In any research that aspires to obtain information on large populations, the basic objective in selecting respondents is to obtain a representative sample of people. ‘Representative’ means that the sample of selected respondents (usually approx. 1000) is a reduction of the population as a whole (e.g. of the 9 million inhabitants of the Czech Republic over the age of 15), and as such it does not differ from the larger population in any of its parameters (e.g. age, education, political preferences, distribution of opinions on crime), with one exception, and that is its size.
The Public Opinion Research Centre selects respondents using ‘quota sampling’, a method that requires prior knowledge about the composition of the total population. We can find that information in the results of the regular Population Census, thanks to which we know, for example, that approximately one-half of the population is male and one-half female, that 11% of Czech citizens live in the Moravia-Silesia Region, and that 34% have upper secondary education. Having determined the structure of the population from this information, we then convert it to a target population of 1000 respondents (in such a sample there will then be, for instance, 500 men and 500 women), and these respondents are divided up among approximately 250 interviewers. The interviewers are selected for the research from the Public Opinion Research Centre’s interviewer network, which is at present made up of around 350 people. By selecting the interviewers while still in the planning stage of the research, the Public Opinion Research Centre is able to ensure that the number of interviews conducted in every region and size category of municipality in the Czech Republic is, in conformity with the principle of representativeness, proportional to what part of the total population that region or municipality represents. An interviewer who is to conduct, for instance, 4 interviews, will then be instructed to conduct 2 interviews with women and 2 with men, to conduct one interview each with a person with basic, lower secondary, upper secondary, and university education, carry out 1 interview with someone in the 15–25 age group, and so on. The interviewers then contact people they do not personally know in the area they live in (which ensures adequate distribution across regions and municipality categories), and if they find a person with the required characteristics (and the person agrees) they conduct an interview with him or her.
Once every interviewer has completed interviews with the designated number of respondents and in conformity with the prescribed characteristics of sex, age, and education, then the resulting set of all respondentswill almost exactly match the composition of the population as a whole. It is expected (and to some degree this has almost been confirmed in research) that if the sample of respondents does not differ from the total population in these basic demographic characteristics, then it will resemble the total population also in every other parameter; and the information obtained from this kind of sample can therefore be regarded as valid for the population as a whole. It is possible to claim with sufficient reliability that, for example, the level of trust in politicians or political institutions measured on a representative sample of people at a particular point in time is also valid for the entire target population of citizens of the Czech Republic.
Quota sampling is just one way of obtaining a representative sample of respondents (the other most commonly used method is random sampling) and in special situations there are also surveys that do not need representative samples. Public opinion polls, however, almost always belong to the type of survey for which a representative sample of respondents is crucial.