Asymmetric Within and Between Popularity Effects: Evidence From Multihoming

Abstract

Observational Learning (OL) in Information Systems (IS) literature, inferring product quality from the popularity as an aggregated summary of the purchase history, differs from Word of Mouth (WOM) effects in that OL offers less information and thus leaves more room for interpretation of the quality signal. Our study empirically tests the asymmetric effect of the popularity of a platform on a sale conducted on that platform (within-popularity effect) and the asymmetric spillover effects of the popularity of a platform on sales conducted on the other platform (between-popularity effect) using multihoming games across two representative mobile platforms, i.e., Google Play and Apple’s App Store. Consideration of the multihoming games gives salience to the asymmetries by controlling for matched game quality, a possible cause of simultaneity. We exploit a diverse panel data framework to systematically address unavoidable econometrical issues and a dynamic panel data model to control endogeneity of autoregression. Finally, by applying z-test to compare two matched pairs of coefficients, we found that the within-popularity of the Google Play is significantly greater than that of the App Store, whereas the between-popularity is significantly less. We speculate that contextual information, that is, information publicly available about a platform’s policy, moderates the interpretation of OL signals, causing asymmetric effects.