sampling

In statistics, taking a small number of cases and assuming that they are representative of the total number. In survey work in market research, for example, it is usually not practical to survey every customer. It is also important in quality control to avoid the expense of testing every item produced. Instead, a sample is taken. A random sample is one where the sample is chosen in a random way.

For example, in a postal survey, one person in every hundred on the electoral register might be chosen. A stratified random sample is one where a group is selected and then a random sample is taken from that group. For example, a manufacturer of baby-care products might select pregnant mothers as a group, and then conduct a random-sample survey on that group. A quota sample survey is one where the size and composition of the survey matches the total population that is being researched. For example, a psychologist may want to interview 20 boys aged 0-10, 20 girls aged 0-10, 20 males aged 10-20, and so on.

Choosing a sample The sample size is important. It should be large enough to be representative of the whole group without becoming unmanageable. However, the larger the sample used, the more accurate the results will be.

The sample must also be a fair representation of the group as a whole, otherwise the data collected will be biased. For example, in a survey of the general population, the proportions of men, women, and children within a chosen sample should reflect the proportions in the whole population. This is known as stratified sampling.

If there is no information available about the characteristics of a group, random sampling is usually employed, where all members of the group are equally likely to be chosen. With random sampling, several samples should be taken and the results averaged in order to improve accuracy.