 Many of our survey data sets include survey weights. This video introduces what we need to know about survey weights when using survey data. Surveys usually collect information from a sample of the population with the aim of inferring information about the population as a whole. Survey designers use rigorous sampling techniques to try and generate a representative sample. Survey weights are also used to adjust sample data to make it better represent the population. One reason for weighting data is that common sampling methods can mean members of a population have different chances of being selected to take part. For example, surveys in the UK often draw a sample from a list of addresses called the postcode address file to get samples of individuals with their sample addresses, and at each address select an individual at random. Using this method, individuals who live alone are more likely to be sampled than individuals who live with others. Using the known differences in the probabilities of being selected, we calculate design weights to adjust the sample. Weights can also help compensate for survey non-response. Not everyone sampled will take part in a survey, some people cannot be contacted and others refuse. Response rates can vary systematically across groups. Non-response weights use information about response rates for subgroups to adjust data and limit potential bias. We can also use weights to adjust a sample to reflect key population proportions. Known as post stratification or calibration weights, they use information from sources such as the UK Census to improve the accuracy and precision of population estimates. Let's take a closer look at how weights work. Usually data producers combine adjustments into a single weighting variable that you'll find in the dataset. The weighting variable will contain a value for each case, which indicates how each case should be weighted during analysis. Higher numbers indicate that the case will count for more. For example, we'll find higher weights for people who live in larger households, and for those coming from population groups underrepresented in the sample, due to either chance or non-response. Methods for using weights vary by statistical software package, but involve indicating the name of the weighting variable before or as part of the analysis. Importantly, results can vary between weighted and unweighted analyses. Here we have the frequencies for a variable indicating household size. The unweighted results indicate that nearly 30% come from a one person household. In contrast, the weighted results suggest it's near a 17%. If we do not use weights when analysing survey data, results might not relate to our population of interest. Weights can also be used to make the sample look the same size as the population. Known as a grossing weight, they can help us describe the prevalence of social phenomena. For example, the crime survey for England and Wales to help us describe rates of crime. These can be useful, but we need to remember the sample is much smaller than it looks, and to check how the statistics software is calculating the precision of the estimates. You will find information about the weights in the documentation that comes with the data. Sometimes you'll find multiple weights in need to establish which to use for your analysis. For example, health survey for England data sets contain a weight of use with core interview data, and weights for the follow-up stages of the survey. However, as you can see, the user guide gives details about which weight to use when. This video is given an introduction to survey weights. To explore the topic further, you can read our more detailed what is weighting guide. We also have guides to the main statistics software packages, which include instructions for applying weights.