nach oben

Erschienen in:

2014 | OriginalPaper | Buchkapitel

3. Analysis and Utilization of Conjoint Data (Ratings Based Methods)

verfasst von : Vithala R. Rao

Erschienen in: Applied Conjoint Analysis

Verlag: Springer Berlin Heidelberg

Einloggen

Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.

search-config

KI-gestützte Suche

Aus

Abstract

We saw in the previous chapter various methods for designing and collecting data in conjoint studies. The data collection procedure used almost invariably dictates the type of analytical method used in conjoint analysis. In addition, analysis methods depend on two major factors: the nature of the scale used for the dependent variable (preference) and the desired level of data aggregation.

Sie haben noch keine Lizenz? Dann Informieren Sie sich jetzt über unsere Produkte:

Springer Professional "Wirtschaft+Technik"

Online-Abonnement

Mit Springer Professional "Wirtschaft+Technik" erhalten Sie Zugriff auf:

über 102.000 Bücher
über 537 Zeitschriften

aus folgenden Fachgebieten:

Automobil + Motoren
Bauwesen + Immobilien
Business IT + Informatik
Elektrotechnik + Elektronik
Energie + Nachhaltigkeit
Finance + Banking
Management + Führung
Marketing + Vertrieb
Maschinenbau + Werkstoffe
Versicherung + Risiko

Jetzt Wissensvorsprung sichern!

Jetzt informieren

Springer Professional "Wirtschaft"

Online-Abonnement

Mit Springer Professional "Wirtschaft" erhalten Sie Zugriff auf:

über 67.000 Bücher
über 340 Zeitschriften

aus folgenden Fachgebieten:

Bauwesen + Immobilien
Business IT + Informatik
Finance + Banking
Management + Führung
Marketing + Vertrieb
Versicherung + Risiko

Jetzt Wissensvorsprung sichern!

Jetzt informieren

Vorheriges Kapitel Theory and Design of Conjoint Studies (Ratings Based Methods)

Nächstes Kapitel Choice Based Conjoint Studies: Design and Analysis

Nur mit Berechtigung zugänglich

The specification will be gx + hx², where x is the amount of sugar. The resulting specification for the part-utility function will be linear in parameters. The ideal point will be positive ideal if h is negative and a negative ideal if h is positive.

Even when one conducts aggregate level analysis, heterogeneity can be included by the use of the componential segmentation approach.

This issue of extensive data is one of the challenges in conjoint analysis. One handles this problem by asking relatively few questions to each respondent.

For example, the partworth function for price can sometimes be upward sloping contrary to expectations. This may be due to the information role of price versus its allocative role. One approach to correct this is discussed in Rao and Sattler (2000) and described in Chap. 8; this method calls for collecting two sets of preferences for profiles that include price as an attribute without and with a budget constraint.

An alternative way to estimate individual-level partworths is to specify heterogeneity using finite mixture (FM) models and to estimate mixture (or segment) level parameters and recover individual-level parameters using posterior analysis. See DeSarbo et al. (1992). See also Appendix 2.

This specification is the same as the “componential segmentation” model (see Green et al. (1989)). The componential segmentation approach involves first identifying significant interactions using an iterative procedure and only the significant interactions are included in the final estimation (in order to minimize the number of parameters). This method is slightly different than including all interactions. We will describe a comparison of it with aggregated and subgroup models later in the chapter.

To understand this, assume that there is only one person descriptor Z. Then, this formulation involves specifying the partworth coefficient for the i-th person for the r-th attribute and q-th level in (3.2) as β_irq = β_rq + γ_rqZ_i. The γ-parameter measures the interaction between the attribute dummy variable and the person descriptor. The set-up for this model is shown is shown in Table 3.4.

We should note that the componential model is slightly different than Approach IIIB as indicated earlier.

Readers may note that this study was probably conducted much earlier than 1996 and therefore some of the technical features and prices may appear not up-to-date.

Although not germane to this comparison, the HB approach enables one to estimate the partworths with data from fewer profiles. In the computer study, the root mean square error of the partworth estimates from the HB model with data from 4, 8, and 12 profiles (randomly chosen) was 0.066, 0.045, and 0.0.20 respectively as compared with the estimates obtained with data from all 16 profiles. Thus, it is possible to design conjoint more economically when the HB approach is used for estimation.

Aside, the HB method enables one to estimate the partworths with data from fewer profiles. In this computer study, the root mean square error of the partworth estimates from the HB model with data from 4, 8, and 12 profiles (randomly chosen) was 0.066, 0.045, and 0.0.20 respectively as compared with the estimates obtained with data from all 16 profiles. Thus, it is possible to design conjoint study more economically when the HB approach is used for estimation.

We will discuss CBC methods in Chap. 4.

If the analyst wishes to incorporate no prior information, one sets the initial βbar and A-matrix equal to zero. In that case, the HB estimates will be asymptotically the same as the OLS results. In a similar manner, constraints on signs or the order of partworths (therefore the β-parameters) are incorporated directly in the posterior distribution of the β-vector.

The notation, [u] represents the distribution of u.

We should point out that confidence intervals are not meaningful in the strict sense.

Allenby, G. M., Arora, N., & Ginter, J. L. (1995). Incorporating prior knowledge into the analysis of conjoint studies. Journal of Marketing Research, 32, 152–162.CrossRef

Andrews, R. L., Ainslie, A., & Currim, I. S. (2002). An empirical comparison of logit choice models with discrete representations of heterogeneity. Journal of Marketing Research, 39, 479–487.CrossRef

DeSarbo, W. S., Wedel, M., Vriens, M., & Ramaswamy, V. (1992). Latent class metric conjoint analysis. Marketing Letters, 3(3), 273–288.CrossRef

Green, P. E. (1981). A hybrid utility estimation model for conjoint analysis. Journal of Marketing, 45, 33–41.CrossRef

Green, P. E. (1984). Hybrid models for conjoint analysis: An expository review. Journal of Marketing Research, 21, 155–169.CrossRef

Green, P. E., & Krieger, A. M. (1988). Choice rules and sensitivity in conjoint analysis. Journal of the Academy of Marketing Science, 6, 114–127.CrossRef

Green, P. E., & Krieger, A. M. (1995). Attribute importance weights modification in assessing a brand’s competitive potential. Marketing Science, 14, 253–270.CrossRef

Green, P. E., & Krieger, A. M. (1996). Individualized hybrid models for conjoint analysis. Management Science, 42, 850–867.CrossRef

Green, P. E., Krieger, A. M., & Zelnio, R. N. (1989). A componential segmentation model with optimal product design features. Decision Sciences, 20, 221–238.CrossRef

Green, P. E., Krieger, A. M., & Agarwal, M. K. (1991). A cross-validation test for four models for quantifying multiattribute preferences. Marketing Letters, 4(4), 369–380.CrossRef

Hagerty, M. R. (1985). Improving the predictive power of conjoint analysis: The use of factor analysis and cluster analysis. Journal of Marketing Research, 22, 168–184.CrossRef

Hagerty, M., & Srinivasan, V. (1991). Comparing the predictive powers of alternative multiple regresation models. Psychometerlica, 56(1), 77–85.CrossRef

Johnson, R. M. (1991). Comment on ‘Adaptive conjoint analysis: Some caveats and suggestions’. Journal of Marketing Research, 28, 223–225.CrossRef

Kamakura, W. A. (1988). A least squares procedure for benefit segmentation with conjoint experiments. Journal of Marketing Research, 25, 157–167.CrossRef

Lenk, P. J., et al. (1996). Hierarchical Bayes conjoint analysis: Recovery of part-worth heterogeneity from reduced experimental design. Marketing Science, 15(2), 173–191.CrossRef

Mallows, C. L. (1973). Some comments on C_p. Technometrics, 15, 661–676.

Moore, W. L. (1980). Levels of aggregation in conjoint analysis: An empirical comparison. Journal of Marketing Research, 27, 516–523.CrossRef

Rao, V. R., & Sattler, H. (2000). Measurement of allocative and informational effects of price. In A. Gustafsson, A. Herrmann, & F. Huber (Eds.), Conjoint measurement. New York: Springer.

Rossi, P. E., Allenby, G. M., & McCullough, R. (2008). Bayesian statistics and marketing. New York: Wiley.

Srinivasan, V., & Shocker, A. D. (1973). Linear programming techniques for multidimensional analysis of preferences. Psychometrika, 38, 337–369.CrossRef

Toubia, O., Semester, D. I., Hauser, J. R., & Dahan, E. (2003). Fast polyhedral adaptive conjoint estimation. Marketing Science, 22(3), 273–303.CrossRef

Toubia, O., John Hauser, J. R., & Garcia, R. (2007). Probabilistic polyhedral methods for adaptive choice-based conjoint analysis: Theory and application. Marketing Science, 26(5)(September–October), 596–610.

Vriens, M., Wedel, M., & Wilms, T. (1996). Metric conjoint segmentation methods. Journal of Marketing Research, 33, 73–85.CrossRef

Titel: Analysis and Utilization of Conjoint Data (Ratings Based Methods)
verfasst von: Vithala R. Rao
Verlag: Springer Berlin Heidelberg
Buch: Applied Conjoint Analysis
Print ISBN: 978-3-540-87752-3

Electronic ISBN: 978-3-540-87753-0

Copyright-Jahr: 2014
DOI: https://doi.org/10.1007/978-3-540-87753-0_3