RGS-IBG Annual International Conference 2015

Add to my calendar:    Outlook   Google   Hotmail/Outlook.com   iPhone/iPad   iCal (.ics)

Please note that some mobile devices may require third party apps to add appointments to your calendar

170 Exploiting New Data for Population Research (1): Demographic Insights
Affiliation Population Geography Research Group
Quantitative Methods Research Group
Convenor(s) Adam Dennett (University College London, UK)
Ian Shuttleworth (Queen's University Belfast, UK)
Nik Lomax (University of Leeds, UK)
Christopher Lloyd (University of Liverpool, UK)
Chair(s) Adam Dennett (University College London, UK)
Timetable Thursday 03 September 2015, Session 3 (14:40 - 16:20)
Room Forum - Seminar Room 9
Session abstract Researchers studying population have long relied on the rich and familiar data contained in national population censuses. However, as the popularity of censuses worldwide is challenged by the ‘data deluge’ and the prospect of free (or at least by-product), real-time (or at least more-timely) and ‘Big’ new datasets, what does this ‘brave new world’ offer population geographers? There is potential to ask and answer new questions but also significant theoretical and methodological challenges in handling and extracting meaning from these proliferating new datasets. The session aims to explore not only these new social and policy questions but also the methods that can be most appropriately used.
Linked Sessions Exploiting New Data for Population Research (2): Global and Health Insights
Contact the conference organisers to request a change to session or paper details: AC2015@rgs.org
Using Satellite Data on Night-time Lights Intensity to Estimate Contemporary Human Migration Distances
Thomas Niedomysl (Lund University, Sweden)
Ola Hall (Lund University, Sweden)
Ulf Ernstsson (Gothenberg University, Sweden)
Maria Archila (Lund University, Sweden)
For well over a century migration researchers have recognized the lack of adequate distance measures to be a key obstacle for advancing understanding of internal migration. The problem arises from the convention of spatially defining migration as the crossing of administrative borders. Since administrative regions vary in size, shape and settlement patterns, it is difficult to tell how far movers go, raising doubts about the generalizability of research in the field. This paper shows that satellite data on nighttime lights can be used to infer accurate measures of migration distance. We first use the intensity of nighttime lights to locate mean population centers that closely correspond to actual mean population centers. We then show that this information can be used to accurately estimate migration distances.
The Annual Survey of Hours and Earnings: potential for new insights into migration and commuting
Tony Champion (University of Newcastle, UK)
The geographical literature on migration and commuting is very rich, but is lacking in at least two significant ways. One is that the primary thrust of these is cross-sectional, relating to a single point in time, as derived from separate censuses, while most time-series work is on aggregate flows rather than tracing individuals longitudinally. Secondly, the work on migration is focused almost exclusively on residential mobility, with very little attention being given to workplace mobility and thus providing very little intelligence about the changes in commuting journeys produced by changes of home or work address. This reflects the fact that quantitative analyses of migration and commuting are highly reliant on the population census, which in the UK as elsewhere normally asks questions about current home and workplace addresses and home address at a date in the past, but not about workplace at the same previous time. This deficiency has proved a stumbling block in previous research that has aimed at finding out whether moving home increases the length of people’s commute and, if so, whether this is an enduring change or leads to a subsequent adjustment through change of workplace or a further change of home address.

The Annual Survey of Hours and Earnings (ASHE) has the potential for plugging these gaps. It comprises a 1% sample of employees based on National Insurance number and has been running since 2002 on its current basis which includes geographical details of employee’s home address as well as of his/her workplace. Survey forms, sent annually to their registered employers to complete, also ask about the employee’s wages, hours of work, pension arrangements, occupation, industry, date of first employment by the current employer, sex and date of birth. Hence, it is possible to develop a longitudinal panel data set for a 1% sample of Great Britain’s employees that permits one to determine if a worker changed residence during any particular year, and then to cross classify residential mobility with changes of workplace in that year or in any succeeding year, and vice versa. Moreover, since ASHE provides the full postcode of workplace and residence, location can be recoded to a variety of geographies including rural-urban location and size of settlement.

