I am developing a scale with total 21 items. there are 7 domains within the test under which there are 3 questions.what reliability measures can be used to test it in SPSS
Hello Naina,
Are decisions to be made from the individual domain scores or from an aggregated score? If domains are thought to be distinct, then reliability estimation should be for individual domain scores.
Two very common possible methods for estimation of domain score reliability (with 3 items) would be Cronbach's coefficient alpha or test-retest. With so few items per domain, the correlations among items would have to be strong to yield a high estimate of internal consistency reliability.
For example, with 3 items, if each correlated with the others at:
1. r = 0.30, then coefficient alpha estimate would be 0.56
2. r = 0.40, then alpha estimate = 0.67
3. r = 0.50, then alpha estimate = 0.75
None of these would be considered strong enough to make decisions about individuals, and several would be judged too low even for research about groups.
Good luck with your work!
Hello Naina,
Are decisions to be made from the individual domain scores or from an aggregated score? If domains are thought to be distinct, then reliability estimation should be for individual domain scores.
Two very common possible methods for estimation of domain score reliability (with 3 items) would be Cronbach's coefficient alpha or test-retest. With so few items per domain, the correlations among items would have to be strong to yield a high estimate of internal consistency reliability.
For example, with 3 items, if each correlated with the others at:
1. r = 0.30, then coefficient alpha estimate would be 0.56
2. r = 0.40, then alpha estimate = 0.67
3. r = 0.50, then alpha estimate = 0.75
None of these would be considered strong enough to make decisions about individuals, and several would be judged too low even for research about groups.
Good luck with your work!
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS