Publication detail

Transition of Votes for Consecutive Elections – Ecological Regression Modelling

Author(s): † prof. RNDr. Ing. František Turnovec CSc., Mazurkiewicz, M., Mercik, J.W.
Type: Article in collection
Year: 2011
Number: 0
Published in: Mathematical Methods in Economics 2011, part I (listed in the Thomson Reuters ISI Index to Social Sciences & Humanities Proceedings (ISSHP))
Publishing place: University of Economics, Prague, Faculty of Informatics and Statistics
Keywords: ecological regression, electorate flows, transition of votes, homogeneity of electorate, decomposition
JEL codes: C44
Suggested Citation: Mazurkiewicz M., Mercik J.W., Turnovec F. (2011), Transition of Votes for Consecutive Elections – Ecological Regression Modelling. In: Dlouhy M. and V. Skocdopolova (eds), Mathematical Methods in Economics 2011, part I, University of Economics, Prague, pp. 468-473.
Grants: GACR 402/09/1066: Political Economy of Voting Behavior: Rational Voter Theory and Models of Strategic Voting
Abstract: The transition of votes is observed when a voter who voted for a given party in a previous election, votes for another party in current election. Ecological regression technique [1, 2, 3, 4, 5, 6] gives an opportunity to obtain quantitative description of electoral behaviour from aggregated data under homogeneity assumption. Aggregated data available from public statistics contains the information about individual voting behaviour indirectly only. Data aggregation bases usually on geographic units such as counties and constituencies. Regional decomposition is an estimation process, used for the description of a voter’s behaviour. It extends ecological regression to larger regions or even for the whole country. For the case 2x2 (two consecutive elections, two parties) the results of econometric approach are almost obvious. For the case nx2 (two consecutive elections, n parties) the statistical approach is necessary. We suggest that in order to estimate the parameters of ecological
regression, the combined regional decomposition of transition matrix and the
maximum likelihood function can be used. Moreover, the assumption of random distribution of residuals is tested too. At the end, we examine the proposed methodology using data from general parliamentary elections in Poland (from 2005 and 2007 elections) using statistical sample containing over 25 000 results from electoral districts.
Downloadable: Mathematical Methods in Economics 2011, part I, 468-473




Patria Finance
Česká Spořitelna