difficult paired comparison puzzle to solve, buyers will seek the easiest way out. They do this by using their very unevenness in prior knowledge as a basis for judgement, and thus defeat the purpose of using multidimensional scaling for the discovery of important perceived product qualities, and the way in which these qualities separate the products.
The existence of this problem, along with some practical problems of questionnaire design, also make it impossible to extract a meaningful average ideal point. As a further consequence of this, the results of a test rank order correlation between product-ideal point distances and actual purchasing behaviour are far from conclusive.
By way of summary, there appears to be a range of theoretical and practical problems associated with multidimensional scaling analysis which stand to severely limit the scope for application of the techniques to marketing research problems. The greatest theoretical problem is that of circularity, or put another way, the necessity for a researcher to have an intimate prior understanding of the product market to enable a confident interpretation of the multidimensional scaling solution. Thus, it is difficult for multidimensional scaling solutions to produce genuinely novel findings. Regarding practical problems, non-academically oriented subjects can find paired comparison questions very demanding - in fact sufficiently so to discourage their further participation in market research studies. Additionally, if the paired comparison question proves to be very abstract, subjects are likely to use the easiest and most obvious product distinctions to dispense with their unpleasant task quickly. Such distinctions may have little to do with their true purchasing rationale.