Un échec des modèles explicatifs de la science politique ?

In 2016, the Leave vote won the referendum on the membership of the United Kingdom in the European Union, a result that few observers had anticipated. Should we consider the fact that most political scientists were unable to predict this outcome as reflecting the failure of their theoretical models? And what are the implications for the state of political science as a discipline if it is judged primarily according to its ability to make predictive claims? Comparing the current condition of political science to that of the discipline of economics following the 2007 financial crisis, this article contends that it is a theory’s ability to retrospectively render a social event or circumstance intelligible – and not to predict it – that should be evaluated as its primary criterion for validity. Re-examining the causes of the Brexit vote, we show that the 2016 vote can be explained by several existing tenets of political science. From this perspective, we discuss what political scientists might – and might not – say about the future, if we admit that prediction does not play the same role in social science as it does in the natural sciences.