The adjusted recall difference after previous lists of 13 and 5 digits was -0.065 positions in 86 ds003838 v1.0.6 participants

This study tested recall after previous lists of 13 and 5 digits in OpenNeuro ds003838 v1.0.6. It used 8,461 adjacent memory trial pairs from 86 people. The model controlled current list length, block, and trial position. Each current length and person had equal weight. The adjusted difference was -0.065 correct positions. Its 95% participant bootstrap interval was -0.154 to +0.024. The raw difference was +0.032. The raw difference minus the adjusted difference was +0.097, with an interval of +0.051 to +0.146. The main interval includes zero and does not establish equal recall. Exact enumeration showed that balanced block order can give a positive raw difference with no previous-length effect, under stated assumptions. Independent checks reproduced the source scores and main model estimates. A fresh download and rerun reproduced all 14 result files. Three figures also had the same bytes in two runs. The package contains fixed source files, hashes, code, trial pairs, model outputs, bootstrap samples, figures, and reproduction steps. The local plan was not externally registered. The estimates are behavior associations in one cohort. They do not identify a neural cause.

serial recall · working memory · trial history · list length · memory task · participant bootstrap · balanced block design · OpenNeuro