One method sometimes used to reduce the time taken to locate a random sample is to choose every tenth or twentieth name on a list. This is known as systematic sampling. It is, however, less random.
Stratified Random Sampling
This method of random sampling is often preferred by researchers as it makes the sample more representative of the whole group. The sample is divided in segments or strata based on previous knowledge about how the population is divided up. So, if the business was interested in how ‘class’ affected consumers’ demand for a food product, it might divide the population up into different class groups, such as working class males, middle class females etc. A random sample could then be chosen from each of these groups by making sure that they were the same proportions of the sample in each category as in population as a whole. So if the population had 10 per cent upper class males, so would the sample.
Quota Sampling
This sampling method involves the population being segmented into a number of groups which share specific characteristics. These may be based on the age and the sex of the population. Interviewers are then given targets for the number of people out of each segment who they must interview. For example, and interviewer may be asked to interview 10 males between the ages of 18 and 25, or 15 females between the ages of 45 and 60. Once the target is reached, no more people from that group are interviewed. The advantage of this sampling method is that it can be cheaper to operate than many of the others. It is also useful where the proportions of the different groups within the population are known. However, results from quota sampling are not statistically representative of the population and are not randomly chosen. They must therefore be treated with caution.
Cluster Sampling
This involves separating the population into ‘clusters’, usually in different geographical areas. A random sample is then taken from the clusters, which are assumed to be representative of the population. This method is often used when survey results need to be found quickly, such as opinion polls.
Multi-stage Sampling
This involves selecting one sample from another sample. So, for example, a market researcher might choose a country at random and then a district of that country may be selected. Similarly, a street within a city may be chosen and then a particular household within a street.
Snowballing
This is a highly specialised method of sampling. It involves starting the process of sampling with one individual or group and then using these contacts to develop more, hence the ‘snowball’ effect. This is only used when other sampling methods are not possible, due to the fact that samples built up by snowballing cannot be representative. Firms operating in highly secretive businesses such as the arms trade may use this method of sampling. Similarly, firms engaged in producing highly specialised and expensive one off products for a very limited range of customers may need to rely upon snowballing when engaged in market research. Examples might include firms engaged in nuclear and power generating industries.
