Self-selection
Self-selection (also self-selectivity ) is a form of random sampling . In opinion research and empirical social research , one speaks of it when the people examined in a sample are not selected by a random or systematic sampling procedure, but rather make the decision themselves to belong to the sample. Compared to random selection procedures, self-selection means that the sample is not representative and conclusions about the characteristics of the population are therefore only permissible to a limited extent, even if the socio-demographic characteristics of the people examined should agree with those of the population.
Example of self-selection: In a survey on smoking, a random sample of the population would not be specifically written to, but a booth would be set up in a pedestrian zone with questionnaires. Non-smokers in particular might want to participate. The opinion of smokers would be underrepresented.
Self-selection is often used as a selection procedure in online surveys , although it is unsuitable for this. The survey participants are passively recruited, for example by reacting to a banner . Such people are usually particularly interested in the subject of the survey, so their information cannot be transferred to the population of Internet users or website visitors.
Delimitations
Self-selection as a selection procedure must be distinguished from systematic unit nonresponse in the case of random selection. If, for example, in social research a particular group of people systematically fail in an investigation in part (unit nonresponse), there will be systematic non-response and thus distortions (see. Item non-response bias ). These distortions can, however, be corrected if suitable features are known (e.g. by using weighting factors ).
Self-selection is to be distinguished from the general error in which certain people have little or no chance of getting into the sample. An example here would be homeless people who would hardly have a chance of being reached via landline if a telephone survey was carried out.