Measuring and Coding Complex Items: (Semi-) Automated Solutions 1

Convenor: Dr Eric Harrison
Affiliation: City University London

Session Details

All social surveys rely heavily on socio-demographic information about respondents and design their instruments with a view to collecting the fullest most accurate measurements of these that are possible. Data about education, ethnicity, occupation, labour market situation - of both respondents and their families - form the essential backdrop to attitudes and behaviours measured elsewhere.

But they are problematic. They are complex to code, either because there is a huge range of possible answers (occupation), or because the context varies enormously between countries (education), or because respondents don't routinely undertake the task of self-categorisation (ethnicity/ancestry). Interviewers have to be trained to explore and probe in order to retrieve the fullest information from the field. This makes socio-demographic time-consuming and expensive to collect. Often the initial 'first pass' field data has then to be recoded by experts into smaller, more sociologically informed schemas.

As interest grows in using self-administered web surveys, we invite papers that report on the development or use of technological solutions to address some of these problems. Contributions addressing any complex categorical variables are welcome. These might include, but not be restricted to, measures of education, occupation, ethnicity, labour market experience, social class, social status, social distance, social networks, or reporting of medical conditions.

While papers will need to define and introduce the problem, the emphasis of this session is on solutions, so submissions will be required to have a practical component and include, where relevant, some demonstration material.