This meta-analysis found that simple memory span measures (e.g., Digits Forward) and complex span measures correlate more strongly in adolescents than in children. Eyeballing Figure 1 in the paper, it appears that the correlation increases from about 0.35 at age 5 to about 0.45 at age 20. It does not appear that the finding is an artifact of reliability. The interpretation of the finding is that both types of tests require storage ability and executive ability but in differing proportions. The author of the paper proposes that the executive control of attention in children develops considerably throughout childhood and adolescence whereas the storage capacity of primary memory stays comparatively more stable over time. Thus, performance on simple memory span increases over time mostly as a function of the maturation of executive ability. Thus, simple span measures correlate more strongly with complex span measures (which measure executive control of attention more directly) in older samples than in younger samples.