Analiza „efektu panelowego” w badaniach wyborczych w Łodzi w 1961 r.

Abstract

During the elections to the National Council and to the Seym in the city of Łódź two successive panels were made, each with a twofold interviewing only; the first panel was arranged before, and the other after the elections. Simultaneously two independent control samples were interviewed in order to find out and to measure, by means of a comparison, the panel effect, i.e. the bias brought about by the first interviews. All samples were drawn randomly from the electoral lists. Interviewers were the students, the situation of interviewing being arranged in such a way as to make respondents act in the role of a citizen who expresses, before the representant of the local press, his opinions about the candidates and the authorities to be elected as well as about municipal problems. Two distributions of answers to 6 questions obtained in two panels were compared with those found in two control samples. (The 6 questions preferred to the elementary knowledge about the authorities to be elected as well as the evaluations of those whose term had expired.) It turned out that the first interviews caused a considerable increase of knowledge, interest and “good” evaluations in the answers obtained in the second ones. This biasing effect of the first interviews was interpreted and analysed as a result of an unintentional “agitation” influence exerted by the interviewers under specific psycho-social conditions created by the political system and the character of the election campaign in Poland. These are some findings significant at the 0,95 level: The increase of knowledge about the authorities to be elected (a difference of 16% of correct answers among men and 22% among women) was accompanied by the fall in the percentage of “don’t know” answers, while the percentage of the erroneous ones remained unchanged. The increase of the declared interest in the elections was found among men only (11%) and that concerning the names of candidates appeared only among women (9%). The increase of "good” evaluations in the post-election panel went along with the fall of “very good” ones. This was explained hypothetically as a result of the drop in the pressure of electioneering propaganda. The comparison of answers to the question on the degree of approval of the aid to Cuba (invaded at that time by the counter-revolutionary forces) revealed the lack of any effect of the first interviews. This result seems to indicate that this effect appears only within the scope of themes associated closely with a specific social role in which act the interviewees.