Based on factor analysis in SPSS, I got these items for organizational commitment. I applied the Allen and Meyer (1990) questionnaire, which consists of three constructs: affective commitment, Continuance commitment and normative, but I got these five components, which name will be take every scale in these factor analysis? Is it possible to extend the effective and normative dimensions? Please find attached PDF file of factor analysis.
Yes, it is possible to break the subscales into finer subscales. I would first look at the items in each of the new subscales to see if they make sense, i.e., can you interpret them? Do you think they would relate to different things? The next step is to see how strongly correlated each pair is and if they correlate differently with other variables. If they are not highly correlated with each other, and if they have different correlations with at least some other variables, then you have evidence that they are distinct. However, if they are correlated with each other (the two affective subscales or the two normative subscales) and if they correlate the same with other variables, it suggests that the two new subscales are better combined.
Because you have a pre-existing model, I would recommend Confirmatory Factor Analysis (Structural Equation Modeling) rather than Exploratory Factor Analysis. The goal in that case would be to determine whether the original 3-factor model was an adequate fit to the data, with an additional option of determining whether the 5-factor model was a better fit to the data.
Also, you should be sure you replicate the specifications of the original factor analysis: Did those authors use Principal Components Analysis (as you did), or a different version of factor analysis? Did they use an orthogonal rotation (as you did), or did they use an oblique rotation with correlated factors? You can get rather different results, depending on how you specific these features on an Exploratory Factor Analysis.
If you are testing an already existing model (Be it a measurement model), you need to run confirmatory factor analysis. If the results aren't up to the mark then you can explore "why", which also involves Exploratory factor analysis.
If you are using pre-existing questionnaire then you have to use CFA rather than EFA. However, if you prefer EFA, then this analysis will help you to determine contextual variation or cultural influences and it will help to you to refine your dimensions into sub scale. However, you have to need theoretical support for such influences and sub-scale. Further you can collect fresh data to run CFA to justify theoretical contribution, if any.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS