I ran a reliability analysis in SPSS on a set of scale items but the Cronbach's alpha values after item removal are the same for all items. Why am I having this problem?
Hi Stephanie,
the value of removing items based on "alpha if item deleted"/item-total correlation is a trade off. On the one hand, you remove an item with (probably!) huge error - on the other hand, you ...well remove one item. As alpha is based on the average relationship among all items, alpha can be the same (or even worse) compared with the version prior to removing the item.
Having said that, be aware the overall calculation and reasonableness of alpha depends of the met assumptions - that is essential tau equivalence (i.e. all items measure one factor and the loadings are equal). I have seen so many cases where people had heterogeneous scales and ran alpha with it. The consquences are: you get a bad alpha (that is much lower than the true reliablity) and you delete possibly highly relevant/*valid* items.
HTH
Holger
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS