发帖
雷达卡
fixed effects:
studies are similar methodologically(there is one true population ES)
all of the variance in ES across the studies is due to sampling variation
to make a statement about the studies on hand, without generalizing to new studies
random effects approach:
For large meta-analyses with large discrepancies in sample sizes between ES and no suspected correlations between sample size and various study characteristics, it may be preferable to use a random effects approach that does not assign equal weight to all.
Goh, Hall and Rosenthal (2016) recomended reporting both.
京公网安备 11010802022788号
