Frankly speaking, I'm facing difficulties in getting the acceptable value for internal reliability. I've searched many articles and also web-sites on how to solve my problem. Finally, I've found one which is written by Art Kendall, Social Research Consultant. He suggests to check 8 things as stated below:
1. Are items properly reflected?
2. Are you having non-trivial negative correlations among items?
3. Please check your squared multiple correlations (SMC). Your SMC should be lower than your corrected item-total correlations
4. Are all items stated reflect a single construct? (In this case, what does a factor analysis with PAF and Varimax rotation show?)
5. Are there items which show an increase in alpha in the "alpha if item deleted" column of the output?
6. Do you have a lot of non-response?
7. Do you have a substantial number of cases?
8. Were the items written for this analysis or are they from previously developed scale?
No comments:
Post a Comment