Results and Discussion of Conjoint Study

Overview

This chapter presents and discusses the results of the conjoint study explained in the preceding chapter. Both chapters together make up the second component of element D of the outline in figure 1.2. The results refer particularly to the logit model estimation (5.1). Providing a basis to assess the goodness of the estimated models, evaluation criteria for logit models are explained in chapter 5.1.1. Following this, estimation results of differently specified models are shown and reasonable models selected (chapter 5.1.2). The different specifications are particularly based on data exclusion, weighting and segmenting.

Addressing a core question of the conjoint study, chapter 5.1.3 describes and discusses the balance contribution of the individual attributes. On the basis of the estimated coefficients, it classifies the incorporated attributes into four different balance categories. Moreover, by utilizing the coefficients, attributes’ relative importance is inferred and compared to respondents’ direct rating of importance. Concluding the logit model part, chapter 5.1.4 reports on sensitivity and validity tests. Sensitivity checks include removals of variables, variations of the conversion of the product category variable and a check for robustness by a bootstrap/jackknife approach (5.1.4.1). Validity checks refer to face validity as well as internal and external validity (5.1.4.2). As for the willingness-to-offer, the experiences of the modeling process are reported in chapter 5.2. The chapter closes with a summary of the conjoint study’s findings and implications, an evaluation of the performance of the employed measurement and model technique and suggestions for improvement for follow-up studies (5.3).

“It’s very scientific. We just sit around and wait for the phone to ring. Sometimes it’s a wrong number.”

—Warren Buffet on Berkshire’s acquisition strategy

“Far more better an approximate answer to the right question, which is often vague, than an exact answer to the wrong question, which can always be made precise.”

—John Tukey, statistician