The biasing effect of overreporting cultural behaviour in standard surveys on cultural and arts participation. How can overreporting be understood and countered?
Department of Sociology Ghent University Gent, Belgium
Social desirability can result in a respondent overreporting his own behaviour. Especially for activities that are generally considered highly pursuable as elements of an active lifestyle, overreporting might be an important confounding issue. Participation rates then can be overestimated and, moreover, when overreporting is not evenly distributed in the sample, coefficients of explanatory models can be biased. Question wording can be specifically tailored to counter these biases. Although the general principles underlying overreporting are well documented, much less is known on the causes why some people are more inclined than others to overestimate their behaviour. Our paper investigates these issues by using a large-scale representative face-to-face survey on cultural behaviour and attitudes in Flanders (n=2.849) in which two alternative measurement instruments for cultural participation are used. The first is the Eurobarometer 56.0 question for cultural participation. The second contains specific measures to counter overreporting, such as the use of a shorter reference period, a detailed categorization of cultural activities, a filtering system, open answers and alternates questions on cultural participation with other questions. By comparing the answers revealed from these two alternative measurement instruments we are able to estimate the excess degree of overreporting induced by the Eurobarometer type of questioning. Next, we use a multivariate model to explain differences in overreporting behaviour. As explanatory variables we include socio-demographic background characteristics, lifestyle elements and cultural interest. Concluding, we discuss the surplus-value of the alternative measurement instrument for reaching substantive insights in the dynamics and social structuring of cultural participation.