Meta-analytic approaches which combine estimates from published studies or raw data have been popular in the medical sciences. Now that meta-analytic methods are gaining popularity in other fields like the social sciences, new methodological questions arise, e.g., those related to the use of non-experimental, survey-based data. Survey data is subject to unequal sampling probabilities and nonresponse problems. Therefore, researchers have to apply weights ranging from design-based weights to nonresponse weights, as well as poststratification weights. Up to now, it is unknown which meta-analytic strategy is best suited for analyzing combined weighted datasets in the context of multiple regression. That is, (1) estimating a regression on the combined datasets or (2) synthesizing regression coefficients estimated from the single datasets. Through extensive simulations, we will explore the performance of these approaches and also examine different weighting schemes (e.g., transforming weights, poststratification on each dataset separately or the combined dataset).