Abstract
Turnout probability measurement is one of the fundamen-tal methodological questions of pre-election polls. Accurate pre-election polls are important not only for polling agencies and politicians but also for social scientists.
Among the op-portunities they indeed provide is verifying the accuracy of questionnaire surveys - one of the fundamental social scien-tific methods. Likely voter measurement is a highly empirical discipline, where studies lack in theoretical background and fail to justify their choices of variables that determine turnout.
The article seeks to outline a theoretical perspective on empir-ical turnout research and, based on that perspective, to pro-pose a new likely voter scale. In the first section, it presents the theoretical background, namely the social-psychological reasoned action approach (RAA) to studying the predictors of social behavior.
It provides a brief introduction to the theory's history and the basic concepts with which it operates. In RAA, the intention to behave in some way is the central predictor of that behavior.
A behavioral intention is determined by three concepts: attitude to the behavior, perceived social norm, and perceived behavioral control. Those three concepts arise from specific beliefs.
Beliefs are shaped by so-called background factors, including sociodemographic variables, individual char-acteristics, and information sources.The article elaborates two proposals to extend RAA. First, it deals with the so-called anticipated affect which, however, has already been measured by RAA authors under the experiential attitude.
Second, it discusses the role of past behavior, which explains a significant portion of behavioral variance and has a clear theoretical justification in the concept of habit. The sec-ond section of the article presents all predictors of turnout used in pre-election polls.
Those include intention to vote, voter registration, past voting, interest in election campaign, and many others. The different predictors are discussed in detail as to both their role in RAA and their performance in empirical studies on pre-election poll methodology.
The article demon-strates that researchers have correctly often included in their procedures the intention to vote, and sometimes also past behavior, but have completely ignored the second layer of RAA predictors.The final third section discusses issues of a likelihood-to-vote scale. It starts by present-ing the Gallup-Perry Index as the only publicly tested and developed scale of voter turnout which, however, turns out theoretically ungrounded in terms of RAA.
Another option is to ap-ply a method that was already used in a previous RAA-based study of the likelihood to vote. It, however, paid insufficient attention to all RAA dimensions as currently defined.
Following this discussion, the article proposes a new eight-item scale for measuring the likelihood to vote. By including the second layer of RAA indicators, the scale promises more reliable meas-urement of the likelihood to vote.
Its expected factor structure is presented and offered for empirical testing. A theoretical estimation is made as to which items should be able to explain the variance in voting behavior beyond behavioral intention.