For a sample of 270 high school students, the differences between positive and negative item stems are studied using three forms of a computer anxiety scale (original scale, negated items, and mixed stems) to ascertain if the items for each form tended to define a single construct or two different constructs. Internal-consistency estimates of reliability for each form yielded alpha coefficients which ranged from .73 for the mixed-stem format to .95 for the original format. Confirmatory factor analysis (LISREL VI) was employed to test the hypothesis that, for high school students, negative and positive item stems are indicative of different latent variables. The findings tended to support the hypothesis and were consistent with those obtained by other researchers (Benson, 1987 Benson and Hocevar, 1985; Schmitt and Stults, 1985). It was concluded that one should view results with caution when the instrument includes mixed stems, as the two sets of items do not appear to define a single construct.
Stored only when both papers are in the corpus — 131 citations exist upstream.