ESRC Data Archive Bulletin:
Northern Ireland Social Attitudes Survey

The Northern Ireland Social Attitudes Survey (NISAS) (Archive study number 33235) was first run in 1989. The survey is an extension of the British Social Attitudes (BSA) (Archive study number 33168) series run by Social and Community Planning Research (SCPR) which began in 1983. In Northern Ireland the agencies involved are SCPR who retain overall responsibility for the study, Central Community Relation Unit (CCRU) who are one of the funders, Policy Planning Research Unit (PPRU) who have responsibility for sampling and field work, and the Centre for Social Research at The Queen's University of Belfast (formerly the Policy Research Institute) who had responsibility for special analyses.

In 1989, the Nuffield Foundation and the Central Community Relations Unit in Belfast agreed to fund three extensions of the BSA survey to Northern Ireland. The survey has now been conducted in three years 1989, 1990 and 1991.

The basic format of the NISA survey in the three years that it has run has mirrored that of the British survey. It has two components. The first a questionnaire administered by interviewers and lasting, on average, an hour. The second component is a self-completion supplement which is filled in by respondents after the interview, and is either collected by interviewers or returned by post.

Each year the questionnaire includes a number of core questions covering areas such as defence, the economy and labour-market participation as well as a range of background and classificatory questions. Further sets of questions, or modules, on attitudes to other issues such as health, the environment or housing are included in the questionnaire less frequently - on a two- or three-year cycle, or at longer intervals. In addition, a special module on Northern Ireland issues such as community relations has been developed for the Northern Ireland questionnaire. Some of these questions are asked in Britain too. Therefore the survey gives us a unique opportunity to compare the attitudes of those living in Northern Ireland with the attitudes of people in Britain.

The series also has an international component, launched in 1985. The International Social Survey Programme (ISSP) (Archive study number 33213) now has 19 self-funding members each of whom conducts some sort of regular national survey of social attitudes. These questions are included in the self-completion section of both the BSA and NISA surveys. The possibilities for comparison then go further than Northern Ireland and Britain and include comparisons with for example, the Republic of Ireland and Germany.

Both the Northern Irish and the British surveys were designed to yield a representative sample of adults aged 18 years or over. In Northern Ireland the sample is drawn from the rating/valuation list, while in Britain the electoral registers provide the sampling frame.

The Northern Ireland survey is drawn as a simple random sample of all households listed on the ratings list. Approximately 1,400 addresses are drawn each year in the anticipation of achieving 1,000 completed interviews. The ratings list provides a good sampling frame of addresses, but contains no information about the number of residents at an address. The interviewers refer to a Kish grid to select one respondent from those eligible for inclusion in the survey living in the household. Because the selection of an individual respondent at each address cannot be conducted with probability proportionate to household size, the sample is weighted before analysis.

This is in contrast to the multi-stage sampling employed with the British survey. This involves four separate stages of selection. Firstly at parliamentary constituency level, secondly at polling district level, thirdly at address level and finally at individual level. A larger sample size in Britain around 4,500 addresses enables two versions of the questionnaire to be run. Both versions contain common core questions but differing modules. The absence of design effects in the Northern Ireland survey, and therefore of the need to calculate complex standard errors, means that the standard error and confidence intervals for percentage estimates for the NISA survey are only slightly greater than for the British survey, despite the smaller sample size. It also means that standard statistical tests of significance can be applied directly to the data.

The response rate to the survey varies each year however the Northern Irish rate has improved over the three year period to reach 70%.

Data from both series is lodged with the ESRC Data Archive after publication of the first report on the years data. Because of the fact that the NISAS is an extension of the BSA there are no gaps between the British and Northern Irish data apart from the larger sample size in Britain. Discussions are currently under way with funders chiefly Government to enable the Northern Irish series to continue. It is very much hoped that this will be the case.