Dear Koyel,

Check this out and the paper will avswer your question:

The Disgust Scale: Item Analysis, Factor Structure, and Suggestions for

Refinement

Bunmi O. Olatunji et al.

Psychological Assessment Copyright 2007 by the American Psychological Association, 2007, Vol. 19, No. 3, 281–297

I would do as you are doing now: EFA and CFA- some results like the factor loadings (>.40) provide information about the relevance of each item for your scale.

Best,

Rita

Care about the rating scale.

If the scores are asymmetrical or has less than 7 options try to do the analises with the polychoric matrix, instead of the pearson one.

Thanks a lot, everyone! I really appreciate the input given by each of you. 

Beatrice, the paper suggested was really helpful. Thanks so much for this!

I also suggest you to check the correlations between every item of the scale. Very high correlations (> .7) indicate item redundancy, and when I administer a questionnaire on a new sample I often find this problem. Redundant items greatly lower CFA fit (they usually show themselves as correlations between measurement errors). Check the distribution of every single item, too: are there ceiling/floor effects? Is the variability too low? Are there possible answers which were never given? These are all indicators of a possibly problematic item. 

I also agree with David: if items have high skewness and/or kurtosis, it would be best to treat them as ordinal and only use polychoric correlations and the WLSMV estimator. Beware, though, as this artificially increases CFA fit, especially if you have many response options. 

Lastly, on the issue of selecting the "best" items based solely on factor saturation, I suggest you the following read. I found it quite enlightening, if frustrating for the lack of practical solutions offered. 

Little, T. D., Lindenberger, U., & Nesselroade, J. R. (1999). On selecting indicators for multivariate measurement and modeling with latent variables: When" good" indicators are bad and" bad" indicators are good. Psychological Methods, 4(2), 192.

Marcello and Julie, 

Thank you so much. I'll go over the journal articles as well (just had a skimming look at them and they seem to be interesting and helpful). 

Here are my two cents on this.

Marcello, sorry but I disagree with your cutoff for item redundancy, most psychometricians would say >.90 is cause for concern. Even with high inter-item correlations, you can check for item redundancy through face validity , or even better, through item response theory analyses (e.g., item characteristic curves).

Koyel, here are few other papers to check. You won't need all, but the titles can help you decide which might be useful, and feel free to email me if you have further questions.

DeVellis, R. F. (2012). Scale development: Theory and applications (3rd ed. Vol. 26). Los Angeles, CA: Sage.

Costello, A. B., & Osborne, J. W. (2005). Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment Research & Evaluation, 10(7).

Henson, R. K., & Roberts, J. K. (2006). Use of exploratory factor analysis in published research: Common errors and some comment on improved practice. Educational and Psychological Measurement, 66, 393-416.

Clark, L. A., & Watson, D. (1995). Constructing validity: Basic issues in objective scale development. Psychological Assessment, 7(3), 309-319.

Hofmans, J., Theuns, P., & Mairesse, O. (2007). Impact of the number of response categories on linearity and sensitivity of self-anchoring scales: A functional measurement approach. Methodology, 3(4), 160-169.

Kim, Y., Pilkonis, P. A., Frank, E., Thase, M. E., & Reynolds, C. F. (2002). Differential functioning of the Beck Depression Inventory in late-life patients: Use of item response theory. Psychology and Aging, 17(3), 379-391.

Lam, T. C. M., & Kolic, M. (2008). Effects of semantic incompatibility on rating response. Applied Psychological Measurement, 32(3), 248-260.

Lozano, L. M., García-Cueto, E., & Muñiz, J. (2008). Effect of the number of response categories on the reliability and validity of rating scales. Methodology, 4(2), 73-79.

Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). New York, NY: McGraw-Hill.

Royston, P., Altman, D. G., & Sauerbrei, W. (2006). Dichotomizing continuous predictors in multiple regression: A bad idea. Statistics in Medicine, 25(1), 127-141.