Keywords: Electoral polls in emerging/incomplete democracies

Electoral forecasting remains a substantial source of trust to public opinion polls . Researchers all over the world invest time and efforts into understanding how people come to decision on participation and how they make a choice. However, the problem remains acute and regularly tops the agenda last years. The presentation will address the question how to estimate turnout and voter’s preferences in non-competitive electoral systems.
The polls tend to overestimate the turnout and to lower this effect various techniques have been tested during different election campaigns in Russia. A series of experiment’s data will be presented, including the results of using “item-count technique” and data from a probability-based national-representative sample of respondents, who were interviewed 7 times prior to national parliamentary and presidential elections and after those to track change in voting intentions and electoral preferences.
Another challenge for incomplete democracies is the loyalty bias, when support of opposition is underreported. A range of techniques have been employed during 2015 regional elections and 2016-2018 national elections to reduce this effect. This included experiments with wording, survey mode and a CATI poll on election day in parallel to traditional ballot-box exit-poll.
A comparative efficiency of questions to provide a range of turnout and election result estimates will be shown as well as an impact of survey mode (face-to-face, CATI, online) to answers on voting intentions.