In this analysis, we compare the determinants of the first-level and the second-level digital divide for private internet use in Germany. Our work offers three important innovations. First, we use the exact weekly duration of internet use to explain inequalities in internet intensity, explicitly controlling for non-users. Secondly, we use the frequencies of five different internet applications to further investigate the determinants of the second-level digital divide. Thirdly, we estimate selection models to control for unobserved characteristics of all individuals. Comparing the determinants of the first- and second-level digital divide shows that socio-economic characteristics (age, education, migration) explain inequalities of the first- but not of the second-level digital divide. By contrast, preferences and attitudes regarding new technologies as well as peer effects turn out to be important for both.
digital divide internet use peer effects inequality