![]() ![]() ![]() is our primary concern migration as a component of population change? If so, are we equally interested in immigration and emigration?.what needs to be defined? How is a migrant defined? Can this be identified in the source data sets?.what are the variables or concepts of interest?.what is the population of interest? Can we account for this population in its entirety? Do we need a control population for comparison?.is the intention to understand processes, to examine the correlates of longitudinal outcomes? Or are we seeking to produce robust measures and statistics?.what are the research questions that the data need to help to answer?.We provide an example from the work we are currently doing on developing methods for measuring international migration. It is essential to have a clear and concise statement of what the aims and objectives are for the data, including the concepts and the variables that will be involved. The three phases of the design process merit expansion. You wouldn’t leave any of the implementation to chance. ![]() From the initial outline sketches down to the minutest detail identify any structural constraints that may affect the design.In building a longitudinal data set, it’s important to begin by thinking about what the data set is to be used for. They come after the design work is complete. Where would you begin? Not with concrete, bricks and mortar. Think of it as you would the construction of a building. The focus here is on the statistical design of a longitudinally linked administrative data set. We like to use a metaphor from the construction industry when we think about statistical design. We welcome comments and suggestions to further progress this exciting and valuable research area. We provide examples from our application of this framework to help us understand administrative data, for estimating international migration and for the production of population estimates based on administrative data. We are still developing our thinking regarding error management and the quantification of statistical error and uncertainty, so this paper reports on work that is still in progress. Some error can be quantified, and we highlight the use of scaling factor analysis, linkage error investigations, edge effects and the analysis of residuals. Some of the errors that we draw attention to are conceptual and so resistant to quantification, in any precise way at least. This is intended as a helpful taxonomy of the potential errors that are integral to administrative data sources, to be aware of in either evaluating admin-based sources or in designing new, integrated, longitudinal data sets. To support this, we propose an error framework. The idea is that the design should represent the optimal balance of user requirements, design features and statistical error management. We emphasise the importance of careful statistical design, prior to implementation. While the focus is on administrative data linkage, it could also be applied to other data blends, for example microdata linkage of survey and administrative data. This paper seeks to offer a template for the statistical design of data sets that are derived through the linkage of administrative data to produce longitudinal data sets. This has further deepened our understanding of this data source and the findings from this research will help us address statistical design challenges of using longitudinally linked administrative data sets and, through the use of quality indicators, inform and support analytical use of these types of data. We therefore developed an error framework to assess statistical error in administrative data sources to help fill this gap and applied it to longitudinally linked Home Office data used operationally at the border. Much of the population and migration statistics transformation research in ONS involves longitudinal linkage of administrative data by diverse teams.įrom an extensive review of the literature we realised that existing error frameworks did not support the use of longitudinally linked administrative data sets. ONS is currently progressing a programme of research which aims to increase the use of administrative data in the production of population and migration statistics. IntroductionĪlongside many other National Statistical Institutes (NSI), the Office for National Statistics ( ONS) is committed to increasing the use of administrative data in the production of statistics to reduce costs, reduce respondent burden and to improve the quality and granularity of our statistics. Louisa Blackwell and Nicky Rogers, Office for National Statistics 1. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |