|
Abstract:
|
In social, biological, behavior, and clinical trials, it is often interesting for researchers to assess whether different observers can reproduce similar results, so that a better one observer can replace another, or, these observers can be used interchangeably at individual levels. The concordance correlation coefficient (CCC) is a popular scaled index (with values between -1 and 1) for assessing agreement between observers for continuous variables. Traditionally, the CCC is usually developed for fixed observer. The observer effect should be treated as random in the measurement model if they are from a pool of observers. This paper proposes the random-observers CCC with repeated measurements for data from a longitudinal study. Simulations and a vertebral body example will be used for assessing the performance of the proposed methods.
|
