Clinical outcome measurement: Models, theory, psychometrics and practice Leah McClimans a, * , John Browne b , Stefan Cano c a Department of Philosophy, University of South Carolina, Columbia, SC, USA b Department of Epidemiology and Public Health, University College Cork, Cork, Ireland c Modus Outcomes, Letchworth Garden City, UK article info Article history: Received 9 December 2015 Received in revised form 1 June 2016 Available online xxx Keywords: Classical test theory Health outcomes Measurement Models Psychometrics Rasch measurement theory abstract In the last decade much has been made of the role that models play in the epistemology of measurement. Specifically, philosophers have been interested in the role of models in producing measurement out- comes. This discussion has proceeded largely within the context of the physical sciences, with notable exceptions considering measurement in economics. However, models also play a central role in the methods used to develop instruments that purport to quantify psychological phenomena. These methods fall under the umbrella term ‘psychometrics’. In this paper, we focus on Clinical Outcome Assessments (COAs) and discuss two measurement theories and their associated models: Classical Test Theory (CTT) and Rasch Measurement Theory. We argue that models have an important role to play in coordinating theoretical terms with empirical content, but to do so they must serve: 1) as a representation of the measurement interaction; and 2) in conjunction with a theory of the attribute in which we are inter- ested. We conclude that Rasch Measurement Theory is a more promising approach than CTT in these regards despite the latter’s popularity with health outcomes researchers. Ó 2017 Published by Elsevier Ltd. 1. Introduction One thread in the contemporary literature in philosophy of measurement emphasizes the role that models play in measuring outcomes. With some notable exceptions (Boumans, 2015), this discussion has proceeded largely within the context of the physical sciences (Mari, 2000). Models, however, also play qualitative and quantitative roles in psychology. Our interest in this paper is with the methods used to develop instruments that purport to quantify health phenomena, specifically Clinical Outcome Assessments (COAs). The umbrella term for the methods used to develop these instruments is “psychometrics”. In this article, we discuss two psychometric theories and their associated measurement models: Classical Test Theory (CTT) and Rasch Measurement Theory. We argue that models have a role to play in coordinating theoretical terms with empirical content. To play this role, models must: 1) serve as a representation of the measurement interaction; and 2) in conjunction with a theory of the attribute of interest, e.g., one that supplies theoretical quantity terms, explain relationships among theoretical terms. We further argue that the Rasch Model provides a representation of the mea- surement interaction, while the CTT model does not. In the context of COAs, both measurement theories generally fail to utilize an attribute theory. Despite this failure, and CTTs popularity in health outcomes research, we conclude that health researchers should explore the use of Rasch Measurement Theory. 2. Philosophy, models and classical test theory 2.1. Measuring time Although physics enjoys more powerful explanatory theories than psychology (Borsboom, Mellenbergh, & van Heerden, 2004; Taagepera, 2008), physics can provide a useful baseline for thinking about the role of models in coordinating measurement. We provide such a baseline with an exploration of the application of theoretical and statistical models in making inferences about the relationship between measurement indications and measurement outcomes in the context of time (Tal, 2011). Similar to psychological constructs, such as intelligence and physical functioning, time is not observable. Moreover, the definition of the unit of time is ideal. The ‘second’ is defined as the duration of exactly 9,192,631,770 periods * Corresponding author. E-mail addresses: mccliman@mailbox.sc.edu (L. McClimans), j.browne@ucc.ie (J. Browne), stefan.cano@modusoutcomes.com (S. Cano). Contents lists available at ScienceDirect Studies in History and Philosophy of Science journal homepage: www.elsevier.com/locate/shpsa http://dx.doi.org/10.1016/j.shpsa.2017.06.004 0039-3681/Ó 2017 Published by Elsevier Ltd. Studies in History and Philosophy of Science xxx (2017) 1e7 Please cite this article in press as: McClimans, L., et al., Clinical outcome measurement: Models, theory, psychometrics and practice, Studies in History and Philosophy of Science (2017), http://dx.doi.org/10.1016/j.shpsa.2017.06.004