Conferences

Although there may be disagreement about the qualities of the standards and instruments developed within survey research, few would disagree that the uncertainty in measurement precision is especially pertinent when performing cross-national research.

The variables that are designed to measure popular attitudes towards progressive taxation in the ISSP “Role of Government” module of 2006 provide an example of this uncertainty for at least some of the countries surveyed. In the vast majority of countries there is a clear pattern indicating support for the concept of progressive taxation in society. However, the samples of a few developing countries display a very deviant pattern with many respondents asking for the poor to carry a higher share of the tax burden. Our research question is whether these outcomes reflect a substantive difference in the support for progressive taxation or if the deviant pattern is a methodological artefact rather reflecting measurement error. The paper attempts to deal with this issue from a number of angles. The analysis will focus on whether, firstly, the deviant patterns could be caused by sampling procedures or methodological indicator non-equivalence, such as shifts in the meaning of items or categories due to translation problems. Secondly, we will try to investigate whether the reasons behind the outcomes depend on a more substantive level of indicator non-equivalence in terms of different interpretations based on different cultural backgrounds, and country characteristics. Not until these sources of errors have been excluded can we believe that country characteristics on the system level or fundamentally different public attitudes towards progressive taxation have led to the aberrant patterns we observe. The presentation will introduce some first results of various “checking analyses”.