Search site

School of Geography

Paul Norman Dr Paul Norman

Contact details

School of Geography
University of Leeds
University Road
Leeds LS2 9JT   UK


+44 (0) 113 34 38199

Student hours:

Work in progress

Current projects

ESRC What happens when international migrants settle? Ethnic group population trends and projections for UK local areas. Understanding Population Trends and Processes Large Research Grant RES-163-25-0032. October 2007 for 24 months. Co-investigator with Rees (PI).

This project aims to understand demographic changes that the UK’s local ethnic populations presently experience and will experience in the future. We are investigating how differences in ethnic fertility and mortality shape current and future trends, how international migration and internal migration impact the size and ethnic composition of local populations.

The project team will build projections of ethnic group populations for local areas and use the projection model to explore alternative futures. To carry out the projections, the project is engaged in: (1) estimating ethnic group fertility using alternative data sources, (2) estimating ethnic group mortality through indirect modelling; (3) building a databank of international migration for local areas from census, survey and administrative data sets and estimating long-term and shortterm immigration and emigration; (4) building estimates of internal migration for ethnic groups using census and register data; (5) developing a population projection model of ethnic populations for local areas that incorporates best current practice, that includes ethnic mixing; origin to destination migration rates and is adapted to handle different ethnic group classifications in the UK’s constituent countries.

Paul Norman’s main contribution to this work is the estimation of fertility trends by ethnic group and in the analysis of infant and overall mortality

Previous projects

Micro-geography of UK demographic change 1991-2001. ESRC grant RES-163-25-0012 Understanding Population Trends and Processes programme (UPTAP)

This project (2005-07) quantified and mapped changes in population size and social characteristics which occurred in small geographical areas across the UK between the 1991 and 2001 Censuses. This research investigated the relationships between natural change and migration and changes in the social characteristics of small geographical areas. This has enhanced our understanding of geographical aspects of population trends and processes through the identification of: locations in which population was ageing in situ; locations which were changing in population size and structure through net migration; and locations which over time changed in social characteristics in terms of their health, unemployment, housing tenure and level of deprivation.

Estimating small area populations for use in medical studies: accounting for population migration

ESRC CASE Award (S00429937028) PhD with partner Eastern Region Public Health Observatory (Norman, 2002). For medical studies, numerators of mortality/disease incidence and denominator 'populations at risk' for the geographical areas of interest are needed. Population migration affects both numerator (event counts) and denominator (population size and age-sex structure).

a.) Methodological problems hinder the derivation of reliable health rates. Outside census years, the production of cross-sectional population estimates to use as denominator populations is difficult because migration patterns are complex and the inclusion of a migration component in population estimates is problematical and invariably omitted.

b.) The migration process itself affects the counts of healthy and unhealthy persons in small areas. Migration is a selective process influenced by many factors including an individual's age and health status. This selectivity potentially affects both the numerator event data if an area receives or loses unusually high numbers of unhealthy migrants, as well as the denominator if an area receives or loses certain demographic subsets so that the age structure of the population at risk is changed.

Research to address issues relating to analysing time-series aggregate social statistics

This research has provided information about influences of migration on the calculation of small area populations. A time-series of small area populations has been estimated using a variety of methods with the differences in output assessed with the populations used as denominators in Standardised Mortality Ratios (SMRs). It is demonstrated that the method chosen and the inclusion of a specific migration element alters the size and age-sex structure of small area populations (Rees, Brown, Norman & Dorling, 2003). This directly affects the resulting SMRs which in turn may lead to inappropriately targeted resources.

For population estimates various data preparation and methodological enhancements have contributed to the literature. Highlights include: a.) a method by which to establish an annual time-series of data inputs and estimates based on the same geography when the boundary definitions change over the time period (Norman, Rees & Boyle, 2003) and; b.) a method of estimating and handling difficult to estimate sub-populations such as students, armed forces and communal establishments (Rees, Norman & Brown, 2004).

During a period when the availability of spatially referenced socio-demographic data has been dramatically increasing, the research demonstrates how the available data can best be utilised and how data compatibility and quality issues can best be addressed (Norman, 1999; Norman, 2004; Norman, 2005; Rees, Parsons & Norman, 2005; Norman, 2006). The same techniques will remain applicable and useful in the foreseeable future, for example, Norman and Simpson (2005) adjusted 1991 Census ward level data to the 2001 Census ward boundary definitions.

Administrative data have great potential to inform on social change. The system of local government itself responds to population change through the Periodic Electoral Review process. The prime focus is the aim for equality in electoral representation (Norman et al., 2007). However, the system itself creates difficulties in estimating and forecasting the population change to which it is responding and creates discontinuities in time-series of data on Incapacity Benefit claims which can inform on changes in small area population health (Bambra & Norman, 2006; Norman & Bambra, 2007).

Research to identify selection effects relating to migration and health

A greater level of clarity about the inter-relationships between health, migration and different types of location in terms of their deprivation characteristics has been established. An individual-level study (Boyle, Norman & Rees 2002) using 1991 Census Sample of Anonymised Records (SARs) shows that younger healthier persons tend to migrate over longer distances to less deprived locations but when younger people are in public housing short distance migrants tend to be less healthy and in deprived locations. Older persons, both migrants and non-migrants, are less healthy than all other persons.

Taking a longitudinal perspective, based on the deprivation of an individual's ward of residence at three different censuses, research using linked microdata from the ONS Longitudinal Study (LS) shows that over the period 1971-91 an accumulation of less healthy persons occurs in deprived areas with more healthy people migrating to less-deprived locations. The LS has proved an invaluable data source and the research has resulted in various journal submissions (Boyle, Norman & Rees, 2004; Norman, Boyle & Rees, 2004) with current research using links to 2001.


Geographical trends in infant mortality: England and Wales, 1970–2006

Norman P, Gregory I, Dorling D & Baker A (2008) Health Statistics Quarterly 40: 18-29
At national level in England and Wales, infant mortality rates fell rapidly from the early 1970s and into the 1980s. Subnational areas have also experienced a reduction in levels of infant mortality. While rates continued to fall to 2006, the rate of reduction has slowed. Although the Government Office Regions Yorkshire and The Humber, the North West and the West Midlands and the Office for National Statistics local authority types Cities and Services and London Cosmopolitan have experienced relatively large absolute reductions in infant mortality, their rates remained high compared with the national average. Within all regions and local authority types, a strong relationship was found between ward level deprivation and infant mortality rates. Nevertheless, levels of infant mortality declined over time even in the most deprived areas with a narrowing of absolute differences in rates between areas. Areas in which the level of deprivation eased have experienced greater than average reductions in levels of infant mortality.

The internal migration propensities and net migration patterns of ethnic groups in Britain

Stillwell J, Hussain S & Norman P (2008) Migration Letters, 5(2): 135-150
Internal migration propensities of ethnic groups are examined using three types of census data. Special Migration Statistics show variation in aggregate propensities whereas commissioned age-specific flow data indicate age variations by ethnic group. Micro data from Samples of Anonymised Records confirm low Asian propensities and suggest convergence between 1991 and 2001. Inter-district net migration reveals familiar counterurbanisation trends for whites but more complex patterns for non-whites. Evidence suggests white net migration at this scale is greater in areas with higher non-white population shares which themselves experience higher non-white immigration rates.
Keywords: ethnic populations; whites; non-whites; internal migration; England; Wales; Britain

Have national trends in fertility between 1986 and 2006 occurred evenly across England and Wales?

