Abramson, C., R. L. Andrews, I. S. Currim, and M. Jones. 2000. “Parameter Bias from Unobserved Effects in the Multinomial Logit Model of Consumer Choice.” Journal of Marketing Research 37(4):410–26.CrossrefGoogle Scholar
Allenby, G. M., and P. E. Rossi. 1999. “Marketing Models of Consumer Heterogeneity.” Journal of Econometrics 89(1–2):57–78.Google Scholar
Andrews, R. L., A. Ainslie, and I. S. Currim. 2002. “An Empirical Comparison of Logit Choice Models with Discrete Versus Continuous Representations of Heterogeneity.” Journal of Marketing Research 39(4):479–87.CrossrefGoogle Scholar
Andrews, R. L., and I. S. Currim. 2002. “Identifying Segments with Identical Choice Behaviors across Product Categories: An Intercategory Logit Mixture Model.” International Journal of Research in Marketing 19(1):65–79.CrossrefGoogle Scholar
Besag, J. 1974. “Spatial Interaction and the Statistical Analysis of Lattice Systems.” Journal of the Royal Statistical Society Series B 36(2):192–236.Google Scholar
Boatwright, P., S. Dhar, and P. E. Rossi. 2004. “The Role of Retail Competition, Demographics and Account Retail Strategy as Drivers of Promotional Sensitivity.” Quantitative Marketing and Economics 2(2):169–90.Google Scholar
Boztuğ, Y., and L. Hildebrandt. 2008. “Modeling Joint Purchases with a Multivariate MNL Approach.” Schmalenbach Business Review 60(October):400–22.Google Scholar
Carlin, B. P., and T. A. Louis. 2000. Bayes and Empirical Bayes Methods for Data Analysis. Boca Raton, FL: Chapman & Hall/CRC.Google Scholar
Carreira-Perpiñán, M. A., and G. E. Hinton. 2005. “On Contrastive Divergence Learning.” In Proceedings of the Tenth International Workshop on Artificial Intelligence and Statistics, edited by R. G. Cowell and Z. Ghahramani. Barbados: Savannah Hotel.Google Scholar
Celeux, G.. 1998. “Bayesian Inference for Mixtures: The Label Switching Problem.” In COMPSTAT 98, edited by R. Payne and P. Green, 227–32. Heidelberg: Physica.Google Scholar
Chib, S., P. B. Seetharaman, and A. Strijnev. 2002. “Analysis of Multi-Category Purchase Incidence Using IRI Market Basket Data.” In Econometric Models in Marketing, edited by P. H. Franses and A. L. Montgomery, vol. 16, 57–92. Amsterdam: JAI.Google Scholar
Cox, D. R. 1972. “The Analysis of Multivariate Binary Data.” Journal of the Royal Statistical Society Series C (Applied Statistics) 21(2):113–20.Google Scholar
Cressie, N. C. 1993. Statistics for Spatial Data. Revised Edition. New York: John Wiley & Sons.Google Scholar
Efron, B., and R. J. Tibshirani. 1998. An Introduction to the Bootstrap: Monographs on Statistics and Applied Probability, vol. 57. Boca Raton, FL: Chapman & Hall/CRC.Google Scholar
Frühwirth-Schnatter, S. 2006. Finite Mixture and Markov Switching Models. New York: Springer.Google Scholar
Geweke, J. 2005. Contemporary Bayesian Econometrics and Statistics. Hoboken, NJ: John Wiley & Sons.Google Scholar
Gupta, S., and P. K. Chintagunta. 1994. “On Using Demographic Variables to Determine Segment Membership in Logit Mixture Models.” Journal of Marketing Research 31(1):28–36.Google Scholar
Hruschka, H. 1991. “Bestimmung Der Kaufverbundenheit Mit Hilfe Eines Probabilistischen Meßmodells.” Zfbf Zeitschrift Für Betriebswirtschaftliche Forschung 43:418–34.Google Scholar
Hruschka, H. 2013. “Comparing Small and Large Scale Models of Multicategory Buying Behavior.” International Journal of Forecasting, 423–434.Google Scholar
Hruschka, H., M. Lukanowicz, and C. Buchta. 1999. “Cross-Category Sales Promotion Effects.” Journal of Retailing and Consumer Services 6(2):99–105.Google Scholar
Kim, B.-D., K. Srinivasan, and R. T. Wilcox. 1999. “Identifying Price Sensitive Consumers: The Relative Merits of Demographic vs. Purchase Pattern Information.” Journal of Retailing 75(2):173–93.CrossrefGoogle Scholar
Kruger, M. W., and D. Pagni. 2008. IRI Academic Data Set Description, Version 1.31. Chicago: Information Resources Incorporated.Google Scholar
Laud, P. W., and J. G. Ibrahim. 1995. “Predictive Model Selection.” Journal of the Royal Statistical Society Series B 57(1):247–62.Google Scholar
McFadden, D. 1974. “Conditional Logit Analysis of Qualitative Choice Behavior.” In Frontiers in Econometrics, edited by P. Zarembka, 105–42. New York: Academic Press.Google Scholar
Mild, A. and Reutterer, T. 2003. “An improved collaborative filtering approach for predicting cross-category purchases based on binary market basket data.” Journal of Retailing and Consumer Services 10(3):123–133.Google Scholar
McLachlan, G. J., and D. Peel. 2000. Finite Mixture Models. New York: John Wiley & Sons.Google Scholar
Moon, S., and G. J. Russell. 2004. A Spatial Choice Model for Product Recommendations. Working Paper, Tippie School of Business, University of Iowa.Google Scholar
Nylund, K. L., T. Asparouhov, and B. O. Muthen. 2007. “Deciding on the Number of Classes in Latent Class Analysis and Growth Mixture Modeling: A Monte Carlo Simulation Study.” Structural Equation Modeling 14(4):535–69.CrossrefGoogle Scholar
Rasmussen, C. E. 1996. “Evaluation of Gaussian Processes and Other Methods for Non-Linear Regression.” PhD Thesis, University of Toronto.Google Scholar
Russell, G. J., and A. Petersen. 2000. “Analysis of Cross Category Dependence in Market Basket Selection.” Journal of Retailing 76(3):319–32.Google Scholar
Russell, G. J., S. Ratneshwar, A. D. Shocker, D. Bell, A. Bodapati, A. Degeratu, L. Hildebrandt, N. Kim, S. Ramaswami, and V. H. Shankar. 1999. “Multi-Category Decision-Making: Review and Synthesis.” Marketing Letters 10(3):319–32.CrossrefGoogle Scholar
Seetharaman, P. B., C. Siddhartha, A. Ainslie, P. Boatwright, T. Chan, S. Gupta, N. Mehta, V. Rao, and A. Strijnev. 2005. “Models of Multi-Category Choice Behavior.” Marketing Letters 16(3–4):239–54.CrossrefGoogle Scholar
Spiegelhalter, D. J., N. G. Best, B. P. Carlin, and A. van der Linde. 2002. “Bayesian Measures of Model Complexity and Fit.” Journal of the Royal Statistical Society Series B 64(4):583–639.Google Scholar
Varki, S., and P. K. Chintagunta. 2004. “The Augmented Latent Class Model: Incorporating Additional Heterogeneity in the Latent Class Model for Panel Data.” Journal of Marketing Research 41(2):226–33.CrossrefGoogle Scholar
Wang, H. D., M. U. Kalwani, and T. Akcura. 2007. “A Bayesian Multivariate Poisson Regression Model of Cross-Category Store Brand Purchasing Behavior.” Journal of Retailing and Consumer Service 14(6):369–82.Google Scholar
Wedel, M., and W. A. Kamakura. 1998. Market Segmentation: Conceptual and Methodological Foundations. International Series in Quantitative Marketing. Boston, MA: Kluwer Academic Publishers.Google Scholar
Wedel, M., W. A. Kamakura, N. Arora, A. Bemmaor, J. Chiang, T. Elrod, R. Johnson, P. Lenk, S. Neslin, and C. S. Poulsen. 1999. “Discrete and Continuous Representations of Unobserved Heterogeneity in Choice Modeling.” Marketing Letters 10(3):219–32.CrossrefGoogle Scholar
About the article
Published Online: 2013-09-28
Examples for this approach which deal with another multicategory problem by focusing on the covariance of coefficients across categories are the probit models of Ainslie and Rossi (1998) and Seetharaman, Ainslie, and Chintagunta (1999) or the logit models of Hansen, Singh, and Chintagunta (2006) and Singh, Hansen, and Gupta (2005). In the latter two studies, a factor analytic structure is imposed on the covariance matrix of coefficients.
