In regards to quantitative evaluations, Stafford et al. (2008) investigated the impact that the New Deal for Communities (NDC) initiative had on the outcomes reported in the table above. These were assessed by sex, age and educational and ethnic group. The study focused both on absolute improvements in health in the NDC areas, and on whether there have been differential changes in health across demographic or socio-economic groups over time within NDC areas. Three questions were addressed: (1) Have there been overall improvements in health or its determinants in the deprived areas targeted by the NDC initiative? (2) Have there been differential changes in health or its determinants for different socio-economic, ethnic, gender and age groups within NDC areas? (3) To what extent do any differential changes mirror what is happening in similarly deprived non- intervention areas?
The design of this study was quasi-experimental and repeated cross sectional. Stafford et al. (2008) made use of longitudinal survey data collected by MORI in 2002 and 2004 in each of the 39 areas that were awarded NDC funding. Only residents who remained in the NDC or comparator area were included in the analysis. In regards to the sample used in the study, women were over-represented, approximately 20% were from non-white ethnic backgrounds and over a third had no formal qualifications. The sex, age and ethnicity profiles of residents in comparator areas were similar to those residents in the regeneration areas in order to ensure these areas were as similar as possible to the regeneration areas with the exception that they have not experienced the programme.
A further quantitative evaluation which adopted a different approach to evaluating the health impacts of the New Deal for Communities programme in the West Midlands was that conducted by Cotterill et al. (2008). This study assessed whether the NDC had improved the health of communities by looking at trends in specified outcomes in time periods before and after the initiative commenced. The authors used a quasi-experimental method that attempted to provide an indication of the counterfactual, i.e. what would have happened in the absence of the regeneration initiative. To do this they constructed virtual area models to be used as control areas not in receipt of NDC funding. These areas (which do not exist as real discrete entities) were constructed using 2001 census output areas (COA). Thus, the authors used national census data to identify COAs out with NDC areas that were as similar in socio-material and demographics as possible to COAs within NDC areas, and combined a range of the non-NDC COAs together to construct a virtual area that was as similar as possible to the NDC areas under consideration. The authors sought to match non-NDC COAs with NDC COAs with regard to age, sex and two further central characteristics: deprivation and
34 ethnicity profiles. In addition the non-NDC COAs were all drawn from areas within the same local authority as the NDC areas. The authors used the Townsend score as a measure of deprivation and the ethnicity profile of each COA was constructed using the measure ‘percentage white in the population’. It was decided that using a ‘white and non-white’ classification was the best, albeit imperfect option for matching areas as breaking non-white groups down by specific ethnicity would have given smaller numbers. All-cause mortality data and hospital episode statistics were then used to compare the health experiences of the NDC areas and the control areas. From these data the authors focused the analysis on two ‘illnesses’, namely coronary heart disease (as it is the most commonly targeted disease of NDC initiatives) and unintentional injuries (accidents) due to the anticipated short lag time between intervention and impact. This is important as if only short –term follow up data are available then it follows that outcomes which are anticipated to respond quickly (following the intervention) should be employed.
Qualitative evaluations
In terms of qualitative evaluations, a study by Mathers et al. (2008) explored resident non- participation in a New Deal for Communities initiative in an area of Birmingham, UK. The findings of the study touched upon mental health issues relating to a regeneration initiative. Ethnographic methods were used to gather information from residents in an NDC area. Three particular methods were employed. Firstly, direct, first-hand observance of daily behaviour recorded via field notes; informal conversations with residents recorded via field notes; and, longer, more structured interviews (Mathers et al. 2008).
A further qualitative study was conducted by Gosling (2008). The main aim of the research was to gain an understanding of women’s experiences of social exclusion and urban regeneration in an area that had experienced both the Single Regeneration Budget (SRB) and NDC initiatives. The focus of the regeneration in this area was to generate new jobs and training, and to improve public spaces through the refurbishment of existing buildings, the demolition of others and their replacement with new houses and low-rise flats built by private contractors, and the transfer of housing stock to privately run housing associations. The author used qualitative methods that involved initial participant observation, preliminary focus groups and 21 semi-structured interviews with women aged 18 to 80 to explore personal understandings and experiences of regeneration and social exclusion. Gosling’s work is premised upon the suggestion that urban regeneration initiatives can undermine and destroy existing local community networks, which can significantly impact upon the lives of many women who rely on local support networks to help manage social exclusion, and who
35 are often more involved in the community than men, especially as mothers and volunteers (Gosling 2008).
The third qualitative study included here is by Thomson et al. (2003). This study considered the mental health impacts of a regeneration initiative in Glasgow. Thomson et al. (2003) assessed how the provision (and closure) of a public swimming pool and leisure facilities through a regeneration initiative impacted on the health of people living in two deprived areas of Glasgow (three miles apart). Fourteen focus groups were carried out 14-18 months after the pool opened (or closed). In Riverside a swimming pool and sports complex was opened in 2000 as part of a Social Inclusion Partnership Programme (SIP). However in Parkview, an area socio-demographically similar to Riverside and with the same level of deprivation (but not receiving SIP funding), a swimming pool facility was closed in 1999 due to upkeep costs. Parkview also had no regeneration programme in place. The design of this study is unique amongst the qualitative studies included here. The comparative element is a particular strength and represents a useful way to contrast differences between regeneration and non-regeneration areas using an in-depth approach in order to understand residents’ feelings about the area they live in.
