Abstract
Simultaneous systems of equations with non-normal errors are developed to study the relationship between customer and employee satisfaction. Customers interact with many employees, and employees serve many customers, such that a one-to-one mapping between customers and employees is not possible. Analysis proceeds by relating, or linking, distribution percentiles among variables. Such analysis is commonly encountered in marketing when data are from independently collected samples. We demonstrate our model in the context of retail banking, where drivers of customer and employee satisfaction are shown to be percentile-dependent.
Similar content being viewed by others
References
Anderson, E. W., & Mittal, V. (2000). Strengthening the satisfaction-profit chain. Journal of Service Research, 3, 107–120 (Nov).
Anderson, E. W., & Sullivan, M. W. (1993). The antecedents and consequences of customer satisfaction for firms. Marketing Science, 12, 125–143 (Spring).
Fernandez, C., & Steel, M. F. J. (1998). On Bayesian modeling of fat tails and skewness. Journal of the American Statistical Association, 93, 359–371.
Greene, W. H. (2003). Econometric analysis. New Jersey: Prentice Hall.
Heskett, J. L., Jones, T. O., Loveman, G. W., Earl Sasser Jr., W., & Schlesinger, L. A. (1994). Putting the service-profit chain to work. Harvard Business Review, 72, 164–170 (Mar/Apr).
Heskett, J. L., Sasser Jr., W. E., & Schlesinger, L. A. (1997). The service profit chain. New York: Free Press.
Kahneman, D., & Tversky, A. (1979). Prospect theory: an analysis of decision under risk. Econometrica, 47, 263–291.
Kamakura, W. A., Mittal, V., de Rosa, F., & Mazzon, J. A. (2002). Assessing the service-profit chain. Marketing Science, 21(3), 294–317.
Koenker, R. (2005). Quantile regression. Cambridge: Cambridge University Press.
Koenker, R., & Bassett Jr., G. (1978). Regression quantiles. Econometrica, 46, 33–50.
Kottas, A., & Krnjajic, M. (2007). Bayesian nonparametric modeling in quantile regression. Working paper, Santa Cruz: Department of Applied Mathematics and Statistics, University of California.
Morgan, N. A., Anderson, E. W., & Mittal, V. (2005). Understanding firms’ customer satisfaction information usage. Journal of Marketing, 69, 131–151.
Mittal, V., Ross Jr., W. T., & Baldasare, P. M. (1998). The asymmetric impact of negative and positive attribute-level performance on overall satisfaction and repurchase intentions. Journal of Marketing, 62, 33–47.
Newton, M., & Raftery, A. (1994). Approximate Bayesian interference by the weighted likelihood bootstrap. Journal of the Royal Statistical Society Series B, 56, 3–48.
Rossi, P. E., Allenby, G. M., & McCulloch, R. (2005). Bayesian statistics and marketing. New York: Wiley.
Rust, R. T., & Chung, T. S. (2006). Marketing models of service and relationships. Marketing Science, 25, 560–580.
Streukens, S., & De Ruyter, K. (2004). Reconsidering nonlinearity and asymmetry in customer satisfaction and loyalty models: An empirical study in three retail service settings. Marketing Letters, 15, 99–111.
Yu, K., & Moyeed, R. A. (2001). Bayesian quantile regression. Statistics and Probability Letters, 54, 437–447.
Yu, K., & Zhang, J. (2005). A three-parameter asymmetric Laplace distribution and its extension. Communications in Statistics: Theory & Methods, 34, 1867–79.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
1.1 Estimation algorithms
Estimation proceeds by recursively generating draws from the full conditional distribution of all model parameters. The likelihood of the data can be factored as:
where the first factor corresponds to Eq. 1 and the second factor corresponds to Eq. 2. θ A and θ B represent other model parameters specific to the assumed distributions of the error terms. All model parameters are estimated using a random-walk Metropolis–Hastings algorithm where the likelihood contribution is of the form:
where
1.2 Model 1: Asymmetric Laplace
Prior distributions for all slope coefficients were specified as normal with mean zero and covariance equal to 100I. Inverted chi-square priors with 3 degrees of freedom and prior sum of squares equal to .01 were used for the scale parameters (σ yA , σ yB ). Priors for p yA and \(p_{y_B } \) were specified as uniform(0,1).
1.3 Model 2: Skewed t
Prior distributions for all slope coefficients were specified as normal with mean zero and covariance equal to 100I. Inverted chi-square priors with 3 degrees of freedom and prior sum of squares equal to 0.01 were used for the scale parameters (\(\sigma _{y_A } ,\sigma _{y_B } \)). Priors for \(\gamma _{y_A } \) and \(\gamma _{y_B } \) were specified as inverted chi-square with 3 degrees of freedom and prior sum of squares equal to 0.1. A uniform(0,150) prior was used for \(\nu _{y_A } \) and \(\nu _{y_B } \).
1.4 Model 3: Mixture of multivariate normals
-
1.
Generate z|μ k , Σ k , φ
$$\begin{array}{*{20}l} {z_i \sim {\text{Multinomial}}\left( {\pi _i } \right)} \hfill \\ {\pi _i = \phi _k \frac{{\varphi \left( {\left. {\varepsilon _{y_A } ,\varepsilon _{y_B } } \right|\mu _k ,\Sigma _k } \right)}}{{\sum\limits_k {\varphi \left( {\left. {\varepsilon _{y_A } ,\varepsilon _{y_B } } \right|\mu _k ,\Sigma _k } \right)} }}} \hfill \\ \end{array} $$
Where ϕ(·) is the multivariate normal density and z i is a latent indicator variable that assigns each observation in sample to one of the K mixture components.
-
2.
φ|z
$$\begin{array}{*{20}l} {\phi \sim {\text{Dirichlet}}\left( {\widetilde\alpha } \right)} \hfill \\ {\widetilde\alpha = n_k + \alpha _k } \hfill \\ {n_k = \sum\limits_{i = 1}^n {I\left( {z_i = k} \right)} } \hfill \\ \end{array} $$
Where α k is a prior value that is set equal to 1 for all k.
-
3.
μ k , Σ k |z, α, β
Conditional on the latent indicator variable z i each observation can be assigned to one of the K mixture components. Inference for μ k , Σ k can then proceeds through the use of the standard multivariate regression:
Where: X k = ι k , the unit vector with length equal to the number of observations assigned to component k and:
and \(\hat M = \left( {X_k^\prime X_k } \right)^{ - 1} \left( {X_k^\prime Y_k } \right)\)
Prior parameters were specified as follows: ν 0 = 3, \(V_0 = \left[ {\begin{array}{*{20}c} {0.1}&0 \\ 0&{0.1} \\ \end{array} } \right]\), A = 0.01, \(\overline M = \left[ {\begin{array}{*{20}c} 0&0 \\ \end{array} } \right]\)
-
4.
α, β|μ k , Σ k , z
α and β can be drawn using a standard random walk Metropolis-Hastings algorithm where the likelihood for the model can be computed as follows:
Prior distributions for both α and β were specified as multivariate normal with mean 0 and covariance equal to 100I. To improve mixing of the Markov chain we found it useful to draw the regression coefficients in two distinct blocks.
Rights and permissions
About this article
Cite this article
Dotson, J.P., Retzer, J. & Allenby, G.M. Non-normal simultaneous regression models for customer linkage analysis. Quant Mark Econ 6, 257–277 (2008). https://doi.org/10.1007/s11129-007-9037-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11129-007-9037-1