Censuses, Random/Systematic/Stratified/Quota Sampling
If you need to find peoples' opinions you need to ask them questions. You could ask everybody. In the United Kingdom there are 60 million people. It is a big task to ask everybody. A task this big can only be done when the issues involved are major, so a census is taken every ten years, and is used for a wide variety of purposes, from forecasting the need for school places to gauging the changing attitude of society towards sex or forecasting how long it can be before enough people have internet access so many government services can be accessed online. If you want to find out people favourite ice cream you would not carry out a census. You might choose instead to ask every tenth person coming out of a shopping centre. You would in this circumstance be taking a sample.
You would take a census if:
The data needs to be comprehensive and accurate
The population is small so that not to much effort and time is needed to ask everyone for example, if the population is that of a small village or a class of 30 pupils.
You would take a sample if:
A census is impractical because of limited time and resources.
The data does not need to be comprehensive and a small margin of error is acceptable. This is the case for example for political opinion polls which tend to be about 1000 – 2000 in size and with a typical margin of error of 2%.
A population is made up of many different segments, but at any one time everybody of every type is somewhere doing something. The population is made up of “sample frames” one or some of which we must pick from to undertake a survey. For example between 9am and 10am, everybody is at work, home, school or somewhere else. Each of these is a sampling Fame.
Each member of the sample frame has an equal chance of being selected.
Each individual is given a number. Random numbers are then generated – by drawing numbers from a hat, using a computer or from a table - and those members of the population whose numbers come out are sampled.
Instead of choosing the members to be sampled using random numbers (which might be difficult and time consuming for large populations), systematic sampling uses a simple rule to choose people. For example, every 10th member of the sample frame -could be selected.
Stratified sampling can be used when the population in question is split up into groups with different patterns of behaviour. For example, if we were trying to find the nation"s favourite radio programme, most children would probably like different stations or programmes to most adults -.
Each group is sampled separately and the results are put together.
In the example given above, if children make up 25% of the population, we would make sure 25% of the sample would be children.
Quota sampling involves splitting the population into groups and sampling a given number of people from each group.
Market research is typically based on this method. For example, if someone is interviewing people at a shopping centre, they may have been told to interview 50 men and 50 women. Once the quota is filled, people from that sampling frame stop being interviewed. It doesn't matter how they choose the 50, as long as they interview that many.
If there is no sampling frame (list of sampling units), the above sampling methods can't really be implemented. Quota sampling might be the only real possibility.