FORS researcher Michael Ochsner will give an overview of his research soon to be published in the series FORS Guides to survey methods and data management.
Representativeness of Surveys and its Analysis
When fielding a survey, we want to find out some unknown characteristics of the population or how the population thinks about specific topics. However, it is not possible to ask each member of the population to give us the desired information. Hence, instead of asking the whole population, the information is asked only from a small selection of the population under scrutiny. As survey practitioners and data analysts we are concerned about the quality of our sample because if the sample is significantly different from the whole population, statements about the population can be biased. To address this issue, representation analyses are conducted. However, the term and concept of “representativeness” has been challenged in the statistical literature for decades. In this presentation, I will give an overview of the concept of representativeness and the problems involved and suggest a multidimensional approach to the problem of sample quality with regard to inference. Instead of a deterministic concept of a sample “representative for the population”, I suggest examining risks for representation bias linked to the purpose of data use.
The FORS Lunch Seminar Series are open to all researchers.