Race as a Constructed Cause Abstract Science and scientists have a checkered record on race. Some scientists (e.g. Linnaeus, 1758, pp. 21–22; Herrnstein & Murray, 1996; Rushton & Jensen, 2008) have argued for “race realism” in a way that aggravates rather than ameliorates racial domination. The response has been to argue that race is a social construct (Kaplan & Winther, 2014; Gannon, 2016), but many scientists have treated “social construct” as synonymous with “nonexistent” (Morning, 2007; e.g. Lewontin, 1972; Bamshad & Olson, 2003; Leroi, 2005; Prontzos, 2019) which also undermines ameliorative projects (Thompson, 2006). I suggest that this unhelpful dichotomy is the result of scientists combining two otherwise innocuous habits without attending carefully to how they interact. The first habit is taking existence or reality as the instantiation of a property which appears in the best scientific explanation of an empirical phenomenon (see Quine, 1960, 1969; Harman, 1967; Cargile, 2003; van Inwagen, 2009). Because genetics is the primary tool for hereditary population studies and race doesn’t appear as a strong genetic cluster within human populations (Romualdi et al., 2002), many biologists simply regard it as nonexistent. The second habit is building multiple related but incompatible causal models in the service of different representational goals (Weisberg, 2007), a practice now widespread in biology (Godfrey-Smith, 2006). These models can be explanatory, because in complex sciences like biology de-idealization is often impossible and they successfully express counterfactual dependencies (Bokulich, 2011). It turns out that race can play such a role in pharmacology, even though and precisely because it is a social construct (Doyle, 2006), which grants it a certain scientific reality. This analysis neglects Bokulich (2011)’s third condition for explanatory modeling, however: the phenomenon to be explained must fall within the model’s domain of applicability. Sociological, pharmaceutical, and other models which make use of race in their explanations are not of unrestricted applicability like the basic equations of physics, or even of applicability to all terrestrial life like genetics. The conclusion, if we are careful in relating the two habits, is that race exists only in a qualified sense. It is an answer to a question that can only be put to certain models. We can therefore see how scientists have so often gone wrong about race. By treating race as an entity which exists (or fails to exist) in an unqualified sense, many scientists have been led to unhelpful articulations of race-realism and race- denialism. Remembering that race is a reality only of certain models with ineluctably social components makes it available for ameliorative projects without pretending that it is a natural kind suitable for generic scientific research.