In the context of psychiatric/psychological care, clinicians who want to test if a single client has significantly improved or worsened on certain test scores calculate Reliable Change Indexes. (RCI) Usually you would calculate an RCI for a posttest raw score minus a pretest raw score.
I have seen some instances where people calculate an RCI score based on several sorts of standardized/norm/scaled score (however you want to call it). It seems counterintuitive to me to calculate RCI (which is kind of like a z-score, a standardized score), on a difference score from another standardized score. Then again I can't really come up with a good reason why you shouldn't.
For example, a type of standardized score very often used in testing clinical patients is a T-score, which is a score transformed from a z score to have a mean of 50 and a SD of 10. Not to be confused with a t-score for t-testing in statistics! I wonder, is there any reason why one shouldn't use a difference score from T-scores instead of raw test scores for an RCI calculation?
Using standardized scores like T-scores for RCI calculation is valid if the same standardization and reference group are used for both pretest and posttest scores. This ensures comparability and allows for easier interpretation across different tests or measures. Be cautious about potential issues if the reference group or standardization method changes between tests.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size