Knowledge Bank

I attempt, here, to determine effect of non-available respondents on results of survey research. This analysis is based on the Polish General Social Survey carried out in Poland in 1994, and on its follow-up realized on respondents who were not approached in PGSS. There were two basic questions: (i) whether non-available respondents differ in significant way from the PGSS sample? (ii) how much, and in what direction. this affects conclusions based on the PGSS? I test effect of 'non-availability' in regression models concerning social stratification and various attitudes. Results of the analysis, which may surprise, show that non-available respondents do not affect results of standard analyses in a significant way.