Advancements of survey design in election polls and surveys 1 |
|
Convenor | Ms Vilma Agalioti-sgompou (ISER University of Essex ) |
Many pre-election polls identify the preferences of people labeled "likely voters". One challenge in pre-election polling is the fact that often a greater proportion of respondents report that they will vote than the proportion of the population who actually end up voting. Consequently, researchers wish to identify the subset of respondents who are truly likely voters, and different organizations use different approaches. Our study evaluates the effectiveness of a variety of different methods for identifying likely voters. The result is evidence pointing to the methods that can be most effective in identifying which survey respondents will eventually vote.
On September 18, 2014, Scotland held a referendum on independence. Sixty-eight polls purporting to measure support for independence were published in the media in 2014, 18 of which were published in September. These polls tracked changes in voting intentions over time and estimated Election Day results. They significantly overestimated support for independence. Drawing from the lessons learned in the 1995 Quebec sovereignty referendum, the paper addresses the question of why these polls went wrong and why they are generally likely to go wrong in referenda on national independence.
Reported past voting behaviour in political surveys relies on 2 assumptions: 1) respondents recall correctly how they voted; 2) respondents reporting a previous vote are representative of the electorate. However, research suggests that all respondents don’t fulfil this ideal.
Between 2012 and 2014, we conducted an experiment in a French online panel in which still a few thousands answered the latest wave, offering a conservative estimation of the memory problem. Hence, we will 1) measure how many respondents do not repeatedly provide the same answer 2) explore these non-congruent answers. We will draw lessons for representative electoral surveys.
Vote recall is a key variable both for analyses of electoral behavior. All these analyses rely on the answers of to survey questions on vote recall, and, quite often, these survey results do not reproduce accurately the election results they are trying to portray.
We will analyse the responses to the vote recall question regarding the previous general election included in a pre and post wave of a panel election survey (2008 and 2011 Spanish general elections), focusing especially in the analysis of non-consistent respondents (i.e. those that declare a different party choice in the two waves).
Policy representation in the 2013 German Bundestag election is analyzed from the viewpoint
of virtual direct democracy. For this purpose, party positions on 36 topical issues are compared with
the results of public opinion polls, and the party and coalition indices of popularity (the average
percentage of the population represented) and universality (frequency in representing a majority) are
constructed. In particular, the 2013 election winner, the conservative union CDU/CSU with their 41.6%
of the votes, is the least representative among the 28 parties considered. To enhance policy
representation, an alternative election procedure is proposed and hypothetically applied to