Modeling age and retest processes in longitudinal studies of cognitive abilities.

Mixed effects models were used to examine the separate effects associated with age and retest on changes in various cognitive abilities. Individuals (N > 800) ranging in age from 40 to 70 years at the 1st measurement occasion were assessed with measures of memory, spatial abilities, and speed on 4 occasions. All cognitive abilities showed decline associated with increased age and improvement across the 4 measurement occasions. The age-related effects were similar across variables, but the practice effects were largest for memory and smallest for speed. When retest effects were not included in the models, the age-related effects were underestimated, with the magnitude of bias depending on the size of the ignored retest effects. It is suggested that both age and retest should be modeled simultaneously when analyzing longitudinal data because part of the change across occasions may be attributable to practice or reactive effects.