Research

Abstract:

The hypothetical nature of choices collected in typical discrete choice experiments (DCEs) for market research has been a source of concern for both researchers in academia and industry. To the extent that processing the information in choice sets requires effort, classical economic theory questions the external validity of inferences from standard hypothetical (HYP) choices as elicited in market research. Recent studies in marketing indeed demonstrate increased external validity of inferences from choices that are properly incentivized (ICA). However, these studies model the difference between HYP and ICA data collected from the same population as if preferences change. Together, the classical economists’ critique of inference from HYP-DCEs and the notion of changing preferences has led to ignoring the information in HYP data when ICA data are available. In this paper we propose a model that links the information in HYP and ICA data collected in the same population. The model we propose is in the class of sequential-sampling models of choice and assumes that ICA leads respondents to increase their decision effort relative to the standard HYP market research setting, but subject to the same set of “deep” preference parameters. We show that increased amounts of cognitive processing under incentive-alignment materially and plausibly change choice probabilities and outcomes in our model, even if underlying, deep preference parameters are invariant. Our model yields a framework that parsimoniously bridges between data from HYP-DCEs and data from ICA-DCEs that potentially decreases data collection cost at the same level of external predictive validity.

Mediation analysis empirically investigates the process underlying the effect of an experimental manipulation on a dependent variable of interest. In the simplest mediation setting, the experimental treatment can affect the dependent variable through the mediator (indirect effect) and/or directly (direct effect). Recent methodological advances made in the field of mediation analysis aim at developing statistically reliable estimates of the indirect effect of the treatment on the outcome. However, what appears to be an indirect effect through the mediator may reflect a data generating process without mediation, regardless of the statistical properties of the estimate. To overcome this indeterminacy where possible, we develop the insight that a statistically reliable indirect effect combined with strong evidence for conditional independence of treatment and outcome given the mediator is unequivocal evidence for mediation (as the underlying causal model generating the data) into an operational procedure. Our procedure combines Bayes factors as principled measures of the degree of support for conditional independence, i.e., the degree of support for a Null hypothesis, with latent variable modeling to account for measurement error and discretization in a fully Bayesian framework. We illustrate how our approach facilitates stronger conclusions by re-analzing a set of published mediation studies.