AUSSDA @ PUMA-Symposium 2017


Last week‘s PUMA symposium focussed on the topic of random samples and whether they still act as the gold standard of empirical research. Members of the AUSSDA-team were there to discuss.

The University of Linz organized the third symposium of the platform for surveys, methods and empirical analyses (PUMA), titled “Sampling procedures and representativeness”. AUSSDA – specifically Lena Raffetseder, Otto Bodi-Fernandez and Dimitri Prandner – outlined the AUSSDA services to the 60 attendees in a short presentation, while the scientific debate focussed on the qualitative differences between random samples and other sampling procedures.

Replicability of data

A prerequisite for data to be archived at AUSSDA is that they must be understandable to others, meaning that the circumstances of the data collection process must be clear. This is vital as the quality of empirical social research is closely tied to the data collection and by outlining their methods and considerations, researchers can make the research process transparent. AUSSDA ensures this by archiving accompanying documentation and metadata along with all datasets, allowing valuable insights to the context of each research project.

Quality criteria in survey research

The keynote by Ulrich Kohler, Professor for methods in empirical social research at the University of Potsdam, showed that quality criteria for his field cannot be absolute, but rather are dependent on the framework of each individual research endeavor. By providing various examples, he discussed problems of response rates and representativeness of a sample.  

Two panels with a total of five presentations followed, covering various aspects of sampling procedures, their quality and the weighting process.

Two days before the election: How valid are opinion polls?

Was the title of the closing discussion with Sylvia Kritzinger (University of Vienna, Head of PUMA), the market- and opinion researchers Edith Jaksch (Jaksch & Partner) and Eva Zeglovits (IFES) and Martin Weichbold (University of Salzburg). The discussants covered topics of margin of error, the fear of publishing results diverging from other polls, the black box of the fieldwork phase, weighting, and the – often underrated – role of interviewers. There was agreement that a representative survey cannot forecast an election result; rather, it can show trends and developments.

The participants of the closing discussion (from left to right): Martin Weichbold, Eva Zeglovits, Johann Bacher, Sylvia Kritzinger and Edith Jaksch. (photo: Dimitri Prandner)