The usual approach to detection and adjustment of nonresponse bias in general social surveys has been post-stratification weights, or more recently propensity score weighting based on auxiliary information. There exists a third approach which is far less popular: using multiple imputed values for survey outcomes of each missing unit. Propensity score weighting may require different models for different survey outcomes if their correlation with auxiliary variables varies. Creating an imputed data set could address bias in many survey outcomes in multi-purpose surveys and therefore could be more practical for the second hand users. The objective of this paper is to evaluate the effectiveness of multiple imputation approach, its performance as well as practicality, in detecting nonresponse bias compared to propensity score weighting. To this end, we use 2010 German General Social Survey (ALLBUS) which provides paradata, frame information and clustered external geographic data (e.g. car ownership, purchasing power parity). We compare the two methods looking at changes in means, variations and covariances of survey outcomes.