Constraints on general slowing: a meta-analysis using hierarchical linear models with random coefficients

Hierarchical linear models (HLM) with random coefficients are more appropriate than ordinary least squares (OLS) methods in the meta-analyses of response time (RT) data and for testing general slowing theories. HLM show significant variance in slowing across tasks and demonstrate a process-specific effect by showing less slowing for memory scanning than for visual-search and mental-rotation tasks. OLS, on the other hand, does not account for the nested nature of multitask RT data, making them unsuitable for testing GS theories.