Enhanced Modeling of Top-Box Performance: Bayesian Binary Quantile Regression Applied to Modeling Customer Feedback

• Activity Number: 608
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Jorge Alejandro*+
• Companies: Market Probe
• Keywords: Bayesian binary quantile regression ; Gibbs sampling ; Performance indicators ; Attitudinal drivers ; Key driver analysis

Abstract

Managers across many industries commonly focus their attention on a top-box or a top-two-box percent score to monitor performance on an attitudinal outcome of interest gathered via a customer feedback program (e.g., overall satisfaction). In order to better inform related activities such as resource planning and allocation, key driver analysis is usually carried out by ascertaining the impact product or service attributes have on the chosen attitudinal outcome. Oftentimes, researchers tend to model the mean response of the dependent variable given a set of covariates, even though the actual concern lies on the top-box or top-two-box response. In this research, Bayesian binary quantile regression is utilized to provide a much richer view of how predictors affect the response while the focus remains centered in the original type of response of interest. Simulation results and an applied case study are provided to illustrate the performance of the method relative to conventional methods as well as to highlight the managerial implications of the findings.