Difference‐in‐differences with ordinal data: analysing the impact of terror attacks on racial harassment fears of UK Muslims

Abstract The method of difference‐in‐differences is central to public policy analysis, but several challenges arise in applying this method to ordinal outcomes, which are routinely collected in surveys and widely analysed in social sciences. In this paper, we propose a user‐friendly estimator to implement the method of difference‐in‐differences with ordinal outcomes to address these challenges. This estimator quantifies the average treatment effect on the treated in terms of response probabilities and allows an assessment of the distributional impacts of treatment. We use this estimator to analyse fear of racial harassment among Muslims living in a non‐Muslim majority country following extremist Islamic terror attacks. Our findings reveal a shift in feeling ‘not at all worried’ to ‘fairly worried’ about racial harassment after terror attacks, with little change in feeling ‘not very worried’ or ‘very worried’.