X hits on this document





10 / 27

of being religious or not (Exclusion vs. Inclusion of Transcendence) can be disentangled from the way in which religious contents are dealt with (either in a literal or in a symbolic way). For the purpose of the present study, we decided to derive the two underlying dimensions of the PCB (Exclusion vs. Inclusion of Transcendence and Literal vs. Symbolic) by means of factor analysis (Varimax rotation) as well as the four subscales (cf. the procedure described by Duriez et al., 2000). A Principal Component Analysis (PCA) was then carried out. A scree test (Cattell, 1966) pointed to a two-componential solution. However, since PCA allows freedom of rotation, the componential structures obtained in different samples cannot be compared straightforwardly. Therefore, this structure was subjected to an orthogonal rotation (Varimax rotation). Hence, these components could be interpreted as Exclusion vs. Inclusion of Tran­scen­dence and Literal vs. Symbolic respectively. Although the first part of the analysis was done by using the traditional unweighted sum of item scores, which indicated a two-factor solution proposed like the one in Duriez, Soenens, & Beyers (2003), the second part of the analysis was done by using factor scores. This guarantees reliability equal to or greater than the one that is obtained by using an unweighted sum of item scores (see Duriez, Soenens, & Beyers, 2003). An unweighted sum score does not allow items to contribute differentially to a construct. In contrast, factor scores allow some items to make a greater contribution to the construct than other items. In this way, they are based on all the items included in the Post-Critical Belief Scale instead of only on some of them. Hence, the scores that are derived to represent an individual's position in Wulff's model are not only more reliable but also more accurate. The construction of these weighted scores is described in Duriez, Soenens, & Beyers (2003). After a Principal Component Analysis (PCA) was carried out on these corrected scores, and a scree test (Cattell, 1966) pointed out a two-componential solution, the structure was subjected to an orthogonal Procrustes rotation (McCrae et al., 1996; Schonemann, 1966) towards the average structure reported by Fontaine et al. (2003). Although the Tucker's Phi indices were not calculated,

Document info
Document views115
Page views137
Page last viewedSat Jan 14 17:59:13 UTC 2017