National Statistics Methodology Advisory Committee
Sixth meeting, 26 March 2004

The following work in progress papers were considered.

Paper 1: Re-designing the Consumer/Retail Price Index sampling plan
The Consumer price Index recently replaced the Retail Price Index as Britain's official measure of inflation. The two indices are used for a variety of official purposes including setting tax/benefit levels, wage bargaining and HM Treasury payments and setting interest rates. With an aim of improving precision, we are reviewing the current sample design.

Committee Conclusions:

• To understand the structure/characteristics of the data, consider variability in, and time series of, price data.
• The design should be optimised with respect to price change, and take into account national and regional requirements.
• Marginal costs of different sample designs, such as all price sampling at an outlet, are needed to understand the cost/precision trade-off.
• The impact of undercoverage should be assessed by sensitivity analysis, direct sampling, or simulation from enumeration results.
• The relationship between PPS size measures and price changes should be investigated.

Paper 2: Measuring PSA progress
A major element of the Public Service Agreement for the Department For International Development is a focus on achieving progress in specific social indicators in a range of key partner countries in Asia and Africa. In order to demonstrate this progress we are reliant on country level data published by the World Bank which we aggregate to provide composite figures for Africa and Asia separately. Our methodology and results are set out in this paper.

Committee Conclusions:

• Consider quadratic interpolation, a geographically larger model (and local covariates for missing data), and a multilevel model.
• Quality indicators (i.e. warnings if necessary) for data and for results were missing.
• Shocks to data should be investigated - their relation to missing data, and extrapolation in their presence.

Paper 3: Qualitative and quantitative methods for evaluating, redesigning and testing ONS business survey data collection instruments
Over the past 2 years Data Collection Methodology, ONS, has researched, developed and implemented methodologies for improving business survey data collection instruments. This paper outlines: the characteristics of business survey data collection instruments in the ONS; how improvements have been made to the development of business survey data collection instruments; and the framework for reviewing business survey data collection instruments.

Committee Conclusions:

• Cognitive interviews and expert reviews should be peopled with care.
• Consider qualitative assessment of questionnaires, and pilot studies in addition to field testing.
• Questionnaires should be coordinated across surveys, and long questionnaires could be split over time.
• Changes to questionnaires should be documented and clearly communicated with users.
• Respondents should have the choice of response mode.
• To estimate measurement error, collect "gold standard" data from a subsample of respondents.

This page last revised: Tuesday, 26 October 2004