English title: Analyzing and measuring Event Data Across Europe: a comparison between weekly and monthly event reporting
Author(s): Theoni Stathopoulou -
Type: Conference paper/poster
The collection of media-reported event data for the European Social Survey (ESS) aimed at adding timing and context to the formation of survey respondents’ opinions and attitudes. In an effort to extract the most out of the database created by the first round of the ESS, we proposed a new model for events analysis. This model used a combination of text mining, time series, and cluster analysis techniques to monitor the evolution of the events in time and space and map the impact of events in various countries, thus tracking the formation and flow of “event identities” across the participating countries of the survey. While the event data of ESS Round 1 were collected on a monthly basis, the event data of ESS Round 2 were collected weekly. On the basis of the methodology developed after ESS Round 1 outlined above, this paper will compare the weekly data with the monthly data in order to: 1. study the quality of incoming information on a weekly basis; 2. study the degree to which the weekly information creates clearly defined groups over time as regards content and quality; and 3. study the degree to which the weekly collection of events increases or decreases noise levels in the analysis as compared to monthly collection.
Conference name: First European Association for Survey Research conference
Start date: Jul 18, 2005