How does the analysis of voter transitions work?

The voter transition analysis works without questioning individuals. Only aggregated data as election results of municipalities, communities, wards, etc. are used. Statistical correlations generated from these results allow predictions for possible electoral behaviour.

The idea is a simple one: if a party gains most votes in those municipalities in which another party had the most votes at the election before, it is interpreted as a vote transition between those parties.

This method is called multiple regression: "Regression" because we trace back the parties' results of the current election to the parties' results of the comparable election, and "Multiple" because we relate a party's current election result to the results of all parties of the comparable election.

The equation of voter transition analysis from the national election 1995 to the national election 1999 would look like this:

ÖVP04 = b1 × SPÖ99 + b2 × ÖVP99 + b3 × FPÖ99 + b4 × LIF99 + b5 × Greens99 + b6 × others99 + b7 × non voters99

More information:

Hofinger, Christoph / Ogris, Günther (2002): Orakel der Neuzeit: Was leisten Wahlbörsen, Wählerstromanalysen und Wahltagshochrechnungen?, in: Österreichische Zeitschrift für Politikwissenschaft (31), 143-158.