Nonprobability Sampling

Non-probability Sampling

A. Nonprobability sampling plans are all those which do not meet the requirements of probability sampling; they result in samples in which all elements do not have an equal chance of selection. These plans are used when the researcher does not have the resources (time, money) to develop probability samples. They are considered inferior but often necessary. It is true that some nonprobability plans result in good samples; many result in poor samples.

B. Types of nonprobability sampling plans:

1. Availability sampling - selection of elements as they present themselves. This often results in a self-selecting sample, one of the worst types of samples and often very biased.

Examples: survey which collects data by leaving questionnaires in a public place and instructs respondents to fill it out if they wish; survey in a magazine.

2. Judgmental sampling - (this is a little confusing, as the text uses the term purposive sampling in discussing this type; I am reserving that term for a different type, see 5. below) selection of elements based on knowledge of the population. This would include the use of "matched pairs" which are often used in experimental and quasi-experimental designs. This is discussed in the text as "purposive or judgmental sampling" but I am using "purposive" differently (see below).

Example: selection of one-half of the clients in an agency which you know to be representative of the population based on their demographic and situational characteristics.

3. Quota sampling - accidental or purposive selection of elements in order to fill quotas based on stratifications in the population. "The poor man's stratified random selection." This is usually considered a better type of NPS and may result in an effective sample.

Example: selection of twenty-year old, female accounting major to fill quota in survey of MTSU students, etc. with other types.

4. Snowball sampling - selection of elements based on information gained from preceding elements. These are used most often in studies of deviance where population is difficult to access.

Example: interviews with victims of spouse abuse: find out about second respondent from first, etc.

5. Purposive sampling - selection of every nth element.