Statistical sampling

Population

The entire set of items of interest

Sample

A selection of observations taken from a subset of the population

Sampling unit

Individual members or items in a population

Sampling frame

A list of all the units in a population

Census

A survey of the entire population

Discrete data

Data that can take certain values within a range, often integer values

Continuous data

Data that can take any numerical value

Qualitative data

Data that have numerical values

Quantitative data

Non-numerical data

Advantages of a sample

• A sample requires fewer resources than a census, i.e. less time and money. A census can be very time consuming and costly

• Results are obtained more quickly since fewer units of the population are observed

• Overall, the data processing is easier because a sample generates a smaller volume of data

• A census cannot be used when observation of a unit would result in destruction of that unit

Advantages of a census

• A census is appropriate if a high degree of accuracy, or a completely accurate result, is required for the entire population

• With a census there is always enough data to gain information about sub-sections of the population. Certain sampling techniques also ensure this is the case

• If the population is small, a census is often practical, since the resources are smaller than they would be for a census of a large population

Advantages of systematic sampling

• It is simple to use

• It can be used for large populations and large samples

Disadvantages of systematic sampling

• It is only random if the ordered list is truly random

• It can introduce bias

Stratified sampling

A form of random sampling in which the population is divided into groups of strata, which are non-overlapping, so that no unit can appear in more than one group. The groups would be decided using one or more criteria

When is stratified sampling used?

• When the population is large

• When the population divides naturally into non-overlapping groups

Advantages of stratified sampling

• It can give a more accurate result than simple random sampling

• Each group within the population is fairly represented within the sample

Disadvantages of stratified sampling

• Within each stratum any of the disadvantages associated with random sampling may occur

• Care must be taken to divide the populations into well-defined, non-overlapping groups

• There is slightly more work involved than with a random sample