Tromans N, Natamba E, Jefferies J and Norman P (2008)  Population Trends 133: 7-19
During the past 20 years fertility patterns within England and Wales have changed considerably. The total fertility rate experienced a prolonged decline during the 1990s and hit a record low in 2001. Since then the level of fertility has increased fairly rapidly. Over the two decades, fertility has been constantly increasing at ages above 30, and as a consequence the mean age of motherhood has been rising. This article explores fertility trends within statistical regions and local authorities to improve our understanding of changes in fertility at the subnational level between 1986 and 2006.

‘Estimating with Confidence’ and hindsight: new UK small area population estimates for 1991

Norman P, Simpson L and Sabater A (2008) Population, Space and Place 14(5): 449-472
Population estimates are used in a variety of applications including the monitoring of social trends, the distribution of financial resources and assessment of demand for housing, schooling, employment and other goods and services. In many countries the primary source informing population estimates is a national census. A census, however, is not a perfect source. Even after imputation of missing households, the UK’s 1991 Census counts were lower than expected. In the 1990s, the Estimating with Confidence project (EwC) distributed official non-response data for local authority districts on a small area-specific basis to allow for underenumeration, timing changes between census day and the mid-year, armed forces adjustments and the transfer of students from home to term-time addresses. The EwC-enhanced census counts became accepted as the ‘gold standard’ for mid-1991 small area populations and were widely used in research.

Following the 2001 Census, evidence suggested that previous official upward adjustments to 1991 populations were too large. Revisions were retrospectively made to 1991 non-response and to the official annual time-series of subnational mid-year population estimates. For the UK’s small areas, here we have revised the original EwC non-response allowances, converted the output to more contemporary geographies and extended the estimates detail to single year of age. As a result, the 1991 EwC small area estimates are now consistent and comparable with the 2001 Census population definition and geography and the 1981-2001 official population time-series. Our revised UK coverage small area population estimates can underpin late 20th Century local area social analyses. Given the very detailed geographic and demographic detail in the estimates, we recommend these resources are used as building bricks for aggregation to larger areas and to application-relevant age bands.

We present example calculations to assess differences which occur when a revised population estimate is used for the calculation of population change and as the denominator in demographic rates. Re-worked research affects demographic, health and social indicators, especially the ranking of areas’ population and population change, although the extent of the revisions may often not result in different conclusions being drawn about trends and patterns.
Keywords: Small area population estimates; population time-series; census non-response; Estimating with Confidence


Representation and local democracy: geographical variations in elector to councillor ratios

Norman P, Purdam K, Tajar, A & Simpson S (2007) Political Geography 26 57-77
In a democracy, one person’s vote should count as much as another's. While a range of factors can affect this, including the electoral system, party support bases, party campaigning and the effectiveness and identity of representatives, a key principal is that for each political unit the number of electors per representative should be as equal as possible. Only when equality in electorate to representative ratios is established can equity in other demographic infrastructures be pursued. To achieve representation equality in English local authorities the Electoral Commission’s Periodic Electoral Review process considers for each electoral ward the number of councillors, current and forecasted electorates and revisions to boundaries. Here, using 2005 boundaries, we examine variations in elector to councillor ratios in England. Comparing these ratios with 2001 Census data, we investigate whether variations relate to ethnic minority population distributions.

We found considerable differences in representation ratios between four types of local authority. Generally, County Districts have fewer electors per councillor and therefore better representation ratios. There are progressively higher ratios for Unitary Authorities and London Boroughs; Metropolitan Districts have most electors per councillor. Comparing each ward’s ratio with the representation of its associated district we found most wards lie within what might be considered an acceptable range of variation. Sub-district representation variability relates to urban-rural variations in ward extent and the use of one-, two- and three-seat systems. There is no evidence that variations in ward ratios relate systematically to distributions of ethnic minorities. Despite this, to capture local population characteristics, we advise utilising ethnic group demographic characteristics when forecasting electorates.
Keywords: Democracy; Representation; Boundary change; Electorate; Periodic Electoral Reviews

The utility of medically certified sickness absence data as an updatable indicator of population health

Norman P & Bambra C (2007)  Population, Space & Place 13(5): 333-352
Despite the availability of mortality data, a lack of annually accessible morbidity information for small geographical areas in England and Wales means that health studies are often restricted to using decennial, self-reported census measures. Whilst the dissemination of census information has enabled much research, the self-assessment of health can be affected by subjective factors. Administrative data on health-related benefit claims, in the form of Incapacity Benefit (IB) data, are more regularly available and claimants are professionally diagnosed. This source may have potential to be an annual small area indicator of population health.

