Hi,
I conducted a principal axis factoring (PAF) with oblimin rotation on a scale with 5 items. 2 factors were extracted (using Kaiser's criteria (eigenvalue >1) and scree plot analysis): 2 items load on factor 1 (F1) and 3 items on factor 2 (F2). Items do not cross-load.
All items are measuring the feeling of efficacy but different dimensions of it. F1 seemingly measures self-efficacy (my family & friends (item 1), and I (item 2) can do something to...). F2 describes collective efficacy (people in general (item 3), businesses (item 4), governments (item 5) can do something to...).
If I force PAF to extract only 1 factor (to measure "general" efficacy), all item factor loadings are still >.5, and Cronbach's alpha of the 5-item scale is above >.7.
However, I am not sure what would be the better way to proceed to build a regression model with "efficacy" as one/or two independent variable/s: Go for the 1-scale solution (to avoid a 2-item scale of "self-efficacy") and simply mention that efficacy has two dimensions that are measured together in one variable? Or split efficacy into two scales? I saw both approaches in research.
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts