An Introduction to Survey Research
What are the different methods of sampling?
Why should we do surveys? How can we obtain representative samples? How can surveys go wrong?
This article is to help students taking the University of Exeter’s POL/SOC1041 unit, and other researchers. This article will focus on statistical thinking, rather than arithmetic.
Why do a survey?
Surveys help researchers answer questions that other data sources cannot provide. There is no complete register of all unemployed people, so the Office for National Statistics conducts the Labour Force Survey to estimate unemployment.
In the UK, a census — a count of all people and households — takes place every ten years. This is too infrequent for most purposes and too costly to repeat more often, possibly missing information that researchers find important.
Usually, the goal of a survey is obtain a representative slice of a population. After steadily cutting a cake with a good knife, we want a proportionate amount of icing, jam and layers of sponge.
We do not need to eat the whole cake to know its taste.
We do not need to ask everyone to roughly understand what a population believes. We only need a sample.