This paper begins by developing the points just made. It then presents data on key dimensions that can be compared with those derived from the census, such as propensity to change home address in a 12-month period, distance of one-year address change and travel-to-work distance. It goes on to demonstrate the value of ASHE in being able to identify frequency of moving home for this population (as far as annual observations allow), as well as whether a change of residence is accompanied by change of workplace in the same year or subsequently and, more generally, what effect any change of residence and/or workplace has on the length of the commuting trip. Policy relevance will also be discussed.
Exploiting administrative data for population estimation and profiling – experiences and applications
Gill Harper (Mayhew Harper Associates / Geocreate, UK)
There has been a wider move in the UK to utilize administrative data sources in research and for the future production of population statistics. The ONS continue to explore this to replace or supplement the census survey of population as part of their Beyond 2011 programme, backing the National Statistician’s final recommendation in April 2014 for a predominantly online census in 2021 supplemented by further use of administrative and survey data.

For over 12 years, the author has been utilizing administrative data at the local level to estimate and profile populations. This experience provides a unique insight of the advantages and methodological challenges of using data for this purpose in a UK setting.

The paper will set out the author’s experience, how the method has developed and modified over time, how it has been combined with other data sources such as surveys when relevant, and the future of the method in light of ever changing data access restrictions and uncertain legislation.

In particular, detailed case studies will be given of how an administrative data derived population database creates a wealth of variables to profile the population and households, providing an evidence base for policy, more strategic intelligence-led decision making, to model demographic change, and commission services more effectively. Application areas include older people services, selective licensing of the private rented sector, energy efficiency, public health and health and well-being.
Studying patterns of socio-economic segregation using diurnal mobility of mobile-phone users in Sweden
Marcus Mohall (Uppsala University, Sweden)
John Osth (Uppsala University, Sweden)
Thomas Niedomysl (Lund University, Sweden)
Register based segregation analyses to a large extent make use of population registers containing information about the residential location of individuals at the time of survey. Most of these databases contain large samples or even all officially residing individuals in the regions of study. However, much of the statistics describing the residential location and socio-economic composition of the population are collected on a discrete basis, depicting the situation for a specific date annually or even decennially. Using a unique register describing the diurnal mobility of all phones in a major telephone network company in Sweden, we are able to see if patterns of residential segregation hold after controlling for the diurnal mobility of the phone holders. By associating the approximate position of each phone with contextual data describing local socio-economic status of the area in which the phone is being used. A comparison between residential, population register based and mobility based phone registers can be made. The analysis indicates that phone mobility decreases segregation in general but also that these patterns vary between locations and population segments.
Geographical Inequalities, Spatial Scale and Small Area Statistics for England and Wales
Christopher Lloyd (University of Liverpool, UK)
Following on from recent debates about the future of the UK Census, which included discussion about alternative data sources and their suitability for constructing small area statistics, this paper assesses some of the implications of selecting different zonal systems for the analysis of geographical inequalities in England and Wales in 2011. The analysis computes the index of dissimilarity and the variogram for a set of nested zonal systems to characterise spatial variation in population sub-groups as represented by Census counts for Output Areas (OAs), Lower Layer Super Output Areas (LSOAs), Middle Layer Super Output Areas and local authority districts (LAs). The analysis shows how much information (in terms of variation) is contained at each spatial scale for counts (or log-ratio transformed percentages) relating to age, ethnic group, housing tenure, car or van ownership, qualifications, employment, limiting long term illness (LLTI) and National Statistics Socio-economic Classification (NS-SeC). It is shown that for all variables OAs, LSOAs and MSOAs display similar spatial structures at a regional scale (within 80 km) but that, for most variables, a considerable amount of variation is lost with aggregation from OAs to LSOAs and from LSOAs to MSOAs. Collectively, the results provide useful information which could be used to guide future design of population surveys and also provide guidance to users in selecting an appropriate zonal system for analysis of particular variables. The results also provide evidence on some of the ways in which the population of England and Wales was geographically distributed in 2011.