Wpływ kategorii „Trudno powiedzieć" na wyniki analiz

Abstract

This analysis attempts to determine some important question concerning answers 'do not know' in survey research. We ask whether taking them into account in constructing variables affects results of analyses. Several decades of methodological studies on this question did not provide unambiguous conclusions. Here, we refer to the Polish data which come from the national random sample carried out in 1998. Our research question is as follows: does elimination of 'do not know’ answers in operationalization of some important variables bring significantly different results as compared with case when such categories are included. We check this with respect to selected, extensively used, scales measuring attitudes, such as self-reliance, support for the welfare state, tolerance for some extravagant behaviours (like homosexuality) and support for occupational activity of women. It shows that taking into account categories 'do not know' - in construction of scales pertaining to these attitudes - does not significantly affect validity and reliability of substantive results.