Reanalyzing the impact of cognitive training over 20 years in the ACTIVE study: Losing the forest in the trees
William T. DickensAbstract
INTRODUCTION
This paper reanalyzes the 20‐year follow‐up to the Advanced Cognitive Training for Independent and Vital Elderly (ACTIVE) cognitive training trial, which evaluated whether memory, reasoning, and speed‐of‐processing training reduce dementia risk in older adults.
METHODS
Published hazard ratios and confidence intervals from the follow‐up were used to conduct Bonferroni–Dunn corrections, sign tests on the pattern of hazard ratios, and parametric modeling.
RESULTS
After Bonferroni–Dunn correction, all confidence intervals include a hazard ratio of 1. Sign tests reject both the hypothesis of no training effect and the hypothesis that only one training type is effective. Generalized method of moments estimation of parametric models suggests a common protective effect for all treatments concentrated in the youngest age group.
DISCUSSION
Preregistration strengthens claims of statistical significance, but interpretation need not be limited to the preregistered analytical framework when alternative analyses reveal consistent patterns.