The Use of Indicators for Unobservable Product Qualities: Inferences Based on Consumer Sorting

Abstract

Using the dietary supplement black cohosh to demonstrate our method, we employ data on a product characteristic unobservable to consumers to decompose the contribution to consumers’ valuations of observable characteristics into surrogate indicator and direct components. Because consumers are not all “expert appraisers” of the unobservable characteristic, the measured relationship of indicators to the unobservable quality is generally not the one consumers perceive. Consequently, biases that depend upon the nature of consumers’ ineptitude are introduced into the component estimation. The researcher’s inference problem is solved by recognizing that consumers with greater appraisal expertise sort disproportionately to higher quality products. This enables feasible measurement of inept consumers’ relative valuations and conjectures through separate hedonic estimation on high- and low-quality product subsamples. We find that, relative to experts, inept consumers likely underestimate the value of most observable characteristics in indicating black cohosh product authenticity; however they overweight online product ratings.