Health, Housing Tenure & Entrapment : Does Changing Tenure and Address Improve Health? Myles Gould (UoL) ac.uk Ian Shuttleworth (QUB) Presentation at British Society for Population Studies 2014 Conference Winchester, Wednesday 10 th September 2014

Structure Introduction Project Aims & Presentation objectives Data & analytical approach – NILS Data – Population bases Data description Multilevel Cross-Interaction Model Results – Staying good health ( ) – Transitioning good to bad health ( ) – Transitioning bad to good health ( ) Conclusions

Introduction Political & policy debates about social-rented housing focus on low spatial mobility & reduced chances of upward social mobility Extensive literature on inter-relationships between: housing tenure, health, and wider dimensions of social wellbeing, and the measurement of these at both the individual and area level (e.g. Marmot, 2010; Macintyre et al, 2002) Smith & Easterlow (2005) consider concepts of housing entrapment & selective placement – Are people entrapped in poor housing & health? – Are people selectively placed in tenures / spatialities in poor health?

Project Aims 1.To explore relationships between changing health & housing tenure in Northern Ireland, To determine whether different tenure trajectories are associated with changes in health status e.g. movements from social rented to owner occupied housing & changing health status 3.To explore whether changing health status is linked to different kinds of spatial move/mobility – moves between different types of place/area (e.g. area deprivation score) – i.e. change SOA geographical area Aims 1 & 3 are todays presentation’s focus Aim 2 was considered in previous presentation (RGS- IBG, & NILS Launch event)

Today’s Presentation Objectives 1.Explore the effects on tenure (2001) on changing health status ( ) 2.Explore the relative importance of changing address and of changing tenure on changing health status ( ) 3.Explore the importance of initial area context (multiple deprivation 2001, & change )

Data and Analytical Approach Restricted set of variables for parsimony Descriptive analysis different kinds of tenure/health transition in the NILS – changes in individuals’ general health Multilevel statistical modelling (individuals nested in SOAs) Later will look at limiting long-term illness (disability ) – will also compare health changes with chronic illness(es) status in 2011

NILS Structure 2001 Core NILS ID Health card registration Sex & age 2001 Census Household data Individual data 2001 SOA code 2011 Core NILS ID Health card registration Sex & age 2011 Census Household data Individual data 2011 SOA code 2001 Aggregate Census Area statistics 2001 SOA code 2011 Aggregate Census Area statistics 2011 SOA code 2005 Area Deprivation (MDM) 2001 SOA code One in three sample of the whole N. Ireland population Born on a selection secret birthdates Treating as individuals, ignoring fact households might have multiple members

Population Bases Movers – changed tenure – may/may not changed home/address – &/or changed SOA – theoretically possible to only change tenure - e.g. (re)mortgage, buy from landlord Movers changed address / SOAs – Recently been analysing this too: 4 possibilities 1.Changed tenure, changed address 2.Changed tenure, not changed address (unlikely) 3.Not changed tenure, changed address 4.Not changed tenure, not changed address

Health Transitions: Health 2011 Good health Fairly good health Not good health Total Health 2001 Good health %13.4%3.2%100.0% Fairly good health %39.8%11.0%100.0% Not good health %42.2%37.4%100.0% Total %22.7%8.9%100.0% Health 2011: 5 categories recoded /combined to 3 to compare with 2001 Consider model results for those cells shaded in red today

Health Transitions ( ) & Age (2011) Reminds us of obvious importance of taking account of age, doing this in our statistical modelling work

Variations in Average Area by Area (SOAs): 2011 Demography varies by SOA, need to take account of this in our statistical models

Modelling Approach Restricted set of variables for parsimony At this stage a number of logistical regression models treating response as binary outcomes – Sticking: 1. Staying Good; 2. Staying Good ( ) – Transitioning: 3. Good to Bad; 4. Bad to Good ( ) Multilevel statistical modelling (individuals nested in SOAs) – Place difference having taken account socio-demographic characteristics – Include area effects for SOA deprivation in 2001; or ratio for change in deprivation Cross-level interactions: individual/household & area effects – Area deprivation with 2001 tenure – Area deprivation with indicator for changed tenure &/or changed address

Model Predictors Having allowed for 2001 tenure [tenure change in previous analysis]… …Also age, sex, occupational status, education level, & community background Plus allowed for response to vary by SOA find small effect, but significant place differences

