ABSTRACT
The CDF taken from the ANES website is a compilation of time series data from the Pre-/ Post-Election Study in presidential election years and the Post-Election Study in midterm years. ‘The CDF is pooled cross-section studies: any respondent for a particular study who is strictly “panel” or “supplement” has been deleted from the Cumulative File.’ For the purposes of this paper, we have used data from the CDF for the years 1980-2004. The sampling is random for these years. We went through the process of recoding all the 2008 main variables of interest as well as the wealth of demographic and socio-economic data described below. In certain instances, the 2008 ANES question phrasing was different than the CDF. In other instances, the list of possible answers from which the respondent could choose was different than the list of answers used in the CDF. In all cases, where there were discrepancies between the 2008 data and the CDF. Public opinion polling, of course, has its inevitable downfalls. Even though several of
our main variables of interest were asked during all of the years in our study, the survey question placement varied from year to year and could have had an effect on the overall results. While most respondents pay close attention in the beginning of a survey, it is conceivable that their minds could wonder as the session goes on. Additionally, the ANES survey was conducted in person in certain years and via telephone for portions of other years. While question placement and type of interview method concerns could be cause for concern, prior research using the ANES data has concluded that data sensitivity for these two issues does not pose a significant bias.6
