Broken Robustness Analysis: How to make proper climate change conclusions in contradictory multimodal measurement contexts.

Abstract:

Philosophers of science discuss how multiple modes of measurement can generate evidence for the existence and character of a phenomenon (Horwich 1982; Hacking 1983; Franklin and Howson 1984; Collins 1985; Sober 1989; Trout 1993; Culp 1995; Keeley 2002; Staley 2004; Weber 2005; Keyser 2012). But how can this work systematically in climate change measurement? Additionally, what conclusions can scientists and policy-makers draw when different modes of measurement fail to be robust by producing contradictory results? First, I present a new technical account of robust measurement (RAMP) that focuses on the physical independence of measurement processes. I detail how physically independent measurement processes “check each other’s results.” (This account is in contrast to philosophical accounts of robustness analysis that focus on independent model assumptions or independent measurement products or results.) Second, I present a puzzle about contradictory and divergent climate change measures, which has consistently re-emerged in climate measurement. This discussion will focus on land, drilling, troposphere, and computer simulation measures. Third, to systematically solve this climate measurement puzzle, I use RAMP in the context of drought measurement in order to generate a classification of measurement processes. Here, I discuss how multimodal precipitation measures—e.g., measures of precipitation deficit like the Standard Precipitation Index vs. air humidity measures like the Standardized Relative Humidity Index--can help with the classification scheme of climate change measurement processes. Finally, I discuss how this classification of measures can help scientists and policy-makers draw effective conclusions in contradictory multimodal climate change measurement contexts.