Data Analysis

We first investigated the association of health-related information-seeking internet use with sociodemographic characteristics using t tests for binary variables and F tests for categorical variables with more than two values. We report the Pearson correlation coefficients (r) for continuous variables. In particular, we focused on gender differences in health-related information-seeking internet use because previous studies have shown gender differences in internet use and health information–seeking behavior [13,15]. Next, we used multiple regression analyses for both binary and continuous outcome values of health-related internet use. We used logistic regression analysis for binary outcomes and ordinary least-squares regression analysis for the continuous health-related internet use score. We used heteroscedasticity robust standard errors for the multiple regression analyses. We conducted all analyses in Stata 14.1, using a 5% statistical significance level criterion.