Educational Measurement (4th ed.)

Click to increase image sizeClick to decrease image size

Notes

¹It is moreover somewhat unclear what Kane's chapter has to do with measurement in particular, as it discusses ways suited to back up any kind of test score interpretation with any kind evidential or theoretical warrant, whereas validity should, in my view (Borsboom, Mellenbergh, & Van Heerden, 2004), be concerned with a very specific kind of interpretation (e.g., this test measures verbal intelligence) and very specific kind of evidence (namely, evidence that bears on the question whether the test is indeed sensitive to variation in the attribute of interest, that is, whether variation in the attribute of interest causes variation in the test scores).

²As a side note, such investigations would probably reveal that consistent interpretations of procedures commonly followed in educational testing require latent variables to function as common causes of the item scores. For instance, latent variables should mediate effects of relevant background variables for the test scores to have measurement invariance. It is doubtful whether instrumentalist conceptions of latent variables, that figure prominently in this book (i.e., latent variables are just overall summaries of performance, narrative concepts, scaling devices, etc.) can sustain such interpretations, as overall summaries of performance are post-hoc constructions of the researcher that one assumes are causally inert with respect to the processes that lead to item responses (Borsboom, 2008; Borsboom, Mellenbergh, & Van Heerden, 2003, 2004).