Internet-based public opinion research
Lamza Posavec V., Rihtar S.
First author's affiliation: Institute of Social Sciences Ivo Pilar, Zagreb, Croatia
Internet-based surveys are becoming widely used in social research and are usually conducted through placing survey questions on web-pages or distributing them via e-mail. In public opinion research, results are usually generalized to the total population. In Internet surveying, a major methodological issue stems from the coverage bias: the sampling frame from which the sample is drawn does not match the target population. The main aim of our research was to investigate the possible effect of coverage bias in Internet-based surveys of public opinion in Croatia. Research was conducted in November 2007, on a representative probabilistic sample of adult population of Croatia. Sample included 1130 respondents from 93 communities or 118 sample points. Results were analyzed using descriptive and inferential statistics and discriminant analysis. According to the results, at the end of 2007, approximately 35% of adult population of Croatia had access to the Internet, and were using it mostly for information search (34%) and e-mail (28%). As expected, Internet users differ significantly from non-users in their demographic and social characteristics (they are younger, better educated, better situated, living in urban communities), but also in their political and social attitudes (they are more critical of the government and their leaders and less satisfied with the situation in society). When comparing the results of the total sample with those of the Internet-users, the observed differences point to a significant bias in almost all of the examined variables of political attitudes and behavior. An additional, important source of bias stems from the unresolved issue of probabilistic sampling, suitable for public opinion research.