do I, if yes then how do I report items that were removed from a scale?
should I include both the original and adjusted Cronbach's alpha?
If I have removed 5 questions from the questionnaire to improve the internal consistency of the scale, do I mention which questions were exactly removed from the total?
Also, while calculating the knowledge score, do I omit the responses to these questions?
Sana Afreen It would help to know the context. Are you trying to assess the reliability of a newly-created measure or a previously validated measure?
If this is a new measure you've created, by all means remove bad items (i.e., items that show little to no correlation with the rest of the scale and/or items that are not strongly associated with the underlying construct). If you wanted to be thorough you could provide the full set of items in an Appendix or supplemental file and indicate which items were removed from the final scale.
If this a previously published measure, you should probably keep the full set of items used to validate the measure.
Hi,
Cronbach’s alpha gives information on the closeness of the relation between a set of test items as a group. In general, a score of more than 0.7 is acceptable. Some authors suggest higher values of 0.90 to 0.95.
A low value for alpha may be because of an inadequate number of questions on the test. When you add more relevant items to the test it increases alpha. Low interrelatedness between test questions leads to low values, or, measuring more than one latent variable. Questions are only measuring one latent variable or dimension, which means the low-correlated questions can be eliminated or not of this dimension.
The final score is more important. A flow diagram of the tests leading to the final score can be depicted.
Blaine Tomkins Bhogaraju Anand Thanks!
This study assessed the gap in knowledge levels of adolescent girls where correct answers were given 1 mark and 0 for wrong.
I am planning on pooling items in a set of various categories in order to check for better alpha values.
And adding a supplemental file!
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Probability
- Probability Theory
- Reliability