期刊名称：Drustvena istrazivanja. Journal for General Social Issues
印刷版ISSN：1330-0288
出版年度：2000
卷号：9
期号：4-5 (48-49)
页码：601-629
出版社：Institute of Social Sciences IVO PILAR
摘要：ln this work the authors present the findings of the validation analysis of results concerning voting intentions at the 2000 presidentiai elections, gathered in two separate telephone surveys on the eve of the first election round and run-off. Based on the results of the analysis a general prognostic validity of the research was established in relation to the official electoral results and so was the influence of certain modalities of gathering and processing data on the validity of the results obtained. The results indicate that, in our current circumstances, the main issue of the validity of pre-election telephone surveysis connected with a relatively large proportion of noncommitment regarding voting intentions (in the research analysed it was 29.8 or 25.6%), thus making it necessaryto correct the original results through different reduction procedures. Therefore, correcting results in two steps has proven the most suitable method, based on additional questions on inclinations towards certain electoral options (this procedure has, in both electoral situations analysed, considerably improved the precision of results and reduced a certain partiality in their main distribution), while proportional conversion by simple elimination of non-committed examinees, has not given satisfactory effects. In the data-gathering procedure, repeated attempts regarding the currently unavailable examinees have contributed to the overall precison of the research, while the differences in the distribution of results according to daily time intervals, have, in this case, not appeared statistically significant.
