Assessing Independent Variables in Cross National Datasets: the Case of Religiosity

JOSÉ RAMÓN MONTERO; MARÍA CELESTE RATTO; GUILLERMO CORDERO

Buenos Aires

Congreso; V Congreso Latinoamericano de Ciencia Política, ALACIP; 2010

Asociación Latinoamericana de Ciencia Política

Comparative cross-national datasets have long been a crucial tool in comparative politics and in public opinion research. In the late twentieth-century, survey research has enjoyed an exponential increase (Heath, Fisher, and Smith 2005:311). Unfortunately, this expansion has not been accompanied by the necessary tools to assess the quality of the data produced or by a rigorous set of procedures to measure their reliability. Oddly enough, the knowledge of the quality of cross-national datasets has been strictly limited (Kohler 2007:405). Recent studies have shown that, despite the increasing number of surveys being used for cross-national research, scholars have not paid much attention to the rules for these data to be really comparable (Long Jusko and Shively 2005: 327; Saris and Gallhofer 2006: 329; Heath, Fisher, and Smith 2005: 297). This problem becomes extremely relevant if we consider the unfinished debates that have taken place on religiosity studies. The use of weak indicators that moreover change in each country and for almost every survey, make comparative analyses virtually impossible and lack any intent to improve on the previous mentioned debates (i.e., Greeley 2003; Norris and Inglehart 2004). The results reported by different cross-national surveys for similar samples from the same country and in the same year show significant differences. From our point of view, evaluating the reliability of public opinion surveys is a pre-requisite in order to carry out a reliable comparison on religiosity. This paper is a first attempt to discuss the reliability of data produced by cross-national surveys. We aim to assess the quality of available indicators on religiosity.