| Sumario: | Using a published correlation matrix (Lawrence & DeCicco, 1997), I re-evaluate the number of dimensions of paranormal belief supported by the revised Paranormal Belief Scale (PBS; Tobacyk, 1988). I use two empirically validated procedures (the minimum average partial and parallel analysis criteria) to estimate the number of latent variables represented in this data set. Both indicate that, for these data, the seven-factor solution favored by Tobacyk and Milford (1983) and the five-factor solution favored by Lawrence, Roe, and Williams (1997) and Lawrence and DeCicco (1997) may represent overextractions. Instead, four latent variables may be all these data can support. Numerous modifications should be made to the PBS before additional effort is expended on the factor structure of its items or on the relative merits of orthogonal and oblique rotation of its factors (Lawrence & DeCicco, 1997; Lawrence et al., 1997). Reprinted by permission of the publisher.
|
|---|