This paper draws on data from the second cross-sectional survey of GoWell, a long-term study of 15 deprived communities in Glasgow undergoing major housing investment and area regeneration over a 10-15 year period. All but one of the 15 study areas is in the 15% most deprived areas in Scotland on the Scottish Index of Multiple Deprivation, with six of the areas in the very worst, or most deprived, 5% (see [34
] for more details about the GoWell study design and methods). The GoWell areas are broadly representative of deprived areas in Glasgow [35
Sampling and recruitment
The 15 areas varied in population size and to obtain sufficiently powered samples for each area we used a mixed sampling strategy (involving a census in six areas and random stratified cross-sectional sampling in nine areas). We selected households in each study area from the addresses in the most recent version of the Royal Mail Postal Address File within the postcode units that define the study areas. Where selected homes had more than one householder, only the householder with the most recent birthday was interviewed (i.e. one householder was interviewed per selected household). Households selected for cross-sectional surveys were posted information sheets and letters inviting them to take part. Study information leaflets were produced in English, Arabic, Urdu, Cantonese and Turkish. Fieldworkers made up to five attempts to contact selected homes in person to seek consent to participate.
A face-to-face questionnaire lasting around 40 min was verbally administered by fieldworkers at participants' homes with responses recorded using CAPI (Computer Assisted Personal Interviewing). Interpreters were available or assistance from other household members obtained if the interviewee did not speak English (or one of several of the languages common amongst UK non-English speakers in which some of the contracted fieldworkers were also fluent). The achieved sample of householders was 4,657, with an overall response rate of 47.5%, which is good for a survey of deprived area populations.
Ethics approval for the study was given by the NHS Scotland B MREC (no. 05/MRE10/89).
The outcome of interest here is mental wellbeing. To consider how regeneration might impact upon wellbeing, the other independent variables are organised into groupings related to residential circumstances and psychosocial factors, and perceptions of residents' house and neighbourhood.
Mental wellbeing was assessed using the Warwick-Edinburgh Mental Wellbeing Scale (WEMWBS) [27
]. The scale has good psychometric properties [36
]. In the validation of WEMWBS, Tennant et al. [27
] examined the relationships between other measures of positive mental health (e.g. WHO-5, SPWB, PANAS) and mental ill health (GHQ- 12, PANAS). They reported relatively high correlations with the other positive wellbeing measures (correlations ≥ 0.7) and moderate, negative correlations with the GHQ- 12 (r = -0.53) measure of mental ill health. They concluded that 'respondents scoring the same on the GHQ-12 had a range of WEMWBS scores, so although lower WEMWBS scores tend to be associated with higher GHQ-12 scores, one is not simply the inverse of the other. The two scales are therefore not measuring the same thing' [27
The scale has 14 items covering: positive affect (feelings of optimism, cheerfulness, relaxation); positive functioning (energy, clear thinking, self-acceptance, personal development, competence and authority); and relationships with others. Respondents are asked to what extent they have been feeling that way over the past two weeks. Responses are summed, with higher scores indicating greater wellbeing. Internal consistency of the scale in this study was high (Cronbach alpha 0.92). Low, average and high wellbeing in the current analyses were defined as: < 1 standard deviation below the population mean in Scotland (50.7), +/-1 standard deviation (9.9) around the mean, and > 1 standard deviation above the mean, respectively. We chose to trichotomise the WEMWBS scores to better show the associations with average and high wellbeing, rather than the effect of a one-point increase in the continuous wellbeing score.
We included the following socio-demographic factors in our analysis: gender, ethnicity, age, household type (young adults, single-parent family, two-parent family, or household with two or more adults over 64 years of age), and educational attainment. Economic status was assessed using three variables: employment status, income source, and regular access to a private vehicle.
In this analysis we used the first item of the SF-12v2 Health Survey [37
]--a self-reported assessment of health from poor to excellent--and a question about whether the respondent considered themselves to have a long-standing illness or disability.
Residential circumstances and psychosocial benefits
Here we include housing tenure and the length of time the respondent had lived in their home and neighbourhood, as well as their perceptions of the internal and external reputation of the neighbourhood, whether their neighbourhood had changed for better or worse in the previous two years, satisfaction with their home, neighbourhood and landlord, and whether their home or neighbourhood made them feel they were doing well in their life.
Tenure can be considered to convey material and psychosocial advantages [38
]. Housing tenure is included here (rather than as a socio-demographic factor), because for our study group, in deprived areas, we believe tenure has greater importance as a psychosocial factor than as a structural or material factor. Whilst we recognise that housing tenure, in particular ownership, can confer financial advantages, for marginal home-owner groups it often results in financial disadvantage and psychological stress [39
Furthermore, and perhaps more importantly, the preference for private consumption, possession and ownership of goods--especially housing--has been explained by sociologists as a 'cultural phenomenon rather than something inherent to the objects themselves', providing benefits of identity, autonomy and ontological security [[40
], p.328-9]. Thus, housing is not simply economic consumption but also 'symbolic consumption' enabling the external communication of identity [41
]. This has assumed ever greater importance as home ownership has become 'normalized' in British society and discourse as a positional good which conveys status, normal aspiration, and notions of being a good citizen and parent [42
Length of residence is also included here as it serves as a proxy for place attachment, the two having been shown to have a strong relationship in recent research in the UK [43
]. Place attachment itself confers psychological benefits for residents by giving lives meaning, value and significance and contributing to identity and self-esteem [44
Dwelling type and perceived housing quality
Residents provided information about type of residence (house, high- or low-rise flat, etc.), the existence of internal and external problems, access to a garden, perception of overall condition of the residence, and specifically about insulation, the external appearance and repair, and the condition of their front door. They were also asked about their satisfaction with any improvements that had been made to their residence. We have included dwelling type here for similar reasons to those mentioned above in relation to housing tenure. Dwelling type also contributes to psychosocial benefits pertaining to identity, status and self-esteem, with strong general preferences for houses over flats, and for larger properties with more rooms and additional space over and above smaller properties [46
]. This is reflected in the strong emotional content involved in the house marketing and selection process [47
Perceived neighbourhood quality
Residents were asked whether the following environmental incivilities in their local neighbourhood were a serious or a slight problem, or not a problem: vandalism, graffiti, other deliberate damage to property or vehicles; abandoned or burnt-out cars; rubbish or litter lying around; vacant or derelict buildings and sites; and untidy gardens. The number of serious problems was summed and then coded as: 3 or more serious incivilities, 1-2 serious incivilities, or no serious incivilities.
Quality of local amenities (play areas; schools; youth and leisure services; shops; banking or financial services; childcare or nurseries; health centre or GP service) was assessed. The number of amenities assessed as 'poor' or 'very poor' was counted and coded as 3 or more, 1 or 2, or none.
Residents were asked to rate the attractiveness of the buildings and environment in their neighbourhood and whether it was quiet and peaceful. Neighbourhood quality was based on the number of 'fairly' or 'very poor' responses to questions about attractive buildings, attractive environment, and a quiet and peaceful environment.
Method of analysis
Multinomial logistic regression was used to assess the associations between housing and neighbourhood environment and average or high wellbeing compared with low wellbeing, adjusting for individual socio-demographic factors, and taking account of clustering within GoWell areas. STATA Version 9 was used for all analyses [48