Reader Comments

Post a new comment on this article

Capturing individual differences appropriately

Posted by Potsdam_EM_Group on 28 Sep 2007 at 22:09 GMT

Pelli and Tillman are interested to claim additivity of rates because, if substantiated, such additivity implies exclusivity. In ANOVAs on individual subjects, they treated the eight conditions as independent groups. The rationale presumably is that trials are not repeated across conditions. This specification ignores, however, that the data were generated by the same subject. Consequently, the test of the interaction may be too conservative. For the analyses over subjects, conditions were specified as a repeated-measures factor. In the interpretation of these results they are inclined to argue the null hypothesis with respect to statistical evidence for interactions, noting that "understanding individual differences in reading rate" would be invaluable. We agree and would recommend to specify both items and subjects as crossed-random factors in linear mixed-effects model (see Kliegl, 2007, for an application to reading fixations). With this procedure, fixed effects are assessed with simultaneous assessment of between-subject and between-item effects, yielding a stronger test for interaction effects than the analyses reported in this article.

Kliegl, R. (2007). Toward a perceptual-span theory of distributed processing in reading: A reply to Rayner, Pollatsek, Drieghe, Slattery, and Reichle (2007). Journal of Experimental Psychology: General, 136, 530-537.

RE: Capturing individual differences appropriately

DenisPelli replied to Potsdam_EM_Group on 26 Oct 2007 at 17:07 GMT

Thanks, we’ll look at the Kliegl paper.

Denis Pelli & Katharine Tillman