Scottish Government

CHAPTER THREE METHODS

3.1 Capturing the complexities of human responses to flooding and assessing 'what works' in terms of flood prevention and warning schemes requires a flexible and wide-ranging methodology which embraces both quantitative and qualitative approaches. The methodology adopted in the project comprises four distinct but complementary strands:

1) Socio-demographic profiling of flood risk areas in Scotland and selection of study areas
2) Questionnaire survey of households in flooded/flood risk areas
3) Focus group discussions with residents in flooded areas
4) Semi-structured interviews with institutional stakeholders

Socio-demographic profiling of flood risk areas and selection of study areas

3.2 In order to establish what type of communities are most exposed to flood risk, and to ensure that the selection of study areas is representative of this, the socio-demographic characteristics of the populations living in flood-risk areas across Scotland have been identified from the 2001 Census of Population. This information has been used for two purposes:

• To assess whether potentially vulnerable groups are over or under represented in flood-risk areas.
• To assess how representative the returns in the household survey are of flood-risk areas across Scotland as a whole.

3.3 Socio-demographic profiling is necessary because certain types of individuals

and households may be more vulnerable to the impacts of flooding than others, for example the elderly, lone parents, children and people whose activities are limited by ill-health or disability ( DEFRA, 2006). In addition, housing tenure may influence the nature and severity of the impacts of a flood, for example local authority landlords often have limited capacity to rehouse flood victims whereas owner occupiers may be able to access accommodation paid for through their buildings insurance.

3.4 It is also important to establish whether potentially vulnerable groups are over

or under represented in flood-risk areas across Scotland (cf Fielding and Burningham, 2005). On the one hand, it could be postulated that local authority housing and lower income households in the private sectors are more likely to be located in flood-risk areas because the land and housing may be cheaper. On the other hand, some attractive riverside locations may act as magnets to higher income households.

Methods

3.5 The key challenge in profiling the socio-demographics of small areas such as flood-risk zones is matching the spatial units used to report Census results to outlines of flood risk. The smallest geographical units for which Census data in Scotland are available are Output Areas. There are 42,604 Output Areas across Scotland with, on average, around 50 households and a population of 120. Because Output Areas are designed to have approximately equal populations, they are of substantially greater geographic extent in rural areas reflecting the lower population density found there.

3.6 Analysis using a Geographical Information System ( GIS) indicates that there are 287 Census Output Areas that are completely contained within flood-risk areas (as defined by the IH 130 indicative flood risk maps and, where available, more recent maps held by the relevant local authority). Although this approach excludes significant areas of flood-risk, most of these are relatively sparsely populated and it has the advantage of not spuriously introducing any Census data from outside flood-risk areas. The 287 selected Output Areas are composed mainly of clusters in Angus, the Scottish Borders, Edinburgh, Glasgow, Moray and Perth & Kinross.

3.7 The main disadvantage with this approach is that the 287 Output Areas largely represent towns and cities at the expense of remoter areas. For this reason, the socio-demographics of the 287 'flood-risk' Output Areas should not be compared against Scotland as a whole, as this would simply reflect the substantial socio-demographic differences between urban and rural areas. For example, 27.5% of households in the 287 flood-risk Output Areas rent their home from the local authority compared to only 21.6% in Scotland as a whole. At first sight, this could suggest that local authority housing is more likely to be built in flood-risk areas, but when large Output Areas in remoter parts of Scotland are excluded the figure for comparison rises to 27.0%, indicating virtually no difference between flood-risk and non-flood-risk areas.

3.8 Consequently, the analysis presented here compares flood-risk areas in built-up areas with urban areas not at risk of flooding. The possibility that the results would be different in remoter areas cannot be discounted.

Socio-demographics of flood-risk areas

3.9 In urban areas, the socio-demographic profile of flood-risk areas is generally very similar to that of areas not at risk of flooding. For example, 27.5% of households in flood-risk areas rent from a local authority compared to 27.0% overall (Table 3.1). The corresponding figures for employed persons in 'elementary' ( i.e. unskilled) occupations are 14.5% and 14.1% respectively. Lone parents with dependent children and households containing a person with a limiting long-term illness both form slightly lower proportions of households in flood-risk areas (7.5% and 37.6% respectively) than elsewhere (8.3% and 38.4% respectively). Another minor difference between flood-risk and non-flood-risk areas is that older people are slightly more likely to live in the former (17.3% of people in flood-risk areas are aged 65+ versus 15.6% elsewhere).

