Estimating the interdependence of television program viewership between spouses: A Bayesian simultaneous equation model

When making product choices, consumers are influenced by the preferences of other consumers, such as family members, friends, neighbors, and colleagues. Preference interdependence among family members is likely to be significant because of cohabitation and strong emotional ties. To estimate the preference interdependence, we specify a simultaneous equation model and propose a Bayesian estimation approach. Unlike existing models that use a spatial autoregressive structure to capture the interdependence of consumer preferences, we are able to estimate the potential asymmetry in the preference interdependence among family members in a more flexible way. In a simulation study, we show that models that ignore interdependence of preferences yield biased estimates of consumers' sensitivity to observed attribute preferences. In an empirical application, we estimate the interdependence of the viewership of television programs between husbands and wives in 481 households. We find that wives' viewing behavior depends more strongly on their husbands' viewing behavior than husbands' viewing behavior depends on their wives' viewing behavior. There exist significant differences in parameter estimates of dependence across categories of television programs. Differences in levels of spousal interdependence across households are partially explained by the age and the education level of the spouses.