• Study series
• 25 years of BSA
• Our methods
• Participants

British Social Attitudes | Our methods

What do you think?

The British Social Attitudes survey series has been conducted annually since 1983 and is NatCen's longest running survey. Each year over 3,000 interviews are conducted with people in Britain. Participants are selected using a technique called random probability sampling. This technique ensures that everyone has a fair chance of taking part in the survey and the results are representative of the British population.

BSA has covered an extensive number of complex social, political and moral issues. Topics include work, transport, health, education, government spending and voting habits, as well as religion, racism and illegal drugs. New areas of questioning are added each year to reflect current issues, but all questions are designed with a view to repeating them periodically to chart changes over time.

You can find out more about the questions we have asked the British public over the last 25 years here.

The datasets are also available for further analysis from the UK Data Archive at the University of Essex.

The way we have carried out the study has changed a lot since it began:

• Originally we sampled addresses from the Electoral Register. But, in 1991, the survey changed its sampling frame to the Postcode Address File, which is a list of addresses maintained by the Royal Mail. We did this because the electoral register was increasingly unrepresentative of the British population.
• When the series began, interviewers had to complete lengthy paper-based questionnaires by hand. This was replaced by computer-assisted interviewing in 1993. This change made life easier for interviewers and meant we could get data more quickly.
• We regularly carry out experiments to see how we can improve the study. For instance, in 2006 we conducted an experiment to see if sending people an incentive in advance affects their willingness to take part in the survey or not (we changed our incentive strategy as a result). In 2009, we are testing whether a new logo and branded documents have any effect on whether people take part in the survey.

Of interest: