Events and reliability of measures: the effect of elections on measures of interest in politics.
English title: Events and reliability of measures: the effect of elections on measures of interest in politics
Author(s): Mihkel Solvak -
Type: Conference paper/poster
Collecting event data is considered necessary in order to interpret comparative survey data about citizen’s interest and involvement in politics. Event data is seen as a vital tool to estimating the reliability of certain variables in between countries, thus making cross-national comparisons more robust. Elections are commonly considered to affect the reliability of variables that measure subjective interest in politics and political involvement. The standard advice being that questions about interest in politics should not be asked during election campaign due to the risk that interest will be artificially high at that time. This paper, using European Social Survey data, aims to determine whether elections affect the level of interest in politics. Fieldwork periods that take place over several months make it possible to compare the change in levels of interest in politics as election time approaches and campaigns intensify (or on the other hand as elections pass and the attention of the public turns toward other issues). The scope and comparability of the ESS makes it possible to determine this effect on a cross-national level. The distance on a logarithmic scale from last national election and the distance to the next national elections is calculated for every interview in every country. This is then compared with the change of levels in variables measuring interest in politics throughout the fieldwork period. The effect of elections on the level of interest in politics seems plausible from a theoretical perspective. Elections are a mayor event in the political life of every country, they are widely covered by the media, the campaigning by politicians is widespread, and the event itself is regular making it easy to see any reoccurring fluctuations in public opinion. The aggregate results however show that, despite weighting the data in order to make respondents of different fieldwork months comparable to the overall country sample, elections do not have a significant effect on the interest in politics. The same applies to individual countries, also in cases where elections are very close and campaigns ongoing. Elections, as a pivotal “political event”, do not seem to compromise the reliability of variables used to measure interest in politics and changes in time-distances do not correlate with changes in levels of interest. The average citizen seems to be interested or disinterested in politics, regardless of whether national elections happen to take place or not. The paper concludes with a discussion of event data collecting and usability of it in cross-national surveys.
Conference name: International Conference on Survey Methods in Multinational, Multiregional, and Multicultural Contexts
Start date: Jun 25, 2008