The statistical population refers to the total set of observations or entities in question. The term universe can also be used here. Although it is ideal to be able to gather information on the entire population, it is not always feasible or sensible. Some reasons why this is so:

• the cost and time involved may be enormous
• impractical - complete set of observations may not be available or accessible.
• impossible - population may be infinite in number
• quality - act of collecting information may degrade quality of material obtained (eg. a long questionnaire may lead to hurried results)
• unnecessary - precision of a sample may be adequate for the intended objectives

A sample is a subset of entities or observations in a population. It is intended to be representative of that population. The relationship between the samples and their parent populations underpins the theory and practice of inferential statistics. When a statistic is calculated and obtained for that sample, it is an estimate of the population.

In the process of sampling, a number of issues arise:

• How do we choose a sample that represents the population?
• What method do we use to identify and access the samples?
• How many samples do we choose?

The key to making sampling decisions...

The understanding of the relationship between populations and samples will help you to make decisions regarding sampling.

We will examine some of the sampling issues in more detail on the next page.

Don't forget...

When we refer to a "population" we tend to think of "people". Don't forget that a "statistical population" can refer to any phenomena under investigation... people, roads, animals, plants, mineral ore bodies, oil and gas refineries, local government areas, soils, etc. etc.