There has been a long discussion on the performance of finite vs continuous mixture models (e.g. Wedel et al. 1999; for a short summary, see Varki and Chintagunta 2004). Whereas Allenby, Arora, and Ginter (1998) and Allenby and Rossi (1999) argue that FM models do not sufficiently represent consumer heterogeneity, especially when the number of segments is small and complete homogeneity within a segment is an unrealistic assumption, Andrews, Ainslie, and Currim (2002) do not find any performance superiority of the continuous over the FM models. Even for a very limited number of segments (one to three), the continuous and the discrete model recover parameters and forecast holdout data equally well. Additionally, Wedel and Kamakura (1998) and Wedel et al. (1999) stress the consistency of the FM model with the way management thinks about consumers in segments.
Segment is the marketing interpretation of a component in a FM model. Therefore, the terms segment and component are used interchangeably in the text.
In contrast to probit models, there is no biasing effect of joint non-purchase that would be the most frequently occurring event. We also remark that we follow the cross-category effect definition by Hruschka, Lukanowicz, and Buchta (1999). In contrast to Russell and Petersen (2000), cross-category effects do not depend on a household’s typical basket size. This modeling decision is justified, because (1) the inclusion of basket size resulted only in a weak improvement of the LL value for holdout data in the RP model and (2) our model already accounts for interaction effect variability by estimating different effects for different segments.
We thank one anonymous reviewer who suggested to discuss this issue and drew our attention to the latent variable interpretation of co-incidence. We also thank the editor for suggesting relevant references.
With , Z has elements, that is, all possible market baskets. Huang and Ogata (2002) observe exponents between 9 and 15 to be the limit of computation.
The general disadvantage of wrong standard errors can be easily adjusted for, as correct standard errors can be computed with bootstrapping (e.g. Efron and Tibshirani 1998).
This two-step approach is conventionally used in FM models for multicategory choice (e.g. Song and Chintagunta 2007).
See, for example, Andrews and Currim (2002) for a complete tabulation including formulas for calculations.
We argue that the homogeneous model smoothes interaction effects. The lower number of interaction effects included for the heterogeneous model might also contribute to such results. Boztuğ and Reutterer (2008) formulate a similar hypothesis, though they motivate it differently. Chib, Seetharaman, and Strijnev (2002) present the opposite effect.
Our model specification does not include RP’s category-specific household variables time since last category purchase (TIME) and loyalty (LOYAL). As this model is estimated over the purchases within one shop only neglecting purchases in other stores, we do not have complete information on a consumer’s shopping history. Therefore, the values of TIME and LOYAL would not be meaningful. Besides, we already account for heterogeneity with the FM model and do not need auxiliary measures of consumer diversity.
Test runs showed that the model parameters and especially the household assignment to segments stabilize quickly.
For reasons of comparability, PLL is also calculated for the independence model whose parameters are estimated by ML.
We only consider interaction effects larger than 0.001 in absolute size.
We thank one anonymous reviewer who recommended to compare the results of our model to those obtained by a correlational analysis.
Household income in thousand US$.
HS: high school; C: college; TS: technical school; PG: postgraduate work. SC: some college, what means that the person left college without a degree.
For model stability, variable selection is not applied to category constants or marketing-mix coefficients.