Return to search

Representing and constructing : psychometrics from the perspectives of measurement theory and concept formation

Social scientific measurement is much desired and much criticized. In this dissertation I evaluate one of the main approaches to social scientific measurement that has nevertheless been virtually ignored by philosophers - the psychometric approach. Psychometric measures are typically used to measure unobservable attributes such as intelligence and personality. They typically take the form of questionnaires or tests and are validated by statistical tests of properties such as reliability and model-fit. My thesis is two-fold. In the first, more critical part, I argue that psychometric instruments normally fail to fulfil plausible criteria for adequate measurement. To make this argument, I define and defend a conception of quantitative representation necessary for measurement. My definition is grounded in the Representational Theory of Measurement but avoids the main critiques this theory has faced. I then show that the typical psychometric process of measure validation fails to produce evidence of such quantitative representation. The upshot is that although a quantitative interpretation of psychometric data is common, it is largely unwarranted. In the second part, I argue that psychometric instruments are nonetheless apt for various other purposes. This argument hinges on a new outlook on how concepts should be formed for psychometric purposes. Philosophers have traditionally thought that concepts should cohere with intuitions and/or pick out so-called natural kinds, while many psychometricians argue that concepts should pick out real as opposed to constructed attributes. I argue that, when it comes to social scientific measurement, it is much more important to focus on the usefulness of the concept, where usefulness can take on different meanings in different contexts. Building on the defended outlook on concept formation, I show what useful functions psychometric instruments can serve even when they fail at quantitative representation.

Identiferoai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:767817
Date January 2019
CreatorsVessonen, Elina Sini Maria
ContributorsChang, Hasok ; Alexandrova, Anna
PublisherUniversity of Cambridge
Source SetsEthos UK
Detected LanguageEnglish
TypeElectronic Thesis or Dissertation
Sourcehttps://www.repository.cam.ac.uk/handle/1810/289446

Page generated in 0.0017 seconds