Research studies are distinct events that involve a particular group of participants. However, researchers usually intend on answering a general question about a larger population of individuals rather than a small select group. Therefore, the main aim of psychological research is to be able to make valid generalisations and extend their results beyond those who participate. For this reason, the selection of participants is a very crucial issue when planning research. Obviously, researchers cannot collect data from every single individual from their population of interest, since this would be extremely expensive and take a very long time! So instead they use a small group of individuals – called a sample. The sample is chosen from the population and is used to represent the population. Researchers use sampling techniques to select the participants for their sample – these techniques help to minimise cost whilst maximising generalisability. So, in this weeks blog I am going to be discussing the different sampling techniques and methods, and considering the issue of sampling bias and the problems associated in research.
There are a variety of different sampling methods available to researchers to select individuals for a study. Sampling method fall into two categories:
- Probability sampling: Every individual in the population is known and each has a certain probability of being selected. A random process decides the sample based on each individual’s probability.
- Nonprobability sampling: The population is not entirely known, thus individual probabilities cannot be known. Common sense or ease is used to choose the sample, but efforts are made to avoid bias and keep the sample representative.
Simple random is an example of probability sampling. This is when a list containing all of the population is created and used to obtain participants by random selection. This random selection guarantees that each individual has an independent and equal chance of being selected. This method is very fair, unbiased and easy to carry out. However, with simple random sampling there is no assurance of complete representativeness of the sample. Another example of simple random sampling is cluster sampling. This is when the sample is gained by the random selection of clusters (pre-existing groups of individuals) from a list containing all of the clusters existing within a population. Cluster sampling is often used to estimate number of mortalities in events such as war and natural disaster¹. This method is easy for obtaining a large and relatively random selection of participants, however, the selections lack independence.
Convenience sampling is a method of nonprobability sampling. With convenience sampling, the sample is made up of individual participants who are easy to get. For example, Milgram (1963) used convenience sampling in his famous study ². The participants were individuals who had volunteered by responding to a newspaper article. Convenience sampling is easy to carry out, but one large disadvantage is that the sample is likely to be biased. Milgram’s participants were all male – which could be agued to be a biased sample. Finally, quota sampling is another method of nonprobability sampling. This is when different subgroups are identified and participants are selected through convenience from each different subgroup. For example, say a researcher wanted to select a sample of students to participate in a study using a convenience sample but wanted to ensure that an equal number of boys and girls were selected – quota sampling would be the best method for them to use. This type of sampling can help to control a convenience sample but may results in a biased sample, which would not be a good representative of the wider population.
As I mentioned earlier, the goal of research is to study a sample of participants and then generalise the results to the larger population. How far we can extend such results to generalise to a population is dependant on how closely the sample resembles the population – the representativeness. The main threat to representativeness is bias. A biased sample is one which contains characteristics that are different from those of the population. This bias may happen by chance, but usually is down to selection bias. Selection bias is when participants are selected in a way that increases the probability of acquiring a biased sample. For example, if a researcher recruits participants from a gym, they are more likely to be healthier and fitter than the rest of the general public.
I can definitely say that the selection of participants is a very vital part of planning research. Without carefully planning and choosing an appropriate method for sampling it is very easy to obtain a biased sample that does not represent the population. When this happens, it is difficult to extend findings to a wider population and the validity of the experiment decreases. In order to produce influential and meaningful results, researchers must ensure that they have chosen an appropriate sampling method to select a representative sample of participants.
¹. David Brown, Study Claims Iraq’s ‘Excess’ Death Toll Has Reached 655,000, Washington Post, Wednesday, October 11, 2006
². Milgram, S. (1963). Behavioral study of obedience. Journal of Abnormal and Social Psychology, 72, 207-217.