New Methods in Survey Research 1998

Proceedings of the ASC international conference, a satellite meeting for Compstat 98.
Organised by the Association for Survey Computing with Centre for Applied Social Surveys (CASS), Office for National Statistics (ONS) and the Market Research Society (MRS), 21–22 August 1998, Chilworth Manor, Southampton

Edited by Andrew Westlake, Jean Martin, Malcolm Rigg and Chris Skinner. ISBN: 0 9521682 3 5, x+164 pages.

Contents

Order Form

The series of Compstat conferences (on Statistical Computing) was started in Vienna in 1974, and quickly established the pattern of moving around Europe every two years. Compstat 1998 was in Bristol: the last time it was held in UK was in 1980, in Edinburgh, with an attendance of about 800. Soon after the series was established it became formally part of the European Region of the International Association for Statistical Computing (IASC, itself a section of the International Statistical Institute – ISI).

ASC is formally affiliated to the IASC, so this seemed an ideal opportunity for another international conference (following two successful conferences on Survey and Statistical Computing, in 1992 and 1996), to be held as an associated (satellite) conference, just before Compstat.

We decided to try to use the conference as an opportunity to enhance the link between Survey Computing and the Statistical kind, and to aim particularly at the ‘Survey Professional’ working in the commercial or public sector.

For the planning of the scientific programme ASC joined forces with related organisations, the Centre for Applied Social Surveys (CASS, an ESRC supported project hosted by Southampton University and SCPR, with links to Surrey University), the Office for National Statistics (ONS) and the Market Research Society (MRS). The conference is also officially sponsored by IASC and by its sister organisation IASS (the International Association of Survey Statisticians). ASC took responsibility for the practical arrangements.

At its first meeting the Scientific Programme Committee decided to organise the conference as four half–day sessions over two days, covering different (though related) topics. Each would start with an invited (keynote) presentation setting out the main ideas and issues involved, and followed by two or three contributed presentations on the practical issues and the benefits associated with the topic. We agreed that each organisation would take responsibility for one session, finding the keynote speaker, taking the lead in choosing, reviewing and editing the contributions, and chairing the session.

The following topics were chosen, falling naturally into two pairs for the two days.

The methods presented, though generally relatively new, are widely used in specialist areas but have not all made their way into everyday use in survey research or analysis. Out hope is that readers should gain some new ideas, together with enough insight to judge whether the methods can be useful in their own organisations.

The first half of the book is concerned with computer-assisted data collection. In the first section, Martin Collins and Wendy Sykes provide the keynote paper, reviewing the take-up and impact of computer-aided telephone and personal interviewing (CATI and CAPI) in the UK survey industry. This is followed by contributed papers on the use of multimedia in CAPI, on computer-assisted coding of complex responses (specifically for occupations) in CAPI, and on the generation of survey documentation for the benefit of data users from CAI programs developed by programmers.

Looking forward, Bill Blyth discusses the organisational issues that are affecting the take-up of technology, and spells out the data collection challenges that face research in the future. This paper is not included in the book, but copies of his presentation are available from the web site. Two of the contributed papers discuss the use of the latest hand-held technology for CAPI (both are travel surveys, though the general points apply much more widely). The third paper discusses options for using the Internet for data capture.

The second half of the book addresses issues in the use of data. Graham Kalton provides the keynote paper on developments in weighting methods, giving particular attention to calibration methods, where auxiliary information is used to adjust survey estimates so that they conform to external values. The contributed papers all present examples of the application of these methods in large-scale government surveys.

The last section is concerned with the place of statistical modelling in the analysis of survey data. Paul Harris offers a review of statistical modelling techniques, and discusses their take-up in survey research. The two contributed papers give examples of situations where the use of a model allows the information content of the data to be shared between subgroups, so that groups with few observations can draw power from related groups with more cases. The first example uses a simple model of fitted means to estimate rates across subgroups defined by three factors. The second applies the more complex technique of shrinkage estimation to rates for small areas, yielding significant improvements in precision.

As even this brief and inadequate summary makes clear, the papers which make up this collection are many and varied in their foci and vantage points. Taken as a whole, however, they provide critical commentary on the nature and use of the methods, as well as salutary lessons, based on hard experience, in their implementation. We hope that this volume will be of relevance to the survey practitioner, theoretician and managing director alike.