Governance mechanisms of analytical algorithms: The inherent regulatory capacity of data-driven automated decision-making

Abstract

I draw on a substantial body of theoretical and empirical research on the inherent regulatory capacity of data-driven automated decision-making, and to explore this, I inspected, used, and replicated survey data from Pew Research Center, performing analyses and making estimates regarding % of Facebook users who say they understand not at all/not very/somewhat/very well why certain posts are included in their news feed and others are not, % of U.S. adults who say that it is possible for computer programs to make decisions without human bias/computer programs will always reflect bias of designers (by age group), and % of Facebook users with no assigned category/fewer than 10 categories/10–20 categories/21+ categories listed on their “ad preferences” page. Structural equation modeling was used to analyze the data and test the proposed conceptual model.