Conferences

Societal events are events like elections, significant changes in laws, demonstrations, but also extreme weather conditions. All such events can have an effect on a society and, consequently, influence the attitudes of its population. The assumption for surveys is that respondent behavior or answers to some questions are influenced by significant societal events in various areas. When conducting a survey, the impact of an event must be considered and, whenever possible, controlled especially in survey projects in which different countries participate. However, manual identification of significant events that occur during the data collection phase is a very error-prone and time-consuming task. Therefore, we have developed a procedure to identify events using a combination of two different approaches of the (quantitative) computer-assisted content analysis: the reference text technique and the statistical association approach. On the basis of distinctive features of word usage in a so called reference text corpus and in newspaper texts of a specific time period in which events should be located, words are selected and classified by means of an exploratory factor analysis. The identified factors used as indicators for societal events. Our procedure offers a new opportunity for research in survey-based practice. I will demonstrate this procedure by means of an example in which we will identify events automatically in a specific time period. As a final point, I will discuss the advantages and the restrictions of such an automatic technique to identify events. Considerations regarding the composition of the text base, language specific problems (for example, lemmatization, or the role of verbs), the selection of the final events (for example, number of events per week, weeks with no events, etc.) will be presented.