Section 1: Sampling
From Research Methods in Psychology
WHAT ARE YOU STUDYING? SAMPLING IS ALL ABOUT OBTAINING A SAMPLE FROM A POPULATION. IN THIS SECTION WE WILL CONSIDER THE APPROACHES THAT CAN BE USED BY PSYCHOLOGISTS TO TRY TO MAKE SURE THAT THE SAMPLE IS AS REPRESENTATIVE AS POSSIBLE. AT THE END OF THIS SECTION, WE WILL LOOK AT HOW SAMPLING ACTUALLY WORKS IN PRACTISE.
|
Crucial Concept: Sampling is the process of selecting research participants from the population. |
In psychology, we want to answer questions about, or we want to now about all five year olds, or all children at school in a particular area, or all people who work in similar jobs, or all pigeons. To find out accurately what we want to know, the ideal way would be to measure every individual in the population. Obviously, this is not possible, so instead of measuring everybody we decide upon what is a manageable number of individuals to measure, and measure them.
The people (or pigeons) we are interested in are called the population; the people (or pigeons) we actually measure are called the sample. We hope to find a sample that is representative of the population. A representative sample will contain the same kinds of individuals, and in the same proportions as the whole population.
In this section, we will look at some of the ways of selecting your sample from your population of interest. There is not necessarily only one correct method to use - different methods have different advantages and disadvantages. At the end of this section, we will discuss some of the techniques, in particular those that don’t fit neatly into one of the standard categories.
Contents |
[edit] Random Sampling
|
Crucial Concept: A random sample is a sample in which every member of the population has an equal chance of being selected and that selection of one member of the population does not alter the chances of any other person being selected. |
Random sampling is the best method of selecting your sample from the population of interest. It is also the most difficult to achieve. Strictly speaking, statistical analyses are only valid if random samples are drawn. However non-random samples, as long as they are not biased in any way can be a reasonable alternative.
True random sampling is very rare, a perfection which can never be quite achieved. If someone says that taken a random sample your first instinct should be not to believe them. If you write in a report that you took a random sample, you almost certainly did not. The ONLY way to get a random sample of a population of people in the British Isles is to find out the name of every person in the population of interest (and if your population of interest is British adults, that is quite a large population), write all their names on individual pieces of paper (or do some computerised analogy), put all the pieces of paper into a hat, and pick a sample out. If you don’t follow a procedure like this one, you haven’t taken a random sample. The same rules apply when you are sampling from some smaller population. Students sometimes write in their reports that they ‘used a random sample of people in the canteen/street/laundrette/library.’ However, they almost certainly did not.
|
Crucial Tip: Always be careful when you use the word random in your practical reports. It has a specific meaning, and is often used to describe things which were not, strictly, random. |
What Looks Like Random Sampling, but Isn’t Each of the following examples contains a description of a method of sampling, which appears as if it might be random, but is actually non-random. Why are each of these non-random?
- To obtain a random sample of Poppleton University students: wait outside a lecture theatre, and pick every third person coming out.
- To obtain a random selection of adult residents of Uttoxeter: use the electoral register, put each name on a piece of paper, and pick names from a hat.
- To obtain a random sample of residents of Camber: go to every third house in Hogwart, and select one person from each house.
- To obtain a random sample of people who shop at Tesco: spend all Saturday at Tesco, rolling a dice every time someone goes past a checkout. If the dice comes up 6, select that person.
- To obtain a random sample of children in Hogwart’s school: ask each teacher to write the names of all the children in their class on a piece of paper, and select three names from each class from a hat.
Because obtaining a random sample is either unfeasible, or prohibitively expensive and time consuming, a range of other techniques are used, which will (hopefully) also give a representative sample from our population.
[edit] Systematic Sampling
|
Crucial concept: a systematic sample involves taking every nth individual from a list of names. |
Systematic sampling, like random sampling, requires a list of all of the members of the population, but is easier to do. To take a systematic sample, you list all of the names of people in the population, and then decide upon a size of sample you would like. By dividing the number in the population by the number you want in your sample, you get a number which we will call n. If you take every nth name, you will get a systematic sample of the correct size. If, for example, you wanted to sample 150 children, from a school of 1500, you would take every 10th name. This is much easier than random sampling if you have a list of names. However, a list of names is not always available.
