Test developers often select items to maximize
the internal consistency of a scale by
Test developers often select items to maximize the internal consistency of a scale by deleting heterogeneous items. Cattell and
others (Cattell, R.B. and Tsujioka, 1964; Rosnowski, 1987) have questioned this practice because it can lead to seemingly highly
reliable scales which actually measure only a very narrow, homogeneous segment of the target construct, or measure it only in a narrow
group of people