ITEMS, CONSTRUCTS, AND INDICES

This appendix lists all indicators from the regression and path analyses presented in Chapters 6 through 12.

Three types of indicator are employed. Single items are used where a concept is measured with one survey question. We report the verbatim wording of questions along with the frequency distributions of responses. The accuracy (validity and reliability) of single-item indicators is based on the correspondence of the item's wording with the underlying concept (face validity), its association with other theoretically expected correlates (construct validity) or, eventually, test-retest reliability through longitudinal analysis. We also use several two-item constructs, the composition of which is reported here. In this case, validity and reliability are established by the methods already mentioned, but also by examining interitem correlation (Pearson's r) and internal consistency (Cronbach's Alpha). Wherever possible, we use multiitem indices, again reporting how these are constructed. These indices allow us to establish validity through factor analysis (which measures how each observed item relates to a hypothesized latent construct) and reliability analysis (Cronbach's Alpha).

There are many different combinations of factor analysis. To err on the side of caution, we almost always apply the most stringent methods, that is, maximum likelihood extraction and direct oblimin rotation, guaranteeing that if a factor solution can be found, it will also be found via all other methods. Test statistics from factor and reliability analyses are cited in order to establish the accuracy of all multiitem indices.