Kontrola greške u istraživanjima javnog mnjenja

Abstract

This article deals with researchers' ability to spot, measure and compensate errors made in survey investigation of public opinion. The error as a general concept is recognized under six commonly used terms: error, bias, variation, standard deviation, standard error and random error. The author relates the production of errors to eight sources that could be controlled in this way or the other. These sources are: the operationalization of basic concepts, the questionnaire, the interviewer, the respondent, external conditions, the sample design, the computer and the interpretation of the results. The conclusion is that reliability, objectivity and precision could be achieved in this type of investigation if all parts of the research process are treated responsibly and professionally.