Ruth’s Sociology Resources Blog

This section is not quite finished – basically I ran out of steam and will back another day to complete it!

A lot of students find the idea of sampling scary - don’t, it’s really common sense and once you understand the key ideas you should find it simple.

When a sociologist is doing a piece of research it is often impractical for them to survey everyone in the group they are looking at. For example, in my research on home education it would be impossible for me to interview all of the 150,000 children that are being home educated and their families - I therefore want to choose a sample of those children - a smaller group that I can work with given the time and resources that I have available.

This is what sampling is all about - how you choose the people that you look at and the things you need to think about when making those choices.

Key Concepts in Sampling

Population - often also called the target population, this is used to talk about the group that you are studying. So in my research on home education in the UK, families in the UK who are home educating their children are my target population - they are the group of people that I want to study.

Sampling Frame - this is the list of my population from which I will choose my sample. Sometimes there is more that one possible sampling frame for a population - for example if we wanted to choose a sample from the whole adult population of he UK there are several lists or sampling frames that we could use - we could use the Electoral Register or we could use the Post Office Address File which lists all the addresses in the UK. Sampling frames are rarely perfect - people may be missed off the Electoral Register and they can also choose to have their details removed from the version that is sold to researchers and businesses; using the Post Office Address File makes it difficult to sample some groups of people - the homeless are a good example.

Sampling frames come in all shapes and sizes and you need to pick one that is appropriate for your target population - so using the electoral register to draw a sample of people under the age of 16 or illegal immigrants is inappropriate as neither of those groups are covered by it. For a smaller research project it may be appropriate to use a much more local sampling frame - for example if you want to study students aged 16-18, an appropriate sampling frame could be a school or college’s list of students.

When choosing your sampling frame you will need to consider how representative it is of the population being studied as this will affect the validity, reliability and generalisability of your findings.

Sometimes there is no suitable sampling frame available for the population you want to study. For example there is no comprehensive list of home educating families in the UK (nobody actually knows definitely how many there are). There are organisations for home educating families , so I could use one of their membership lists as a sampling frame, but I am also interested in studying families who are not members of home education groups. The fact that there is no suitable sampling frame for my study will affect the sampling technique that I use.

Sampling Technique - this is the method you use to choose your sample. There are several different techniques available and we will have a look at some of the most common.

Representativeness - how representative a sample is describes how well it represents, or stands for, the target population. For example, if the target population contains 40% women and 60% men then a representative sample would also contain 40% women and 60% men. Obviously if we want to make generalisations about the whole population from the findings about our sample we need our sample to be as representative as possible.

Sampling Techniques

When talking about sampling techniques, how good they are is often judged by 2 main criteria - whether the technique produces a random sample and whether it produces a representative sample.

Random Sampling This is like picking names out of a hat, there is no system and you don’t know what will be picked. These days random sampling is often done by a computer. If a sampling technique is random it means that every unit in the sampling frame has an equal chance of being chosen (we often talk about units as we could be drawing a sample of families or schools instead of individual people). Random sampling is therefore good as it can prevent bias when choosing a sample. If the sample is large, random sampling will usually produce a sample that is representative of the population being studied, however this is not certain - remember in random sampling you have no control over which units are chosen from the population for the sample.

The National Lottery is a good example of random sampling - the number balls are chosen randomly by a computer and each ball has the same chance of being chosen.

Stratified Random Sampling
Stratified random sampling is a technique where, using your sampling frame, you split your target population up by characteristics and then choose randomly from within the groups created. So, going back to my population which contains 60% men and 40% women - I would divide the sampling frame into two groups - men in one and women in the other, I would then choose 60% of my sample at random from the men and 40% of my sample at random from the women. This would guarantee me a representative sample.

So stratified random sampling is great for gaining a representative sample that has at the same time been chosen randomly and is therefore relatively ‘fair’, however, with most samples you would need to stratify (ie divide) your sample in many different ways - ie by class, ethnicity and age as well as gender, which can make the sampling process very complex and long-winded. There is also the problem that you have to know the characteristics of your population in order to divide them up and that information can be hard to get.

Cluster Sampling
Cluster sampling is similar to stratified sampling but in this case the groups are geographical. So if, for example, I am carrying out a study of people in the whole of England instead of sampling people from all over England I will pick some geographical areas that form a relatively representative picture of England as a whole. So I might select a city, a rural area, a few small towns and villages and draw my sample randomly from within them.

Cluster sampling is useful in that it can help reduce the time and cost of research - so instead of interviewing people scattered all over the place I interview several people in each a few different places. However, cluster sampling is not truly representative and it is also not truly random as different units have had different chances of being chosen.

Systematic Sampling
Systematic sampling does what it says on the tin - you have a system for choosing the sample. Usually this involves picking every Nth person from the list that is your sampling frame - so if you want a sample of 50 out of a population of 100 you would pick every 2nd person, if you wanted a sample of 25, you would pick every 4th person. Systematic sampling is not random (obviously!); it can produce a representative sample depending on how the sampling frame is arranged. An obvious example of this if you have a population which is 50% male and 50% female and the sampling list is arranged male-female-male-female - if you pick every 2nd person you will end up with a sample that is 100% female!

The big advantage of systematic sampling is that it is quick, cheap and easy.

Quota Sampling

Snowball Sampling