Modelling probability staying in good health

Cross-level interaction of individual & area effects: (Response: Staying good health) SOA Multiple Deprivation (md=17.8), 2001 Owner Occupier Social renter Private renter Prob. Remaining in Good Health ( )

Ratio of SOA Deprivation 2001 compared 2011 (Area change with respect to deprivation) Owner Occupier Social renter Private renter Prob. Remaining in Good Health ( ) Cross-level interaction of individual & area effects: (Response: Staying good health)

SOA Multiple Deprivation (md=17.8), 2001 Prob. Remaining in Good Health ( ) Changed tenure, not changed address Changed tenure, changed address Not changed tenure, changed address Not changed tenure, not changed address Cross-level interaction of individual & area effects (Response: Staying good health)

Modelling probability staying in bad health

SOA Multiple Deprivation (md=17.8), 2001 Prob. Remaining in Bad Health ( ) Cross-level interaction of individual & area effects (Response: Staying bad health) Owner Occupier Social renter Private renter

Modelling probability changing from good to bad health

SOA Multiple Deprivation (md=17.8), 2001 Prob. Transitioning from Good to Bad Health ( ) Cross-level interaction of individual & area effects (Response: Transitioning good to bad health) Owner Occupier Social renter Private renter

Modelling probability changing from bad to good health

Cross-level interaction of individual & area effects: (Response: Transitioning bad to good health) SOA Multiple Deprivation (md=17.8), 2001 Owner Occupier Social renter Private renter Prob. Transition from bad to good health ( )

Ratio of SOA Deprivation 2001 compared 2011 (Area change with respect to deprivation) Owner Occupier Social renter Private renter Cross-level interaction of individual & area effects: (Response: Transitioning bad to good health) Prob. Transition from bad to good health ( )

SOA Multiple Deprivation (md=17.8), 2001 Prob. Transition from bad to good health ( ) Changed tenure, not changed address Changed tenure, changed address Not changed tenure, changed address Not changed tenure, not changed address Cross-level interaction of individual & area effects (Response: Transitioning bad to good health)

Summary Results (1) Age decreases the probability of transiting from bad to good health, and of remaining in good health in Age slightly decreases the probability of remaining in good health in , & increasing probability remaining in bad health in Males less likely to stay good, more likely to transition from good to bad health ( ) Catholics more likely to transition from good to bad health; & more likely stay bad health ( )

Summary Results (2) Those in owner occupation are more likely to remain in good health ( ); ditto transition from bad-to- good health ( ) Those in social renting are more likely to remain in bad health ( ); & generally transition from good-to-bad ( ) Not changing tenure more likely to remain good health - irrespective of changing address ( ) Changing tenure less likely to remain good health, & more so if have changed address ( ) Probability of transitioning from bad to good highest for those both changed tenure & address, other patterns more complex depend on area deprivation

Conclusions (1) Other model results reassuring – finding similar/consistent patterns with different specifications of Y-variable (consistent, logical & plausible results) Seeking to model considerable complexity: transitional states, compositional & contextual effects, & cross-level interactions – possible because of large & rich variable detail of NILS We think self-reported general health is a good Census question that captures people’s well-being / happiness – but will do more investigation/work – Also need to give some more thought to harmonizing categories to 3 categories

Conclusions (2) Evidence of selective placement of the (un)healthy in different tenures / spatialities – Implications: tenure and spatial mobility (or its lack) linked to social residualisation Can’t assign causality/directionality between health/tenure, or tenure/health – Requires quite different research designs (c.f. Smith & Easterlow, 2005)

Future Work Will look at limiting long-term illness (LLTI) transitions Unified multinomial models of the different health transitions simultaneously Will compare general health / LLTI with NI 2011 chronic illness question

Acknowledgement The help provided by the staff of the Northern Ireland Longitudinal Study and the NILS Research Support Unit is acknowledged. The NILS is funded by the Health and Social Care Research and Development Division of the Public Health Agency (HSC R&D Division) and NISRA. The NILS-RSU is funded by the ESRC and the Northern Ireland Government. The authors alone are responsible for the interpretation of the data and any views or opinions presented are solely those of the author and do not necessarily represent those of NISRA/NILS. NILS-RSU Contact Details Web: