Sampling Bias Background

Differences between a sample and the population it represents should result only from random chance. When differences arise for reasons other than chance, you have introduced sampling bias into your research. Administrative convenience may tempt you to use the major source of bias: nonprobability sampling techniques. Here are descriptions of nonprobability sampling techniques:

Accidental Sampling

Weakest of all sampling procedures, it involves using what is available, and most convenient as a sample pool. Interviewing the first individuals out the door after a training session, using classmates that you know real well, or asking for volunteers to take part in a study, are examples of accidental sampling. In each case there is no way of estimating the probability of any member being chosen.

Judgment Sampling

An "expert" selects a sample based on her expert judgment. We assume that the expert can select elements judged to be typical, or representative from the population. The critical question is the extent to which judgment can be relied on to arrive at a typical sample. We hope that errors in judgment counterbalance one another, but how can we check an expert for systematic bias?

Quota Sampling

If a population has diverse segments within it, and you want to ensure that your sample reflects this diversity, you could use quota sampling to select typical cases. The quotas are based on known characteristics of the population. For example, if you conduct research that focuses on the typical computer user and marketing data indicates that 10 percent of personal computer users prefer the Macintosh operating system, then 10 percent of the sample should be Macintosh users. The steps in quota sampling are:

1. Determine the important variables that characterize your population. Variables such as sex, age, income, and education are typical ways of stratifying your population.
2. Use external information, such as census data, to specify the size of each segment of the population.
3. Compute quotas for each segment of the population that are proportional to the size of each segment.
4. Select typical cases from each segment of the population to fill the quotas.

Step 4 is the weak point in quota sampling. How do you know that the individuals chosen represent the given segment? Going back to our operating system question, how different would the sample be if Macintosh users were chosen from the instructional media lab of a large university versus people that have recently purchased Macintosh computers from an electronics store?

On to sampling activity.