Zhodnocení vlivu práce výzkumných agentur na konstruktovou validitu škál

This article uses a multiple-indicators multiple-causes model to assess the validity of an attitude scale in different survey contexts. Data used in the article come from a methodological experiment that was carried out in the Czech Republic in May 2002. A set of standardised questions was added to the omnibus surveys of three major Czech agencies that routinely conduct public opinion surveys and publicize voting preferences. As all survey instruments in the experiment were standardised, the author argues that the only possible source of variable validity across individual agencies is their fieldwork procedures. Therefore, the model compares not only the construct validity of scales, but also the quality of the fieldwork in individual agencies. It turns out that the scales are valid in all survey contexts, and the assessment fails to identify any major differences between agencies in the quality of measurement.