Abstract
Longitudinal studies are permeating clinical trials in psychiatry. Therefore, it is of utmost importance to study the psychometric properties of rating scales, frequently used in these trials, within a longitudinal framework. However, intrasubject serial correlation and memory effects are problematic issues often encountered in longitudinal data. In the present work the authors study, via simulation, the impact of uncontrolled sources of serial correlation on newly proposed measures, designed to evaluate reliability in a longitudinal scenario. This study also addresses the relationship between serial correlation and memory effect. The simulations illustrate that ignoring serial correlation can have a severe impact on the estimates of reliability and on inferences related to it. Importantly, the authors show that the underlying modeling framework used in this new approach allows correcting for this type of correlation and avoiding bias. Moreover, it can adjust for the presence of a memory effect. Nevertheless, to achieve that, a careful model building is required.
Original language | English |
---|---|
Pages (from-to) | 255-266 |
Journal | Applied Psychological Measurement |
Volume | 34 |
Issue number | 4 |
DOIs | |
Publication status | Published - Jun 2010 |
Keywords
- hierarchical model
- memory effect
- rating scales
- reliability
- serial correlation