An empirical study of the choice of meta analysis model on the overall estimates for continuous data with missing standard deviations

The choice between tile fixed aud raudom dfed5 model for pl'O'fidiDg am o'Ytl'd mda auIysis edimate iB continuous data may affect tile accuracy of tllese estimates. For studies witIt complete iDformatio., tile Cochrane's Q-test could provide some guide on tile choice. although the power ofbthis test is quite low. If tile study- level standard deviations ~SDs) an ROt cOlllpletdy reported or 'missing"', selection meta aualysis model should be dOJle witIt more cautio.. Many studies suggest that imputation is a good way of recovering the lost information in the effect size estimate aud the corresponding standard error. lD this article, we compare empirically,.the effeds ofimputatioa ofthe missing SDs 011 the overaB meta analysis estimates based on both the fixed and random effett modeL Tile results saggest imputation is recommended to estimate tile overall effett size. However, to estimate its corresponding staDdard error (SE), imputation is recommended for tile estimates based 08 the random effed model Ifthe fixed eft'ert model is used. impDtation may lead to bias estimates of the SE