I am doing a validation study for my master thesis, but am having some difficulties in figuring out my next steps. My reliability is showing a very high chronbachs alpha, but my inter-item correlation is between r = .41 and r = 84 (high reliability, but low validity). When I look at the correlation matrix, there are so many correlations above .7 that I have a difficult time distinguishing what items might be the best to possibly remove from the scale.
Does anyone have any articles or suggestions as to what steps would make more sense at this point?
Hi Anneli Lambrechts I would recommend Exploratory Factor Analysis or like that for this purpose. You cant say directly anything about validity and reliability just looking inter-item correlations. .41 and .84 range is medium and high level values but it is not sufficient.
According to James Gaskin,
Discriminant validity refers to the extent to which factors are distinct and uncorrelated. One of the methods for determining discriminant validity is to examine the factor correlation matrix, and to make sure that the correlations between factors should not exceed 0.7. A correlation greater than 0.7 indicates a majority of shared variance (0.7 * 0.7 = 49% shared variance). It is suggested to do a separate EFA with just the items from the offending factors.
Check the link below for details
http://statwiki.kolobkreations.com/index.php?title=Data_screening
Regards
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Probability
- Probability Theory
- Reliability