In this paper, we examine IB as an indicator of population health at local government district and sub-district levels by investigating distributions and relationships between IB and other health measures from the census and from mortality statistics. We found that relationships in 2001 between IB, census measures and mortality suggest that using IB as an indicator of population health will give very similar results, especially with those reporting themselves in the 2001 Census as ‘permanently sick or disabled’ and in the more urban areas. Time-series analysis also suggested IB was a useful and updateable source of data on population health.

Although IB should be an objective measure as it is professionally diagnosed, willingness to take time off work due to sickness, to consult a doctor or to claim benefits may be affected by cultural and socio-economic factors. Also, IB may be an incomplete count of ill people because some may be unable to claim benefits. We recognise that IB may be hiding unemployment and have inferred an estimate using census data on economic activity by exploring the relationship between IB and those reporting themselves permanently sick or disabled. This estimate suggests that previous assertions about the relationship between IB and unemployment may have been over zealous. On balance, IB is a useful indicator of relative health for small areas. However, the utility of Incapacity Benefit as an ongoing updatable indicator of population health depends on the future of the benefit system in view of pending reforms.
Keywords: Incapacity Benefit, sickness absence, LLTI, Census, mortality, morbidity

The use of spatial analytical techniques to explore patterns of fire incidence: a South Wales case study

Corcoran J, Higgs G, Brunsdon C, Ware A and Norman P (2007) Computers, Environment and Urban Systems 31: 623–647
The application of mapping and spatial analytical techniques to explore geographical patterns of crime incidence is well established. In contrast, the analysis of operational incident data routinely collected by fire brigades has received relatively less research attention, certainly in the UK academic literature. The aim of this paper is to redress this balance through the application of spatial analytical techniques that permit an exploration of the spatial dynamics of fire incidents and their relationships with socio-economic variables. By examining patterns for different fire incident types, including household fires, vehicle fires, secondary fires and malicious false alarms in relation to 2001 Census of Population data for an area of South Wales, we demonstrate the potential of such techniques to reveal spatial patterns that may be worthy of further contextual study. Further research is needed to establish how transferable these findings are to other geographical settings and how replicable the findings are at different geographical scales. The paper concludes by drawing attention to the current gaps in knowledge in analysing trends in fire incidence and proposes an agenda to advance such research using spatial analytical techniques.
Keywords: GIS; Fire incidence; Spatial Statistics; Regression; Visualisation

Cytotoxic drug use in treatment of dogs and cats with cancer by UK veterinary practices (2003 to 2004)

Cave T, Norman P and Mellor D (2007) Journal of Small Animal Practice 48(7): 371–377
OBJECTIVES: To describe the range and frequency of cytotoxic drugs prescribed within UK veterinary practices to treat dogs and cats with cancer, determine the effect of practice demographic variables on this practice and determine the frequency with which intravenous catheters were used during administration of parenteral cytotoxic drugs.
METHODS: A postal survey of 1838 veterinary practices providing care for dogs and cats within the UK.
RESULTS: Prescription of cytotoxic drugs to treat dogs and cats with cancer during the preceding 12 months was reported by 70.8 per cent practices. The most widely prescribed agents were cyclophosphamide (65.4 per cent) and vincristine (63_5 per cent). Twenty-three per cent of responding practices had prescribed an antitumour antibiotic and 8.3 per cent had prescribed a platinum agent. The median frequency of prescription was between once a month and once every three months. Increasing frequency and range of cytotoxic drug prescription were associated with practice employment of higher numbers of veterinary surgeons and increased levels of pet insurance among practice clients. Almost a quarter of practices administering vesicant parenteral cytotoxic drugs failed to always use intravenous catheters to do so. CLINICAL SIGNIFICANCE: Prescription of cytotoxic drugs, and therefore the potential for occupational exposure of staff, was widespread among UK veterinary practices providing care for dogs and cats.