Mixed-methods evaluations
A study by Huxley, Rogers and colleagues (2004, 2005, 2008) conducted an evaluation of the SRB area regeneration programme in a disadvantaged area in Manchester (UK) specifically in relation to mental health outcomes. This study aimed both to further understand the role played by urban regeneration in altering the degree and distribution of socio-economic variations in mental health and the impact of socio-economic and locality changes on mental health (Rogers et al. 2008). It hypothesised that the SRB would lead to more changes in the programme area than in a comparator area (Huxley 2005).
The design of the study was longitudinal using mixed methods, with an initial quasi-experimental quantitative phase followed up with qualitative interviews. For the quantitative phase the SRB area was matched using the Index of Deprivation with wards not subject to regeneration to compare the SRB area to an area where no such initiative existed (Huxley 2005). A sample of residents selected from the electoral register using computerized random selection was surveyed by post. The mental health and quality of life of the sample was measured before the SRB intervention had begun and 22 months thereafter. The main outcome measures used were the General Health Questionnaire 12, GP use for ‘nerves’, and the ‘satisfaction with health’ domain of the MANSA quality of life assessment. The MANSA is a brief quality of life assessment covering objective and subjective well-being in 8 life
36 domains including health (Huxley et al. 2004). Predictor variables of mental health that were employed included age, sex, marital status, ethnicity, car ownership, unemployment, housing tenure and restricted opportunities. In addition, respondents were also asked about their perception of the various types of improvements that had been planned using a 5-point categorical scale (a lot worse, worse, no change, a little better, a lot better).The qualitative phase selected 20 of the regeneration area respondents from the quantitative phase of the research in order to obtain further details about subjective views of the locality, effects of the urban regeneration programme, psychosocial well-being and perceptions of mental health (Rogers et al. 2008). This strategy could perhaps have been enhanced by also selecting a sub-section of the comparator area residents to compare how views differed among the two groups to obtain a more rigorous assessment of the programme. However, Rogers et al. (2008) make it clear that their primary interest is solely on the mental health impacts of the programme on the regeneration residents.
A further mixed methods study was carried out by Kearns and colleagues (2008). This study investigated the impacts of being housed in new-build socially rented properties on housing conditions, neighbourhood and social outcomes and the health and well-being of tenants. The full list of outcomes assessed is in included in Table 2-1 above. In addition to these outcomes the authors provide limited detail on independent variables used in analyses; choosing to mention only housing tenure, household type (e.g. lone parents, couples with dependent children) and location change. The authors studied the Scottish social housing investment programme as a whole, not just housing programmes involved in regeneration initiatives as in Scotland nearly 40% of the output from Scottish Homes urban investment in the year 1999/2000 (when the initial plans for this study were made) were in the Priority Partnership Area (PPA) phase of the SARPs programme, meaning that 60% of housing output was to occur either in urban areas not in receipt of regeneration funding or in rural areas. The authors thus noted that this offered the team involved in the study an opportunity to investigate whether the wider regeneration context produces health gains over and above housing investment in itself.
In terms of design, three survey waves were combined with two stages of in-depth qualitative interviews, implemented after survey waves 2 and 3. Wave 1 (baseline) and wave 3 (2 years after) were face-to-face interviews with 334 households who had been re-housed (intervention group) and 389 households who had not been re-housed (control group). Wave 2 was a postal survey one year after the baseline for the intervention group only. In-depth qualitative interviews were also conducted to explore the longer-term impacts of moving house and neighbourhood (n=28 for the first wave of qualitative interviews and n=22 for the second wave). The study focused on two
37 groups: (1) An intervention group, which consisted of people re-housed into a new general-purpose socially rented home let by a Registered Social Landlord; and (2) A comparator group, which comprised people residing in the same locality as the re-housed subjects, but who had not been re- housed from their existing dwelling. The intervention group was further divided into those who moved to another neighbourhood to get a new house (Re-locators) and those who moved house within the same area (Non- Re-locators). Lastly the 2 study groups were split between regeneration areas and other areas. The authors noted that at the time of the study these regeneration areas had evolved from PPAs to the Social Inclusion Partnership (SIP) phase of the SARP programmes. Three- quarters of the samples were in urban areas, and 44% (Wave 1) and 51% (Wave 3) were in SARP regeneration areas. The two groups were matched for location, household type and tenure, which according to the authors ‘worked well’ as for example, 91% of the intervention group and 99% of the control group were renting at Wave 1 of the survey, and 72% of the intervention group and 63% of the control group were families. However it was noted that whilst the groups are matched on certain criteria, they are not identical and differ in some respects, which limited the ‘purity of the study’ (Kearns et al. 2008). It is unclear which criteria that the intervention and comparator residents were poorly matched on. This could well represent a serious flaw in this study and undermines the ability and authority of the study to comment on what would have happened in the absence of the intervention.