| Data on 1,801 Internet users in California were used to examine the impact of online health information seeking on the number of self-reported physician visits. Two-stage instrumental variables (IV) methods were used to correct for the potential endogeneity of online health information seeking. The instruments used (the location of Internet access and the frequency of Internet use) proved to be strong and valid instruments for identifying the effects of online health information seeking. However, a Wu-Hausman test found no evidence of endogeneity, which contradicted the reverse causality theory presented in the conceptual model and the view that selection bias is an inherent feature of observational data. Nonetheless, the estimates in the IV and naive ordinary least squares (OLS) models were consistent. Online health information seeking was associated with a 0.4 and a 0.6 increase in the self-reported number of physician visits in the IV and OLS models, respectively. These results suggest that online health information seeking serves as a complement to, rather than a substitute for, physician services. |
