Daniel McNeish is correct in that borth KR20 and Cronbach's alpha are measures of internal consistency. Internal Consistency refers to the degree to which you are measuring a single construct, but it does not indicate in any way what construct you are measuring. KR20 is used when dealing with dichotomous (binary) data.
Both KR20 and Cronbach's alpha are measures of internal consistency (broadly referred to as coefficient alpha). Under ideal circumstances (e.g., no missing data, unidimensionality of items) and with binary items, KR20 and Cronbach's alpha should essentially be the same. If items are not binary (e.g., test questions where examinees may receive partial credit), KR20 is not appropriate and Cronbach's alpha is the better choice. In psychology, I've rarely seen KR20 reported and Cronbach's alpha is far, far more common.
Daniel McNeish is correct in that borth KR20 and Cronbach's alpha are measures of internal consistency. Internal Consistency refers to the degree to which you are measuring a single construct, but it does not indicate in any way what construct you are measuring. KR20 is used when dealing with dichotomous (binary) data.
Which internal consistency reliability procedure--K-R #20 OR Cronbach's alpha--fits a situation where data are collected on a 5-point scale that goes from "Strongly Agree" to "Strongly Disagree"?
Please kindly see the attached article for more explanation about Cronbach's alpha by "Nichola Ritta".
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics