Meta-analysis of reduced feedback frequency experiments fails to detect significant change from immediate to delayed retention


In the late 20th century, motor learning researchers began to prioritize delayed ~24-hour retention tests and de-emphasize performance curves and immediate retention tests. Augmented feedback research has been central to this transition since Salmoni and colleagues published the guidance hypothesis in 1984. Although widely accepted that feedback frequency effects depend on the time of measurement, the available evidence has not yet been analyzed appropriately via multivariate mixed effects meta-analysis. The present study addressed this gap by completing a pre-registered meta-analysis of experiments that included either an immediate or delayed retention test and compared a 100% augmented feedback frequency to a reduced frequency. Following (1) a search of PubMed and PsycInfo databases, (2) forward and backward tracing of identified articles, and (3) a targeted author search, a total of 74 articles were identified as eligible for inclusion. Data were missing from a majority of articles, which we addressed with two approaches: First, we emailed the authors of articles with missing data to request the required information. Second, we digitized the plots of any figures that included error bars if the authors were unable to provide the missing data. As a result, we were able to calculate Hedges' g effect sizes from k = 37 studies (N = 1145). Although the effect of a reduced feedback frequency was larger at delayed retention (g = .21) than immediate retention (g = .06), the difference between timepoints was not statistically significant, p = .47. The present data fail to support timepoint-dependent feedback frequency effects.