The representativeness of the household survey returns

3.10 Previous experience indicates that retired people and those of higher socio-economic status are more likely to participate in household surveys. While older people (aged 70+) are inevitably over-represented in our returns (22.2% of survey returns versus 11.2% of the Scottish population), lower socio-economic groups are relatively well represented. For example, 9.9% of our survey returns fall into the 'elementary' ( i.e. unskilled) occupational category versus 12.7% in Scotland as a whole, a difference of only 1.8%. The under-representation of local authority tenants is slightly greater (15.8% versus 21.6% nationally) but again is not as substantial as is sometimes the case with this type of survey.

Table 3.1 Socio-demographics of flood-risk areas

*For flood-risk areas,'urban Scotland' is defined by the Census Output areas that are completely contained within flood-risk zones (n=287; mean=1.90ha; median=1.59ha; total pop=30,089).
** For areas not at flood-risk, 'urban Scotland' is defined as Census Output areas less than 3ha on the basis that this gives a size distribution of Output Areas comparable to that of those in flood-risk areas (n=25,311; mean=1.64ha; median=1.63ha; total pop=2,770,719).

Selection of flood risk areas

3.11 Having listed Scotland's major floods over the period 1993-2005 and coded them as fluvial (over-topping of river banks), pluvial (surcharging sewers combined with overland flow) and coastal, a representative sample of flood risk areas was compiled. Floods of differing origins were sampled from small urban areas (Brechin, Elgin, Forres, and Hawick), large urban areas (Glasgow, Edinburgh and Perth) and scattered inland and coastal rural locations. The number of households sampled in each of these locations is given in Appendix B; Table 1. The number of households located in large urban, small urban and rural areas is given in Appendix B; Table 8. The number of households recording each type of flood (fluvial, pluvial and coastal) is given in Appendix B; Table 9.

Questionnaire survey of households in flooded/ flood risk areas

3.12 In order to generate information relating to flood impacts (objective 1) and attitudes to floods, knowledge and awareness of floods and behaviour during floods (objective 2), a survey was undertaken on households which had been flooded and households located within flood risk areas but not flooded (as defined on the first generation IH 130 indicative flood risk maps and, where available, more recent maps held by the relevant local authority).

3.13 The survey was undertaken via a questionnaire designed to be completed by respondents with minimal or no assistance from a member of the project team. In urban areas the questionnaire was delivered by hand to each household and usually collected the next day, with the team member occasionally assisting in its completion. Collection was focused on evenings and weekends to maximise the number of returns; two return visits being made where necessary. FREEPOST envelopes were provided when follow up visits still failed to generate a response.

3.14 Address lists of households which had been flooded and households, not flooded, but located in flood risk areas were generated in advance of visiting each survey site. For smaller urban areas (Brechin, Elgin, Forres and Hawick) questionnaires were delivered to all households within the historic flood envelope. For larger urban areas (Edinburgh and Perth) a sample was drawn of properties closest to the river with a variable upper limit in any one survey site of c. 250-550 properties. Scottish Water provided addresses of properties in the east end of Glasgow which experienced pluvial flooding in 2002 from which a representative sample of 262 properties was drawn. Across all survey sites a maximum of 10% of upper floor properties was included in the total sample. The number of households visited in each survey site is listed in Appendix B; Table 1.

3.15 For scattered rural inland and coastal locations, the questionnaire was delivered by post, respondents being provided with an envelope for return by FREEPOST. A sample of addresses of flooded households in Orkney, Shetland, Culloden, Menstrie, Eyemouth, Dunoon, Newcastleton and Corpach was generated from information in local authority 2003 and 2005 Biennial Reports and telephone contact with council staff and SEPA representatives. Reconstructed outlines of flood extent were often available, which allowed accurate selection of properties.

3.16 The questionnaire (which will be provided with a statistical annex due to be published on the SE website in summer 2007) was divided into five sections:

• Your experience of flooding (information on the actual experience of being flooded).
• Impacts of the flood (immediate and lasting impacts of the flood).
• Living with floods (awareness of flood risk and measures to reduce the impacts of flooding).
• Your thoughts and opinions (views on flood protection measures and where responsibility for providing protection lies).
• About you and your household (information on composition of household, housing tenure, occupation, educational qualifications and household income).

Households which had been flooded were invited to complete all five sections. Households which had not been flooded, but which were located within the flood risk area, were invited to complete the last three sections.

3.17 A draft questionnaire was piloted in Brechin in order to:

• assess what sampling strategy would generate an appropriate proportion of flooded versus 'not flooded but at flood risk' households;
• assess response rates from different risk groups, in particular households that lived within an historic flood envelope versus those outside the envelope; and
• test the suitability of the questionnaire, for example in terms of topics covered, question design, routing and overall length.

