Abstract:
The main objective of this paper is to present a set of non-parametric tests that allow the analyst to detect the presence of purchase-event feedback. The principle is to set aside the potential impact of heterogeneity by analysing data at the individual level, i.e. consumer by consumer. The null hypothesis is that each person has a constant, personal, likelihood of purchasing a brand. If this hypothesis is true, then the purchases would be allocated randomly over time. On this basis, distributions can be computed for quantities such as the total time span separating the first and last of a series of purchases, the time between the initial trial and the first repurchase, the time between the initial product trial and the last repurchase. The observed values for these quantities can be then be compared, consumer by consumer, to the hypothesized distributions. Rejecting the hypothesis implies that probabilities are affected by purchase-event feedback and tend to be higher, for a given consumer, just after a purchase. One can combine test results at the individual level (thus avoiding the heterogeneity problem) into an overall assessment of the hypothesis.
Research Papers
Testing marketing action with scanning data
Catalogue: Seminar 1988: New Methodologies In Test Marketing
Authors: Denis Bied-Charreton, Isabelle Le Roy
 
June 15, 1988
Research Papers
The retail shake up in Asia
Catalogue: Asia Pacific 2000: Redefining Business In Asia Pacific
Authors: Laurent Zeller, Helen Passingham-Hughes
Company: KANTAR TNS Malaysia
November 1, 2000
Research Papers
Marketing analysis on the basis of consumers' degree of involvement
Catalogue: ESOMAR Congress 1984: What We Have Learned From The Recession
Authors: Gilles Laurent, Jean-Nöel Kapferer
Company: IFOP
June 15, 1984
