Sociologie Românească (Romanian Sociology), X new series no. 4/1999, pp. 48-76.
Despre măsurarea intenției de vot în sondajele de opinie
How to Measure Vot Intention in the Survey
Abstract (Rezumat în limba engleză): The bulk of this paper is dedicated to investigate the causes and effects of non-responses in the political surveys. I used 27 data bases provided by different Institutes and private companies to prove that the predictions on the results of elections are better if there are taking in account not only the estimations from the vot item but also the characteristics of the survey by itself. Romanian mass media noticed that different surveys conducted in the same period of time, are producing different estimates of the vote intentions. Usually, the "disfavored" political parties are incriminating of malevolence the agency that was collecting the data, while the public opinion formulates doubts about the probity of this agency. In this paper, I am arguing against this point of view using theoretical explanations doubled by empirical evidences. My first point is that the proportion of the people with "no declared vote intention" varies among surveys as it is determined by the position of the vote intention item in the interview. On its tum, this is determined by the type of survey (focused or not on evaluations of the political process and its actors) and the length of the questionnaire.