Notes
Digital Inequalities and Online Privacy: Using Bayesian Inference to Investigate Differential Access to Privacy Information as a Possible Cause
114Digital inequalities (DIs) are social inequalities transferred to the online world and describe systematic differences in internet users’ behaviors, skills, and experienced outcomes. Caused by sociodemographic factors, DIs have, for instance, been found regarding people’s online privacy skills, knowledge, and behaviors. As an example, education is positively while age is negatively related to internet users’ privacy literacy and protective efforts. Because legal approaches like the GDPR still shift responsibility for privacy-management primarily onto the users, it is essential to understand which factors contribute to digital inequalities in internet users’ privacy behaviors. Therefore, we investigate DIs among two factors that are important for privacy protection, that is privacy literacy (i.e., factual protection knowledge) and privacy self-efficacy (i.e., confidence about one’s protection abilities). Because people can have access to privacy information as part of school-based media literacy education programs, due to personal privacy-management experiences, due to their media diet, and probably also through their professional and social environment, we argue that this access to privacy information is also shaped by sociodemographic factors. According to the homophily principle, people tend to be embedded in social networks characterized by, for example, a common sociodemographic background which might also affect people’s access to privacy information. Therefore, the present work pursues the questions how the sociodemographic variables age, sex, education, and migration background are related to internet users’ access to privacy information, privacy literacy, and privacy self-efficacy and whether access to privacy information is an important avenue for people to gain both literacy and self-efficacy. To analyze our assumptions, we collected a representative sample of the German population (N = 3981). We will use a Bayesian structural equation model to overcome limitations of frequentist inference. Altogether, the current work offers an important analysis of the potential reasons for DIs in the realm of online privacy.
115Online privacy / privacy literacy / digital inequalities / Bayesian inference