Digital trace data have become one of the central pillars of media research methods. Despite the opportunities for better understanding individual users’ true behaviors in the personalized media environment, many scholars have pointed out the potential for bias in trace data collections, questioning the generalizability of findings based on them. This talk will discuss two different kinds of such bias—i.e. sample selection bias and measurement error—and their correction methods. First, Inverse Probability Weighting (IPW) and Type-II Tobit, two popular correction methods for research participants’ differential propensity to ‘donate’ their personal data, will be compared based on full and partial simulations and real trace data from Facebook users. Second, the potential for developing measurement error models for news exposure over/underreported in the survey and trace data by explicitly modeling different error-generating mechanisms embedded in the measures—what Pak recalls error and device access error—will be presented.
Chankyung Pak is an assistant professor in the Department of Media and Communication at Kyungpook National University (KNU) where he teaches data journalism, media economics, and ethics and policies for algorithmic media. He is a computational social scientist who’s interested in the evolution of public discourse that emerges from the conjunction of changing information technologies, media economy, and journalism practice. He has been also working to develop computational methods to overcome the non-standard properties of digital trace data and make full use of them for media study. Before he joins KNU, he worked at University of Amsterdam and Beijing Normal University-Hong Kong Baptist University United International College.