What is the association between sickness absence, morbidity and mortality?

Bambra C and Norman P (2006) Health & Place (728-733)
This paper examines the area level relationships in England and Wales between sickness absence (‘Incapacity Benefit’), mortality and morbidity. It uses a random sample of Incapacity Benefit claims, and population counts of mortality and Census morbidity for local government districts. Although, there is little correspondence between sickness absence claims by specific cause and mortality, all cause sickness absence has a strong relationship with all cause mortality (male r 0.74, p=0.00; female r 0.64, p=0.00) and it also has a very strong relationship with the Census measures of morbidity: LLTI (male r 0.98, p=0.00; female r 0.97, p=0.00) and ’not good health’ (male r 0.99, p=0.00; female r 0.96, p=0.00). Incapacity Benefit claims by all causes has the potential to provide an ongoing measure of area level health in England and Wales.
Keywords: Incapacity Benefit, sickness absence, LLTI, Census, mortality, morbidity

Sociodemographic spatial change in the UK: data and computational issues and solutions

Norman P (2006). GIS Development special issue Maps & Census 10(12): 30-34
In the UK there has been a large expansion in the availability of spatially referenced Census, Vital Statistics and administrative data over the last few decades. This has been paralleled by increases in computer power, the sophistication of analysis packages and of programmer and user skills. To investigate trends and identify change over time we need data consistent in definition over time and space. Unfortunately, a variety of technical issues need to be overcome before even a rudimentary analysis can be carried out.

Attribute data may vary over time in terms of topic availability, definition and the demographic detail for which variables are released. Similarly, the geography for which data are available may change through time either due to a decision about the geographic scale of release or because a boundary change has occurred. These difficulties are compounded by the variety of geographies for which data may be disseminated. For time-series analysis, harmonisation of both attribute and geographical information is essential. In this paper some examples of problems and solutions are given.
Keywords: Boundary change; Data harmonisation; Census; GIS; UK


Selective migration, health and deprivation: a longitudinal analysis

Norman P, Boyle P and Rees P (2005) Social Science & Medicine. 60(12): 2755-2771
Population migration is a major determinant of an area’s age-sex structure and socio-economic characteristics. The suggestion that migration can contribute to an increase or decrease in place-specific rates of illness is not new. However, differences in health status between small geographical locations that may be affected by the inter-relationships between health, area-based deprivation and migration are under-researched. Using the Office for National Statistics (ONS) England and Wales Longitudinal Study (LS) 1971 to 1991, this research tracks individuals to identify any systematic sorting of people that has contributed to the area-level relationships between health (limiting long-term illness and mortality) and deprivation (Carstairs quintiles).

The results demonstrate that among the young, migrants are generally healthier than non-migrants. Migrants who move from more to less deprived locations are healthier than migrants who move from less to more deprived locations. Within less deprived areas migrants are healthier than non-migrants but within deprived areas migrants are less healthy than non-migrants. Over the twenty year period, the largest absolute flow is by relatively healthy migrants moving away from more deprived areas towards less deprived areas. The effect is to raise ill-health and mortality rates in the origins and lower them in the destinations. This is reinforced by a significant group of people in poor health who move from less to more deprived locations. In contrast, a small group of unhealthy people moved away from more deprived into less deprived areas. These countercurrents of less healthy people have a slight ameliorating effect on the health-deprivation relationship. Whilst health-deprivation relationships are more marked for migrants there are also health (dis-) benefits for non-migrants if their location becomes relatively more or less deprived over time.

