Original meta-analysis claimed anti-Flynn effect for backwards digit span but contained unreplicable regression error. Author replicated other model parameters like intercept 7.74 and country effect 0.88. Categorized 599 effect sizes by country and SES while weighting by square root of sample size. Fitted spline controlling for age, SES, country showed backwards digit span gain of 5.5 IQ points (p=0.0063). Forward digit span gained 9.9 IQ points (p=0.00000023). Both gains statistically significant and similar in magnitude. Age peaks performance around 19.5-22.5 years after controls. Data quality low but adjustments improved analysis. Criticism of Flynn Effect as test artifact invalid. Priors slightly shift toward Flynn gains reflecting real intelligence increases.