Featured Former Fellow Lunchtime Talk: Maria Serban, University of East Anglia

The trouble with construct validity

Abstract: An important part in the methodological discourse of psychology focuses on establishing the field as a hard science. The longstanding operationalist tradition emphasised the commitment to experimentalism, to identifying and individuating measurable variables, and to developing local and general theories on the go and on the back of these measurements. However, the growing reputation and interest in psychology was often accompanied by a matching distrust of a science lacking general theories and formal frameworks for explaining the phenomena argued to fall into its domain. To acknowledge the problem of theory avoidance in psychology, the methodological discourse has been expanded with the notion of construct validity. Its main purpose was to flag that in validating the measuring procedures used routinely in psychology one can achieve the theoretical progress desired by many.

This paper challenges the idea that construct validity has played this role of bridging the gap between measurement and theory assigned to it in the general methodological discourse of psychometrists.

I will start with a brief historical overview of how and why the notion of construct validity was introduced in experimental psychology. Then I will situate this notion in the broader methodological discourse of experimental psychology, which will allow me to articulate some challenges facing the practises developed around the notion of construct validity. I illustrate these challenges with the help of two case studies, following the outline of research into general intelligence and memory research. I conclude with some reflections on the proper place of construct validation in the methodological discourse and practices of psychology.

This lecture is an online only event, registration is required. Please register at https://pitt.zoom.us/webinar/register/WN_RG66U8R0QVivD9gZkg0Mmw

Featured Former Fellow Lunchtime Talk: Naftali Weinberger, Munich Center for Mathematical Philosophy

Signal Manipulation and the Causal Analysis of Racial Discrimination

ABSTRACT: Discrimination is, in part, a causal concept. To say that an individual was discriminated against based on race entails that her race made a difference to how she was treated. Yet demographic variables such as race elude standard causal analyses. Whereas one typically tests causal claims by manipulating one feature of an experimental setup while keeping others fixed, it is unclear whether one can sensibly talk of changing an individual’s race in isolation from her other properties. Those seeking to address this problem face a dilemma of either adopting more limited conceptualizations of race that allow it to be manipulated or more sociologically sophisticated ones on which it cannot be. To resolve this dilemma, I develop the signal manipulation approach. On this approach, one tests for discrimination not by manipulating race itself, but rather by manipulating the signals by which race influences discriminatory behaviors. I argue that my proposal is able to account for audit studies of discrimination better than alternative proposals and that it is compatible with understanding race as socially constructed. Additionally, the proposal helps differentiate between the cases where it is and is not fruitful to model race causally.

This will be an online-only event and registration is required.  Register here: https://pitt.zoom.us/webinar/register/WN_UbFcyFpTTQW0ReTKG6LXeg

Featured Former Fellow Lunchtime Talk: Erik Curiel, Munich Center for Mathematical Philosophy

This will be an online event.

Math Does Not Represent

ABSTRACT:  On the standard—almost universally (albeit often only implicitly) accepted—picture of the relation of mathematics in a physical theory to the world, mathematical entities represent physical entities, mathematical structures represent physical structures, and so on.  The relation of representation is—again, almost universally, often implicitly—taken to be one of a designative, depictive or verisimilitudinous character.  I first present eight problems for this standard picture, which I consider damning.  I conclude that math does not represent, at least not in any standard sense from formal semantics, philosophical logic or ontology, nor even in any sense based on more informal ideas such as similarity.  The essential relation to study to comprehend the nature of physical theories and to understand the structure and character of our knowledge in physics is that between our concepts and the world, something I argue to be very like a Peircean symbol.  Mathematics provides us a wealth of different tools to use in order to bring our concepts and the world into contact (and that itself in a number of different ways), nothing more, nothing less.  Some of those tools function in ways that superficially resemble standard ideas of representation, even though they are nothing like it in any important sense; others do not.  I conclude with a few consequences of this view for the epistemology of science and for the numbingly endless debates about realism.

Register here: https://pitt.zoom.us/webinar/register/WN_SGyslNXrSbCbDESWBcq64g