Voter transition analyses show the changes in voters' decisions compared to the previous election. They represent the migrations of voters between candidate parties - or of or to the group of non-voters. So it becomes visible:
- how many voters remained faithful to their party,
- how many voters have voted for another party,
- where did new parties get votes from,
- how many voters of which party did not vote this time or turned from non-voters to voters.
Analysis without survey data
FORESIGHT voter transition analyses are based on “aggregate data”, i.e. on the results of districts, municipalities, etc. The statistical relationships calculated there are used to deduce the behavior of the voters.
For instance: If at the current election a party performs on average strong in exactly those communities in which another party in the compared election was strong, FORESIGHT interpretes it as an indication that many voters have changed between these parties.
Method of calculation
The procedure for this is based on probability calculation and is called multiple regression: “regression”, because the party results of the current election are attributed to the party results of the compared election(regressed). “Multiple” because FORESIGHT relates the current election results of one party to the results of all parties at the compared election.
The equation for a voter transition analysis from the national election 2008 to the national election 2013 looks like this for the SPÖ 2013:
SPÖ2013 = b1 × SPÖ2008 + b2 × ÖVP2008 + b3 × FPÖ2008 + b4 × BZÖ2008 + b5 x Grüne2008 + b6 × Other-parties2008 + b7 × Non-voters2008.
With regard to demographic changes, it is assumed for the calculation that the persons entitled to vote remain the same, i.e.. Additions by immigration and first-time voters are equated with departures by removal and deceased. Differences in the number of eligible voters between the two election years considered are compensated by the number of non-voters. If the number of eligible voters increases, these “new” voters in the comparison are considered non-voters.
Extrapolation
The FORESIGHT extrapolation, the assessment of pre-election results, is also based on voter transistion analysis. By means of the voter inflows, which are calculated on the basis of the counted votes, FORESIGHT closes on the overall result.
Behind this is the assumption that trends in socially similar commutities will be similar. In the run-up to the election, therefore, similar groups of parishes must be identified in which comparable voter movements can be expected on election day.
Due to the electoral secrecy voter transistion analyzes can not be validated later. However, comparative studies with so-called exit polls, i.e. polls after a poll, showed that the described method based on aggregate data is very reliable.
Further literature
- Hofinger, Christoph / Ogris, Günther (2002): Orakel der Neuzeit: Was leisten Wahlbörsen, Wählerstromanalysen und Wahltagshochrechnungen?, in: Österreichische Zeitschrift für Politikwissenschaft (31), 143–158.
- Longchamp, Claude (2004): Das unausgeschöpfte Potenzial der gleichzeitigen Verwendung von Individual- und Kollektivdaten. gfs.bern.
- Laurenz Ennser-Jedenastik (2016): Wie Wählerstromanalysen funktionieren. derStandard.at, 27. April 2016.
