We know what they think,but do we know what they do?

Abstract

This article examines the way we go about building conceptual models to analyse quantitative market research data. It positions the discussion within the context of the scientific paradigm and develops some thoughts about whether the correct form of this paradigm is being used. The study considers the relative values of using four different methods of classifying people: demographics, life-stage, geo-demographics and a multivariate value system. It then questions whether these methods are surrogates for deeper values (in particular whether they help in investigating needs that are psychologically driven as distinct from those needs that may be driven by circumstances), or whether they simply duplicate each other. In the majority of cases, the paper demonstrates that the personal values manifest in the life-stage and geo-demographic groups were largely as would be expected based on their mixture of demographics, and to that extent they were mostly duplications of standard demographics. This suggests that the use of additional systems gives very little gain in understanding over what is already captured in conventional demographics. This was further supported by an analysis of activities and brand use where in most cases the demographics gave most of the information. This is not to say that when the exceptions occurred they were not important, or that the unique aspect of geo-demographics, which is to physically locate them geographically, is not of great value. All this immediately suggests that we should try and use conventional demographics in a more sensitive and intelligent way. At the moment they tend not to be used in combination with one another, and it was the combination of demographics that predicted the value system so well in most cases. The obvious solution is to conceive analysis as being on people in the terms we mostly think of them; for example, as a young, downmarket man or a middle-aged, upmarket woman. To do this is obviously valuable and has many practical advantages: groups can be easily envisaged, and are therefore more easily marketed to. This kind of analysis requires large sample sizes, but we are in an age where price constraints have continuously pushed the sample sizes of quantitative surveys down such that only the most pedestrian analysis can be done. A way needs to be found of increasing survey sample sizes, so that combinations of demographics may be routinely used in analysis. The question of whether the paradigm we are all implicitly using to build conceptual models of analysis requires reframing to bring it in parallel with modern scientific thought remains intriguing and is in need of longer-term discussion.