Sampling methods and Surveys - Topic 5 (part 1) - Theory and Methods

Surveys and Sampling Methods

Surveys are used to collect primary data from large numbers of people, usually via questionnaires or structured interviews.

Positivists prefer the survey method because it produces quantitative data in statistical form. The government often carry out surveys like the census, and other such as market researchers and election pollsters to find out how people intend to vote.

Sociologists rarely have the time or funding to complete large scale surveys of the entire population, so they instead opt for a smaller group known as a sample.

To achieve a representative sample, all the relevant characteristics of the whole group must be included (survey population). 

For example, to get the voting intentions of women aged 30-40 in the UK, one woman aged in this range could be selected from every postcode in the UK, then the results of the study could be generalised to the wider population. It is very important for a sample to be representative, otherwise generalisations cannot be made. If a sample size is too small will mean that is is not representative. In the example above, taking one woman at random from each postcode may not be enough as the sample did not include other characteristics like ethnicity, social class, etc.

A sampling frame is a list of all people included in the survey population from which the sample is selected, e.g. if you were researching your school then the school roll or class registers could be used.

Other commonly used sampling frames include the electoral register or phone books. Again the sampling frame should be representative with no group left out. Just remember that it important that your sample should be representative of the population you are studying!

Lets explore some more sampling methods!

Random Sampling

Random sampling means that every person in the survey population has an equal chance of being picked, this can be done by pulling names out of a hat or by asking a computer to select names (numbers) at random from a sample population. 

The problem with random sampling is that it can leave an unrepresentative sample, e.g. each person selected at random could be male.

Systematic Sampling 

Systematic sampling is when names are selected from a sampling frame at regular intervals until the desired size of the sample is reached. 

For example, if you wanted to interview 50 pupils from your school, you could pick every tenth name from the school roll until you have reached the 50th pupil. 

The downside of this method again is that it could be unrepresentative , by chance each name you selected could've been male.

Stratified Random Sampling

Stratified random sampling is a way of ensuring that a sample can be representative and avoid the makes that can happen in a random sample.

For example, if you knew that 20% of your school were from minority ethnic backgrounds and you wanted the sample to be representative you could separate out the pupils from minority ethnic backgrounds and then take a random sample from this group to make up 20% of the pupils in the study. The remaining 80% could be a random sample from the rest of the school population. 

To help remember this - think of what stratified or stratification means in Sociology. It refers to strata (levels, subdivisions) in society like social class, age, ethnicity, sexuality, education, etc. By selecting the appropriate number of people from each strata you make the study representative. 

Quota Sampling

Quota (quota - meet a number) sampling is when researchers are told to go and find/select a certain number of people who fit into certain categories, e.g. go find 20 people aged 18-30 to interview. 

The problem with quota sampling is with representation - if the interviewer is approach people who are shopping, or at a sporting event, they will not get the views of others did are not shopping or at that event. Bias can enter quota sampling as the interviewer might not approach certain people based on their looks. 

Cluster Sampling (multistage)

Cluster sampling involves selecting a sample of people in various stages. It often gets confused with stratified random sampling to take care to note the difference. 

In conducting a study into school students, the researcher could randomly select 5 schools from all schools in UK, then randomly select one year group from each school, then randomly selected 5 people from each year group. As you can see, it is done in stages and narrowed down to a manageable sized group for the interviewer to access and research. 

Snowball Sampling

This is the easiest to remember! 

Imagine a snowball rolling down a hill, getting larger and larger as it gathers more snow. This is how snowball sampling works! The researcher may find one person who is willing be be interviewed, and then that person could introduce the researcher to another person, and on and on it goes. 

This type of sampling is used when a research group is difficult to reach, e.g. when researching crime and difficult or dangerous people/situations. Snowball sampling is not representative nor random and it doesn't try to be.