Temporal stability and aggregation bias in the NBD-Dirichlet parameters
Date
2010
Authors
Stocchi, L.
Wright, M.
Editors
Beckmann, S.C.
Ringberr, T.
Ritter, T.
Ringberr, T.
Ritter, T.
Advisors
Journal Title
Journal ISSN
Volume Title
Type:
Conference paper
Citation
Proceedings of the 39th EMAC conference, 2010 / Beckmann, S.C., Ringberr, T., Ritter, T. (ed./s), pp.1-7
Statement of Responsibility
Conference Name
The Six Senses - The Essentials of Marketing (1 Jun 2010 - 4 Jun 2010 : Copenhagen, Denmark)
Abstract
Despite the large number of empirical studies revolving around the NBD-Dirichlet Theory of buying behaviour, there is little research into the stability of the parameters of the model's distributions. We focused on two aspects of parameters' stability: temporal stability and aggregation bias. We fitted the model repeatedly for each available month over nearly three years of panel data for two categories, and also for successive quarters, half years and years. Results showed that the parameters are indeed stable, showing very low coefficients of variation. Aggregation bias was minimal, besides a reasonable scale effect arising with longer periods of analysis. Therefore, the market stability claimed by the NBD-Dirichlet can be extended to the model's parameters too, with little or no aggregation bias.
School/Discipline
Dissertation Note
Provenance
Description
Access Status
Rights
Copyright 2010 EMAC