Quota sample
The quota sample (Quota samples) is a special systematic sample .
Quota samples are not random samples, but are based on a conscious selection of target persons. In the case of a quota sample, an attempt is made to create a representative composition of the sample by defining quotas for certain characteristics, the distribution of which in the population must be known. The interviewers are given precise specifications as to which characteristics the interviewees must have.
The quality of a quota sample can be checked by comparing the distribution of unquoted characteristics in the sample and population statistics that are known for the population. For example, if the characteristics age , gender and place of residence were used for a quota sample based on census data , the additional characteristic denomination can be used to determine to what extent the sample corresponds to the population.
Comparison of quota sample and stratified random sample
The quota sample is to be distinguished from the stratified random sample . With the quota sample, certain characteristics are drawn until the desired quotas are reached. With the stratified random sample, the drawing proportions of certain characteristics are determined in advance and then drawn at random. The stratified random sample thus has a definable drawing probability . In the case of the quota sample, no drawing probability can be specified for the quota characteristics; the drawing depends on the willingness to participate in previous cases.
application
Quota samples are used in the social sciences and especially in psychology . One reason for using the quota sample in survey research is the low response rates for random samples, which are now below 10 percent in telephone surveys in the USA. In a quota sample, those who refuse to interview are replaced by so-called statistical twins, which have the same quota characteristics. This is based on the assumption that statistical twins are also similar in the non-quoted characteristics (see above). In this way, there is no (visible) sample failure with quota samples , which can significantly distort the results with random samples.
Advantages over simple random sampling
Suppose you want to compare the monthly income of students in a pedagogical degree. A simple random sample would be unsuitable here, since courses with an educational focus usually only have a small proportion of men. If the study participants were selected at random, more women than men would be surveyed. Under certain circumstances, no tenable statements can be made (if the proportion of men in the degree program is 10%, for example, only 10 out of 100 randomly selected students would be male). Instead, a quota sample can be used, in which the respondents are selected in roughly equal parts based on a certain characteristic (in this case gender).
Scientific discussion about the quota sample
The scientific nature and usefulness of quota samples have been controversial in social science for decades.
Critics and opponents of the procedure base their rejection on several arguments. The first argument is that the math of random sampling does not apply to quota sampling. Thus it is also not permitted (but nevertheless technically possible) to calculate key figures for the quality of an examination such as confidence intervals . Another argument is the influence of the quality of the interviewer's work on the result: The choice of the specific target person is left to the interviewer in the quota sample. This procedure gives the interviewers a great influence on the course of the study. Only if they work correctly does the sample meet the quality criteria. Since the interviewers are paid for each interview carried out and since a follow-up check is difficult because of the protection of the anonymity of the interviewees, quota samples are prone to manipulation. In addition, the procedure increases the likelihood of being selected for people who can be reached easily, or people who absolutely cannot be questioned are not reached with the quota sample any more than with a random sample.
Users and supporters of the quota sample, on the other hand, argue with the results in research practice, which are at least as good as those obtained with random samples. In view of the decreasing response rates for random samples, an improvement in representativeness should also be sought by avoiding random samples.
Under certain conditions, quota samples are faster and - depending on the survey mode ( by telephone , in person , post , online ) - more economical than random samples. This is especially true if there is no list of people to be interviewed from which a random sample could be drawn. The quota specifications also restrict the interviewer's discretion to such an extent that a random selection is approximated.
Individual evidence
- ↑ people-press.org
- ^ R. Schnell, PB Hill, E. Esser: Methods of empirical social research. 8th, unchanged. Edition. Oldenbourg, Munich 2008, pp. 306-317. In more detail on the subject of sample failures: H. Proner: Is no answer an answer? Participation in political polls. 1st edition. VS-Verlag for Social Sciences, Wiesbaden 2011.
- ↑ z. B. the overview with Karl-Heinz Reuband: Quota and random samples in the practice of social research. Similarities and differences in the social composition and response patterns of the respondents. In: ZA-Information. 43, 1998, pp. 48-80.
- ^ R. Schnell, PB Hill, E. Esser: Methods of empirical social research. 8th, unchanged. Edition. Oldenbourg, Munich 2008, p. 303.
- ^ Dieter Roth: Empirical election research. Origin, Theories, Instruments and Methods . 2nd Edition. VS Verlag, Wiesbaden 2008, ISBN 978-3-531-15786-3 , p. 68 f .
- ^ R. Schnell, PB Hill, E. Esser: Methods of empirical social research. 8th, unchanged. Edition. Oldenbourg, Munich 2008, pp. 302-304.
- ↑ On this problem cf. also F. Newport: Presidential Address: Taking Aapor's Mission To Heart. In: Public Opinion Quarterly. 75 (3), 2011, pp. 593-604.
- ^ R. Schnell, PB Hill, E. Esser: Methods of empirical social research. 8th, unchanged. Edition. Oldenbourg, Munich 2008, pp. 302-304.