Our response variable (Y) is a biomarker of stress that was measured three times (reliability check). Researchers in the field tend to take the average score of the observed score across 3 observations and use the average score as a predictor or outcome. Interestingly, the standard deviation of the 3 observations of Y were also recorded.
Is there a procedure in which I can weight the average of Y for each person by the standard deviation (across the 3 measurement period for that person). For example, if the standard deviation is 0, I would just use the mean score (weight of 1). If however, the the standard deviation is .03, I would shrink the score down a little (maybe .97 * MEAN SCORE) to weight the average.
I am curious to know if there is a formal procedure that has been developed already...or if there are other ideas regarding the matter. Thanks much!
See full answer in a document I produced (to respond to this question) - under "Method" on my home page.
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Physiological Processes
- Stress