Election poll

from Wikipedia, the free encyclopedia

The polling day , more precisely also the post-election poll (also exit poll according to the English-language technical term) is an instrument of election research . On the day of the election , voters are asked by interviewers how they cast their votes when they leave the polling station . In addition to the current voting decision, socio-demographic data and the voting decision of the last election - recall question - are collected. The data are used for projections in election reporting and for post-election analysis , including the analysis of the voter flow .

history

The instrument was first used in 1967 in the elections to the Second Chamber in the Netherlands by the sociologist Marcel van Dam . A short time later, this also happened in the gubernatorial election in Kentucky, USA . It has also been used in Germany since 1978. The initial use by Infas in the state elections in Hamburg and Lower Saxony was preceded by an unpublished test in 1976. The directors of ARD and ZDF then decided not to continue using the instrument in order to prevent misuse of the data during election day. Exit polls have been used again in Germany since 1990 .

While in Germany the prognosis of the result for election reporting is in the foreground, the data of the exit polls in the USA are mainly used for election analysis. A joint facility of ABC , CBS , NBC , CNN and Fox News Channel collects the data there on election day.

In the case of Bundestag elections, the publication of the results of voter surveys after voting is not permitted until the polling stations are closed and can be punished as an administrative offense with a fine of up to € 50,000 (Section 32 (2), Section 49a of the Federal Election Act ). Similar rules apply to many other elections.

In the run-up to the 2009 Bundestag elections , the possibility of such unauthorized premature publications came into the focus of politics and journalism. Although timely publication is already possible without any problems on conventional websites, particularly new types of internet services such as Twitter (micro-blogging) are viewed with skepticism in this context - especially since the previous publications in this election were made via Twitter.

Exit polls also leaked through Twitter by 5:03 p.m. in the 2013 federal election .

Since the results in Germany are only published after the polling stations have closed (usually at 6 p.m.), the term 6 p.m. forecast is common.

application

Depending on the election, different numbers of electoral districts are selected at random for the polls. In federal elections this is around 400 out of a total of 80,000, in state elections between 120 and 200. In exit polls , between 20,000 (in federal elections) and 5,000 to 10,000 (in state elections) interviews are carried out. This is significantly more than the 1000 or so cases that are recorded in opinion polls and voting intent surveys . When it comes to the composition of the respondents, the exit polls also differ from the voting intention surveys, because only real voters are interviewed. Non-voters and undecided people do not appear in polls on election day. Postal voters cannot be taken into account, which means that some voters are systematically excluded from the sample. The survey takes place in written form with the help of a short questionnaire, which the respondent fills out independently and then throws into a box that resembles a ballot box.

forecast

In contrast to surveys before the election, the polling day does not ask about voting intent, but about actual voting behavior. The raw data is weighted by the polling institute according to an internal key before publication. These weighting procedures are not publicly known, which makes the intersubjective traceability, the prerequisite for a scientific prognosis, impossible.

In election reporting, the polls on election day are used as a basis for forecasts and projections. These are not real prognoses, since the question is not about future behavior, but about actions that have already been carried out. The forecast is only published after the actual event.

Election polls can be an indication of election fraud ; nevertheless, even without election fraud, the result of an election day survey can deviate significantly from the actual election result.

literature

  • Jürgen W. Falter, Harald Schoen (Hrsg.): Handbuch Wahlforschung . Wiesbaden, VS Verlag für Sozialwissenschaften 2005.

Individual evidence

  1. ↑ Klopt dit wel? - Voor nieuws, achtergronden en columns. Accessed February 21, 2017 (nl-NL).
  2. ^ A b c d Dieter Roth: Empirical election research. Origin, Theories, Instruments and Methods . 2nd Edition. VS Verlag, Wiesbaden 2008, ISBN 978-3-531-15786-3 , p. 82 f .
  3. ↑ Jochen Groß: The prognosis of election results. Approaches and empirical performance . VS Verlag für Sozialwissenschaften, 2010, ISBN 978-3-531-17273-6 , p. 60 .
  4. ↑ www.welt.de: Bundestag election The fear of politics of Twitter
  5. ↑ https://twitter.com/christianmutter/status/381802003757690880/photo/1
  6. ↑ Jochen Groß: The prognosis of election results. Approaches and empirical performance . VS Verlag für Sozialwissenschaften, 2010, ISBN 978-3-531-17273-6 , p. 61 .
  7. ↑ http://www.forschungsgruppe.de/FAQ/haeufig_stellen_Fracht/#Hochrechnung
  8. ↑ a b Jochen Groß: The prognosis of election results. Approaches and empirical performance . VS Verlag für Sozialwissenschaften, 2010, ISBN 978-3-531-17273-6 , p. 65 .