Notes > Foundations of Computing > Sampling 
Taking a sample of a population within a survey is a very useful technique in representing the whole population without gathering information from every single individual. There are various types of sampling methods which are outlined below. Simple random sampling and stratified random sampling are both examples of random sampling.
Simple random sampling is where members of the population are selected completely randomly so that they all have an equal chance of being selected. Random numbers can be generated by computers to choose which items within a population are selected. This method of sampling is a relatively simple method and is particularly suitable for small populations.
Simple random sampling requires a complete "sampling frame" which means all the members of the population need to be known and identifiable. A sampling frame is the structure which enables information about all the population to be stored. For example, a student database is a sampling frame and also, personal records held by a business about its employees is another example of a sampling frame.
Stratified random sampling is where the population is split up into groups or "strata" and then simple random sampling is carried out within each group or stratum. Each item within the population is classed into one group only. Stratified random sampling can give a better overall representation of the population compared to simple random sampling especially if there are well defined strata already existent in the population.
Systematic sampling involves selecting items at specific intervals within the list of the whole population. Systematic sampling is a partially random sampling technique. For example, every 10th record starting from the 5th record in a student database would be an example of systematic sampling. The first record is chosen at random but the rest are chosen systematically which could of course introduce the possibility of the results being biased in some way as the sample would not be fully random.
Nonrandom sampling may be necessary if taking a random sample is too expensive or will take too much time. Two examples of nonrandom sampling are Cluster sampling and Quota sampling.
Cluster sampling involves splitting a population up into segments such as by geographical area or by households for example. A random selection of these clusters can then be used and random sampling can be carried out within them or alternatively, every item within each cluster can be used. This method does contain elements of random sampling at the bottom level but overall it is a nonrandom sampling method. It is useful for where a population is spread over a very large area as it would be very costly to try to take a fully random sample of households spread out across a whole country. Unfortunately there is a risk of an increased sampling error with this technique.
Quota sampling involves a certain number of people who have particular characteristics being interviewed. This requires good control of the interviewers being used to ensure that the questions are asked properly and valid answers are gained. This method is ideal for getting results quickly but it cannot give a fully representative picture of the population.
Search for "Sampling" on:
Google 
Kelkoo 
Amazon 
eBay (UK) 
eBay (US)
Search for "Sampling" on the rest of Computing Students: Sampling