Overall we found that between 1971 and 1991, inequalities in health increased between the least and most deprived areas, compared with the health-deprivation relationship which would have existed if peoples’ locations and deprivation patterns had stayed geographically constant. Migration, rather than changes in the deprivation of the area that non-migrants live in, accounts for the large majority of change.
Keywords: Health selective migration, Deprivation, Limiting long-term illness, All-cause mortality, ONS England and Wales Longitudinal Study

Making an estimate of the number of people and households for Output Areas in the 2001 Census

Rees P, Parsons J and Norman P (2005) Population Trends Winter 2005, 122: 27-34
The 2001 Census of Population Key Statistics and Census Area Statistics contain a large number of tables for 233,060 Output Areas in the UK. Users of the data often want to use one number for the total population count for each area or one number for the total count of households. Because of the disclosure control measures applied, different tables may report different population or household totals. This article describes an investigation into the best way to estimate robust totals for people and households at output area level. We show that computing the average of people and household counts across the set of tables in which the same count appears is a good strategy. When we compare the mean or median person and household counts with those produced by summing from the postcode directory, the two sets of numbers are usually very close. Using averages in this way for other population bases in the sets of tables is therefore recommended. This technique can also be applied to larger areas.


Changing places: do changes in the relative deprivation of areas influence limiting long-term illness and mortality among non-migrant people living in non-deprived households?

Boyle P, Norman P and Rees P (2004) Social Science & Medicine. 58: 2459-2471
Numerous studies have investigated the relative importance of contextual (place) and compositional (person) factors in explaining health and mortality variations. Commonly, these studies control for a range of individual characteristics before testing whether one or more contextual variables have a significant impact on the health or mortality outcome. The findings have been inconsistent, although the growing consensus is, first, that contextual effects are significant but are less important than compositional factors and, second, that contextual effects have a stronger impact in studies of morbidity than in studies of mortality. Here we use longitudinal data to examine a related, but rather different question. Extracting a select group of people from the ONS Longitudinal Study for England and Wales who had not moved house between 1971 and 1991 and who were living in non-deprived households throughout the twenty year period, we tested whether a change in the relative deprivation of the area in which they were living influenced their health and mortality status. The results demonstrate that changes in the relative deprivation of areas are related to health and mortality outcomes in a consistent way for both outcomes, although the results were more significant for morbidity. These findings suggest that public health and regeneration programmes designed to improve neighbourhoods may have demonstrable effects on the health of the residents who live there.
Keywords: Deprivation, Limiting long-term illness, All-cause mortality, ONS England and Wales Longitudinal Study, Context versus composition debate

A framework for progressively improving small area population estimates

Rees P, Norman P and Brown D (2004) Journal of the Royal Statistical Society A. Vol. 167 Part 1: 5-36
The paper presents a framework for small area population estimation that enables users to select a method fit for purpose. The adjustments to input data needed before use are outlined, with emphasis on developing consistent time-series of inputs. We show how geographical harmonisation of small areas, crucial to comparisons over time, can be achieved. For two study regions, the East of England and Yorkshire and The Humber, the differences in output and consequences of adopting different methods are illustrated. The paper concludes with a discussion of how data, on stream since 1998, might be included in future small area estimates.
Keywords: Population estimates; Small area populations; Estimation methods; Geographical harmonisation.


