uni'wissen 01(3)-2011_ENG

Prof. Dr. Uwe Wagschal studied political science and economics at the ­University of Heidelberg. After completing his doctor- ate in 1996 he served as a research assistant at the University of Bremen. From 2001 to 2003 he worked as a political advisor at the think tank “Avenir Suisse,” before accepting a chair in empirical political research at the University in Munich. In 2005 Wagschal was ­appointed as professor for comparative political science at the University of Heidelberg, and since 2009 he has served as ­professor in the same area of specialization at the Uni- versity of Freiburg. His ­research interests include comparative public policy research, ­direct democracy, government budgets, and conflict research. What turned out to be decisive was above all social-structural factors. Although chance de- livers good results, the composition of the sam- ple did not accurately reflect the population structure of the state. To compare the two, the election researchers studied data from the State Statistics Office of Baden-Württemberg on fac- tors like the distribution of age, sex, level of ­education, income, and household size. If they found that a certain age group was not sufficient- ly represented in the sample, they weighted the data from this age group more strongly in the analysis. However, the researchers would have had to weight a few respondents who belonged to several underrepresented groups 20 times more strongly to accurately reflect their repre- sentation in the state. They thus decided to limit the weighting factor to 2.5. They also decided against a so-called political weighting, which would have affected the typical Sunday poll question in particular. “Election re- searchers typically reason that voters’ old pre­ ferences become stronger as the election draws nearer,” explains Wagschal. They assume that people who supported a certain party in the pre- vious election, particularly those who supported the currently ruling party, will be less likely to change their minds and support the other party in the last weeks and days before the election and that opinion polls conducted shortly before an election thus often tend to overestimate sup- port for the opposition parties. In order to coun- ter this tendency, pollsters often give more weight in the analysis to respondents who claim to support the ruling party. Since the students did not employ political weighting, they stated in their presentation of the results at the end of February that they thought the ruling CDU/FDP coalition would improve its approval ratings in Neck and neck: the results of the Freiburg opinion poll in comparison to the official results of the regional election on 27 March 2011. Source: State Statistics Office of Baden-Württemberg/University of Freiburg Graph: qu-int the final weeks before the election and attract enough votes to continue governing, while the Greens were rated too highly. Then came the earthquake and the nuclear catastrophe in Ja- pan. “That cost the CDU a couple of percentage points and put our poll right on target with its prognoses for the Greens and the Left.” This example also demonstrates why Wag- schal believes that election research will be- come more difficult in the future: Traditional al- legiances to political parties are becoming less common, and social milieus are disintegrating; many voters are more willing to change their vote and make their decision more spontaneous- ly. In addition, it is becoming more difficult to conduct representative polls over landline tele- phone connections because young people often only have a mobile phone number. “All of the polling institutes are looking for ways to reach people over the Internet,” says Wagschal. He predicts that social scientists with training in sociology, psychology, statistics, and market ­research will soon enjoy excellent career oppor- tunities in the area of election research, a claim that he also backs up with evidence: The tutor who helped supervise his election research ­project recently landed a job in the strategy de- partment of the national executive committee of the German Green Party. Further Reading Wagschal, U./Grasl, M./Jäckle, S. (2009): ­Arbeitsbuch Empirische Politikforschung. ­Berlin/Münster (Einführungen Politikwissen- schaft 6). Documentation of the project on the Internet: http://www.landtagswahl-umfrage.de 50 % 40 % 30 % 20 % 10 % 0 % CDU Die Grünen official results (27 March 2011) University of Freiburg opinion poll (26 February 2011) SPD FDP Die Linke Andere 39,0 % 41,1 % 24,2 % 24,2 % 23,1 % 22,7 % 5,3 % 6,0 % 5,6 %2,8 % 2,8 % 3,2 % 11