Sampling method

If a probability method is chosen it will involve some kind of random sampling. The benefit of using a probability method is that the results of the survey can be subject to statistical analysis, the most important being the estimate of sampling (standard) error. The researcher may wish to state the precision of the measures obtained, i.e. the confidence interval. (In newspaper reports of opinion polls, for example, an error of plus of minus 3% is often quoted).

Random sampling

This may take the following forms:-

Simple random sampling, where every unit in the population has a known and equal chance of being chosen. Computer generated random numbers are useful here.

Stratified sampling, where the population is divided into strata and each stratum is sampled separately. This occurs where the researcher anticipates important differences between variables within the population. For example, if the population consists of organisations the researcher may want to make sure that manufacturing, service, retail etc. organisations may be sampled proportionately to the number of that type in the population as a whole, or the researcher may wish to 'over-sample' certain strata, giving a disproportionate stratified sample.