Under most circumstances, a population cannot be portrayed as a whole in a survey, simply because it is too large or too difficult to access. Sampling procedures are therefore a common investigation tool in market, public opinion, sales and empirical social research.
MARMAS selects its representative samples using the following methods: quota sampling and area sampling, address random sampling, as well as the master sample of the ADM Working Group of German Market and Social Research Institutes e.v.
All representative surveys also aim at the production of a simple random sample. If the initial approach only aims at institutions such as companies and households, subjects have to be selected using more than one step. Only during the evaluation process, the sample of the institutions can be “transformed” into a simple random sample.
The level of accuracy of these transformation processes can vary: It depends on the degree of difficulty in regards to determining who should be selected in the institutions. MARMAS focuses, among other things, on the examination of multi-level research approaches that aspire to statistically represent difficult target groups.