NotesFAQContact Us
Collection
Advanced
Search Tips
Peer reviewed Peer reviewed
Direct linkDirect link
ERIC Number: EJ920201
Record Type: Journal
Publication Date: 2011
Pages: 4
Abstractor: ERIC
Reference Count: 12
ISBN: N/A
ISSN: ISSN-1536-6367
Invoking Arbitrary Units Is Not a Solution to the Problem of Quantification in the Social Sciences
Barrett, Paul
Measurement: Interdisciplinary Research and Perspectives, v9 n1 p28-31 2011
The article by Stephen Humphry (this issue) is a technical tour de force. At one level, the author marvels at the ingenuity and sophisticated logic and argument on display. This is impressive work and thinking whichever way one looks at it. However, after twice re-reading the manuscript, the same question arises on the author's mind: What exactly has any of this to do with the measurement of psychological attributes? Humphry has produced an interesting line of argument and a mathematical derivation that augments the Rasch IRT model so that it can incorporate differential discrimination parameters for sets of questionnaire items composing a "measure" of a latent variable. As part of this derivation, a standard unit of measurement is apparently created for an arbitrary, statistically constructed, latent trait. Within the simulation example Humphry provides, the unit is never even defined; it's presumably invoked somewhere but apparently cannot be specified as unambiguously as one might specify a physical unit. The author would have been more sympathetic with Humphry's work if he had just presented his derivation as something of relevance solely to atheoretical psychometricians/statisticians working with item responses. But, in his abstract, Humphry claims the work is of relevance to measurement in the social sciences. On the contrary, the author doesn't think it has any relevance. As he sees it, the problem remaining for any social scientist is, not one of developing yet more derivations of existing statistical item response models or even new such models, but one of creating bodies of evidence that demonstrate that a psychological attribute does indeed vary additively.
Psychology Press. Available from: Taylor & Francis, Ltd. 325 Chestnut Street Suite 800, Philadelphia, PA 19106. Tel: 800-354-1420; Fax: 215-625-2940; Web site: http://www.tandf.co.uk/journals
Publication Type: Journal Articles; Opinion Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A