3.18 Having delivered questionnaires to 67 properties in Brechin which yielded 46 returns (68.7% response) it was concluded that sampling within the historic flood envelope would generate a representative sample of households living in a flood risk area. Many properties which had been flooded, but had subsequently changed occupancy, yielded a robust sample of households 'at risk but not flooded' alongside other properties within the historic flood envelope which were 'at risk and had been flooded' but with no change in occupancy. The questionnaire was revised in light of the Brechin returns following discussions within the project team plus contributions from the Project Steering Group.

Focus group discussions

3.19 Focus groups provide a means to elucidate detailed and nuanced information on people's attitudes and motivations in much more depth than is possible via a questionnaire survey.

3.20 Focus group participants were recruited from the household questionnaire survey. Respondents to the survey who had been flooded were asked if they would be willing to participate in a focus group to explore some of the issues in greater depth. Expenses of £15 was paid to each respondent who attended a group. Focus groups were held in Elgin, Forres, Edinburgh, Glasgow and Perth with 10-12 respondents being invited to each venue to explore their views on:

General attitudes towards flooding and flood risk:

• the balance of responsibility for mitigating flood losses - individuals versus public authorities
• the range of options for reducing losses available to individuals living in flood risk areas
• the types of flood proofing and/or action individuals would be willing to take to mitigate flood losses
• individual's attitudes to flood insurance

Experiences and perceptions of flood prevention schemes and flood warning schemes:

• awareness of flood alleviation schemes either in place or being planned
• changes in behaviour due to campaigns designed to raise awareness of these schemes
• perceived effectiveness or potential effectiveness of these schemes and how far they instil confidence or anxiety
• the accuracy and reliability of warnings
• the effectiveness of the warnings which resulted in their evacuation

3.21 Each focus group, held in a hotel, lasted around 90 minutes with 5-10 of the invitees attending alongside two or three members of the project team. Each session was recorded onto audio tape and then transcribed to provide a verbatim record.

Semi-structured interviews with institutional stakeholders

3.22 The purpose of the interviews with institutional stakeholders was to examine 'what works' in terms of flood prevention and flood warning (particularly in relation to the social aspects of such campaigns) and engagement with the community and other stakeholders in flood risk areas. These issues were discussed with practitioners at both national (Scottish Executive, SEPA) and local levels (local authorities). Interviewees and dates are shown in Table 3.2 and the issues covered in Appendix C. Most interviews lasted around 90 minutes and, with permission, were recorded.

Table 3.2 Key institutional stakeholders' interviews and dates

Data analysis

3.23 The returns from the Household Survey were coded and then analysed using SPSS (a statistical package with routines specifically designed for analysing social science data). Summary tables, often involving cross-tabulation of responses against socio-economic status, provided concise and efficient outputs of the main findings. For some responses it was possible to determine statistically significant differences between sub-groups (eg 'flooded' and 'not-flooded') using a Chi-Squared test. The transcripts of the Focus Group discussions were examined following the same sequence of issues as for the Household Survey (impacts of flooding, living with floods and managing flood risk). Verbatim quotations from participants' contributions were extracted from the transcripts to provide illustrations to each of these themes. In a similar manner the tape-recorded semi-structured interviews with institutional stakeholders were examined to collate answers across a range of questions and provided quotations to amplify specific findings.

Summary

3.24 This chapter describes the methods adopted in this study. Households in locations variously flooded between 1993 and 2005 by rivers, coastal storms or failed urban drainage comprised the target population. A self-complete questionnaire-based survey was designed to elicit responses on flood impacts and the experience of living with floods from households which, during the study period, had either been flooded or were at risk of being flooded. The draft questionnaire was piloted on households in Brechin within an area flooded in 2002. In a revised form, it was subsequently delivered by hand to a sample of households in small urban areas (Elgin, Forres and Hawick), large urban areas (Glasgow, Edinburgh and Perth) and, by post, to scattered inland and coastal rural communities. Focus groups were organised at five locations with participants recruited from respondents to the household survey who had been flooded. The group's experiences and attitudes were recorded onto audio tape and then transcribed to produce a verbatim record. Semi-structured interviews (also recorded) with a range of institutional stakeholders were designed to elicit views on 'what works' in terms of flood prevention, flood warning and engagement with those who live in flood risk areas. Socio-demographic profiling of the flood risk areas yielded a slight over-representation of single pensioner households and a slight under-representation of lone parents with children, but in other categories lower socio-economic groups were well-represented when compared with urban Scotland as a whole.