Taylor & Francis (Routledge), Measurement, 1-2(10), p. 50-54
DOI: 10.1080/15366367.2012.681973
Full text: Download
Newton (this issue) proposes various fruitful directions in which to divert the current consensus definition of validity and its implications as entertained in the 1999 Standards for educational and psychological testing. In both Newton’s proposed clarification as well as in the Standards definition, validity is closely associated with measurements of certain attributes. For many researchers in psychology, measurements and attributes refer to the measurement model that is the foundation of the generic latent variable model: an attribute causes responses to a set of items, for example insomnia and fatigue are caused by major depression; and as such, construct validity is often assessed by fitting a confirmatory factor model. Alternatively, the network perspective views psychological phenomena, for example psychological disorders, as networks in which items/symptoms act as autonomous causal agents that mutually influence one another (insomnia → fatigue). From such a perspective, the terms attribute and measurements gather a different meaning and as a result, validity becomes a whole different ball game.