1
روان شناسی و مشاوره::
همبستگی بین گویه ها
When used to develop scales, regardless of the specific procedures implemented, EFA involves reducing the matrix of interitem correlations to a set of factors or components to be used as a basis for forming provisional scales.
Because internal consistency estimates such as coefficient alpha confound internal coherence with scale length, scale developers often use a variety of alternative approaches - including examination of interitem correlations (Clark & Watson, 1995) and conducting confirmatory factor analyses to test the fit of a single-factor model (Schmitt, 1996) - to assess the homogeneity of an item pool.
To establish homogeneity using interitem correlations, one must examine both the average and distribution of the interitem correlations.
According to Clark and Watson, the average interitem correlation should fall somewhere between 0.15 and 0.50, depending on the theoretical breadth (closer to 0.15) or narrowness (closer to 0.50) of the construct one is trying to measure.
In addition, the distribution of interitem correlations should be inspected to ensure that they cluster narrowly around the average, since wide variation among the interitem correlations suggests a number of potential problems.
واژگان شبکه مترجمین ایران