This essay explores the history of election forecasting alongside the history of survey sampling. In doing so, the following contributes to contemporary scholarship on cultures of prediction, suggesting the notion of predictive stability as a way to conceptualise predictions in social science. In taking an ANT-informed perspective, this essay shows how the development of a stable culture of prediction hinges on the assembling of heterogeneous actors, which stabilisation often takes place in the

more »

... aftermath of major elections. In order to arrive at this conclusion, the essay will proceed as follows: I will first introduce the topic of cultures of predictions in the social sciences and opinion polling, through which I develop the concept of predictive (in-)stability. After this, I will briefly draw on the history of election forecasting and the history of survey sampling to show that new sampling methods are usually not adopted when their superiority becomes apparent, but when predictive instability of the old ones comes to the fore. In doing so, I will show how the evaluation of pre-election polls informs the way polling is done in general, which in turn, leads to closure regarding the general accepted methodological approaches. This closure is oftentimes reached in the aftermath of major elections.