[edit] =Quota Sampling
| Crucial Concept: A quota sample is an attempt to make a sample representative by having the same proportions of different groups of people in the sample and in the population. |
When you cannot take a random sample of the population, but you want to make your sample as representative as possible on a small number of variables, a quota sample can be used. You may know the proportions of people of different ages, salaries and sex in the population of interest. To make your sample to be as representative as possible, you want to reflect the proportions of the characteristics in the sample population.
|
Box : Quota Sampling at Work Scenario A: Thursday afternoon, nothing to do, wandering around the city centre, when you see a market researcher. You amble by, looking like you didn’t have a care in the world. They look through you. You stroll past again, and are ignored. You try and do someone a favour, and they don’t appreciate it. Scenario B: Tuesday morning, job interview. Not sure of the exact location, and a little anxious to make sure you’re not late. See market researcher ahead - walk briskly ahead, looking concerned and busy, but they head towards you. “Excuse me, could you spare a couple of minutes…” Do you look like you could spare a couple of minutes? The market researcher is driven by a quota sample, they must get the correct number of each kind of person. If you are not the person they are looking for, they will ignore you. If you are, they can go home, if only they can get to interview you. |
[edit] Cluster Sampling
|
Crucial Concept: a cluster sample takes a naturally occurring group. |
Many populations occur in clusters, and these clusters often contain a range of types of individual. It may be a safe bet that each of those clusters contains a fair mixture of the kind of people who are to be found throughout the population, and it will be much easier to study one cluster, than to take one sample from many different groups. Examples of clusters may be all children in a class, all GPs in Derbyshire, all members of a WI group, or all people drinking in a pub.
[edit] Opportunity Sampling
| Crucial concept: an opportunity sample is a sample of individuals who happen to be available. |
An opportunity sampl;e is obtained by asking members of the population of interest if they would take part in your research. At some point in your university career, students asking if you would mind filling in a questionnaire have probably accosted you. These students were asking you to be part of an opportunity sample.
[edit] Volunteer Sampling
|
Crucial concept: Volunteer sampling involves selecting for research those people who put themselves forward. |
You don’t have to hang around very long anywhere psychology is taught before you see requests pinned to noticeboards for participants to take part in experiments. Sometimes people go into lectures to request that you take part in their research. Both of these methods of recruiting participants are examples of volunteer sampling. Psychologists may sometimes advertise in local newspapers or radio for participants. You may have come across Milgram’s work on obedience to authority (Milgram 1974/1997) - he used a newspaper advertisement to attract participants.
[edit] Comparing Opportunity and Volunteer Sampling
Opportunity and volunteer sampling appear to be quite similar things, and some textbooks will put them together (calling them opportunity sampling), however I think that it is worth considering them separately. The type of person who volunteers for a research project may be different from the type of person who would be willing to take part in a research project if asked.
[edit] Snowball Sampling
| Crucial Concept: A snowball sample is used when a participant is used to introduce the researcher to further participants. |
Snowball sampling is a very different sampling technique. It is a networking technique used when the population of interest is very difficult to get hold of, but it is likely that one member of the population of interest will know other members of the population, who can be introduced to the researcher. Although the sampling technique is far from ideal, for some types of sample it is not possible to approach people or to advertise for individuals who would make suitable participants. Cusick (1998) studied drug use amongst prostitutes in Glasgow. To recruit her sample, she used a snowball technique. Cusick would interview one respondent, who would then introduce her to another potential respondent. In this way, Cusick was able to gain the trust of the respondents more quickly than if she had approached them individually without an introduction from another member of the group.
[edit] Sampling in Practice
Sampling techniques are usually described in textbooks (including this one) as if only one technique is used at a time, In fact, in any one study, psychological researchers usually use a combination of techniques. If you want to study children in a particular school, you cannot pick the children randomly from the school population. There are many hurdles to prevent you having random choice. First, you have to build up enough rapport with a school so that they will allow their pupils to participate in your research. If you are allowed access into a school, the head teacher may know which teachers are less busy and do not have something planned and are in a good mood that day. When you are given access to a class, you may be able to randomly select children from that class (but not the ones who are off sick).
As an example of this sort of combination research, Goldstein, et al. (1993) wanted to sample children in schools in Britain. They used London as a cluster from which they selected schools randomly, and then classes randomly selected from those schools, and children randomly from those classes.
|
Crucial Tip: In your practical report you will often want to write that you used a specific type of sampling. Often, what you actually did, as we have seen, will not fit easily into one of the categories. If you find yourself in these circumstances, describe what you actually did, and try to explain why you did that. |
A final problem with sampling is that when you have selected participants, they may choose not to participate. If the participants refuse to answer your questions, they cease to be participants and you can’t use them. (The way around this is to pay people to participate, but this gets expensive.)
[edit] Section Summary
This section has considered sampling techniques that can be used to obtain a representative sample. We have looked at the ‘textbook definitions’ of random sampling, systematic sampling, quota sampling, cluster sampling, opportunity sampling, volunteer sampling and snowball sampling. We have also considered how the sampling technique that is actually used by psychologists varies from the textbook definitions.
