Empirical underdetermination: A bigger problem for the social sciences?

Abstract

The now familiar idea that the detection of an empirical phenomenon is inferred from a complex collection of data (Bogen & Woodward 1988, Woodward 1989, 2000, 2010, McAllister 1997, 2011, Glymour 2000, Harris 2003, Massimi 2007, Leonelli 2015, 2019, Bokulich 2020) entails the recognition that not only theories, but also the description of empirical phenomena is underdetermined by evidence. Empirical underdetermination, understood as the underdetermination of empirical phenomena by data, emerges as a major challenge still to be fully acknowledged and carefully approached in the philosophy of science. To face this challenge, it is essential to be able to identify the multilevel theoretical assumptions underlying the production of data models and thus the inference to empirical phenomena. Despite the many difficulties, this kind of analysis has already been attempted with some success in the case of the natural sciences (Kaiser 1991, Leonelli 2009, Karaca 2018, Bokulich & Parker 2021, Antoniou 2021), where background knowledge about instruments and empirical procedures is often explicitly available. However, the situation seems quite different in the case of the social sciences, where the opacity of instruments (Borsboom et al. 2009) and the highly conjectural nature of background assumptions, renders the challenge of empirical underdetermination more dramatic.