In a famous APSR article, Schneider and Ingram (1993) have argued that social constructions shape policy choices and group representation. While this article has been cited innumerable times and attracts growing attention due to the growing salience of group politics, this effect remains difficult to test. This paper (1) develops a theoretical argument on how media framing and representation drive policy change via public opinion and draws on recent advances in natural-language-processing to provide an empirical test. Social constructions of groups are measured based on NLI models trained to identify mentions of each group and measure the framing and tone in French and German newspapers (from 1995 to 2023).
Policy change is operationalized through group-specific scales (0–10) measuring the restrictiveness or supportiveness of policies across time. The analysis focuses on three groups over time - women, immigrants, and investors -each associated with distinct media narratives and policy contexts. Regression models test the impact of media coverage on policy outputs while controlling for public opinion trends, economic indicators, and political context. The findings provide valuable and unique insights into how media narratives shape public perceptions and translate into public policy change.