Achieving data compatibility over space and time: creating consistent geographical zones

Norman P, Rees P and Boyle P (2003) International Journal of Population Geography. Vol 9, Issue 5, September-October 2003: 365-386
The importance of boundary specification and problems of using arbitrarily defined sub-national areas for the collection and dissemination of socioeconomic data are widely recognised. Much research has focused on the modifiable areal unit problem and on custom zone design but the practical difficulties created by temporal inconsistencies in zonal boundaries have received less consideration. The UK experiences frequent administrative boundary changes causing difficulties in producing comparable statistics through time. Unless a consistent geographical approach is taken with time-series data it cannot be known whether changes in the relationships between variables collected for areas at different points in time are real or an artefact of boundary changes. After illustrating the nature of the boundary change problems to be overcome, this paper appraises strategies that can be adopted and develops a method by which time-series data can be established on a consistent geographical basis.
Keywords: boundary changes, geographical conversion tables, time-series data

Are socioeconomic inequalities in mortality decreasing or increasing within some British regions? An observational study, 1990-98

Rees P, Brown D, Norman P and Dorling D (2003) Journal of Public Health Medicine. 25(3): 208-214
Background This paper evaluates claims in a recent study that inequalities in small area mortality rates have lessened. We examine the effect of differently estimated populations on time trends in age specific mortality rates for Yorkshire and the Humber and East of England.
Methods Populations were estimated for wards using four methods that introduce increasing amounts of information. Age-specific mortality rates for age groups 45-54, 55-64, 65-74 and 75-84 for both sexes were calculated for population weighted deprivation quintiles. Inequality was tracked using ratios of rates in the most deprived quintile divided by those in the least.
Results When constant 1991 populations are used, rate ratios decrease for all age-sex groups, indicating shrinking inequality. When a method adjusting small area populations to official district estimates is used, both decreases and increases are observed in the mortality rate ratios. These results differ from Trent region findings of decreases in inequality. When small area populations are cohort-survived and adjusted to district populations, most differences in rate ratios indicate increasing inequality. When a method is used that includes information on migration and special populations, then seven out of eight age-sex groups exhibit increasing inequality.
Conclusions A judgement about trends in mortality inequality is highly dependent upon the denominator population used. Simpler estimation methods result in convergence of rate ratios, whilst more sophisticated methods result in increasing inequalities in most age-sex groups.
Keywords: population estimates, estimation methods, mortality, inequalities


Does migration exaggerate the relationship between deprivation and limiting long-term illness? A Scottish analysis

Boyle P, Norman P and Rees P (2002) Social Science & Medicine 55: 21-31
Few epidemiological studies of the links between health and environmental variables account for the potentially confounding effects of population migration. Here we explore the relationship between self-reported limiting long-term illness and material deprivation, using individual-level 1991 census data extracted for Scotland. The aim is to investigate whether the migration patterns of ill individuals influences the relationship between limiting long-term illness and material deprivation. Specifically, we seek to determine whether well individuals are more likely to migrate away from deprived areas and that ill individuals are more likely to migrate towards deprived areas. If true, this would suggest that the apparent relationship between deprivation and limiting long-term illness is exaggerated by the effects of migration. We then examine the issue controlling for individual level characteristics expected to influence limiting long-term illness and pay special attention to the role of public housing in these relationships.
Keywords: Scotland, Migration, 1991 British Census, Limiting long-term illness, Deprivation


Putting iterative proportional fitting on the researcher’s desk

Norman P (1999) Working Paper 99/3, School of Geography, University of Leeds, Leeds. Online: N.B. Software for two dimensional IPF available from the author by email request
Iterative Proportional Fitting (IPF) is a mathematical procedure originally developed to combine the information from two or more datasets. IPF is a well-established technique with the theoretical and practical considerations behind the method thoroughly explored and reported. In this paper the theory of IPF is investigated with a mathematical definition of the procedure and a review of the relevant literature given. So that IPF can be readily accessible to researchers the procedure has been automated in Visual Basic and a description of the program and a ‘User Guide’ are provided. IPF is employed in various disciplines but has been particularly useful in census-related analysis to provide updated population statistics and to estimate individual-level attribute characteristics. To illustrate the practical application of IPF various case studies are described. In the future, demand for individual-level data is thought likely to increase and it is believed that the IPF procedure and Visual Basic program have the potential to facilitate research in geography and other disciplines.
Keywords: Iterative Proportional Fitting (IPF); Population estimates; Census; IPF using Excel