All the studies were undertaken with populations in flood risk areas where flooding had occurred in recent years. The locations were spread across seven of the eight Environment Agency Regions and drew upon different populations in those areas. The 30 interview locations used in the Intangibles Survey and the target number of interviews set and achieved at each location are shown in Appendix 1. At the time the studies were undertaken, the Environment Agency did not hold accurate or up to date records of properties where flooding had taken place in the different areas. The Agency did have a list of ‘at risk’ properties, both residential and non-residential, used for flood warning purposes. However, this information provided only a very approximate guide to the addresses where flooding had taken place. Thus, none of the studies was based on a probability or random sample because of this lack of information on flooded properties to serve as sampling frames, and the process of fully screening populations to develop such sampling frames was too expensive and time consuming to be employed.
In the Intangibles Survey, reports on the Easter 1998 and autumn 2000 floods, were used to draw up an initial list of locations where a substantial number of properties were believed to have been affected by flooding. Much flooding in England and Wales occurs in small pockets and to scattered properties but in order to facilitate the organisation of fieldwork, areas where at least 20 properties were thought to have been flooded were selected. Since the Agency’s at risk data base offered only approximate information on where flooding had occurred, 24 of the 30 locations included in the Intangibles Survey were partially pre-screened through site visits to identify more accurately the scale and extent of flooding in the area. Detailed information had already been obtained for the remaining six sites. On the basis of the site visit data lists of potential addresses for the survey were drawn up.
The Intangibles Survey involved two separate samples. First, it required a large sample of residents who had experienced flood events inside their homes since April 1998. Then, for comparison, a second sample was required of people who lived in the same areas and were also at risk but had not been flooded inside their homes in the recent flood event, although their gardens and driveways might have been affected. These samples were needed in order to investigate whether actually being flooded inside the home led to long-term psychological effects, and if these were therefore not experienced by those who were not seriously affected by the flooding.
The Warnings study was carried out in two phases in order to take advantage of a Post Event Survey being undertaken for the Environment Agency involving the main areas where flooding had most recently occurred (in 2003-4). In the two phases of the Warnings Survey, interviews were conducted in all eight Environment Agency Regions in England and Wales and in a very large number of different locations (Appendix 1). For the first phase of the survey the Environment Agency provided a sample list of addresses and the survey aimed to interview all those listed as affected by the most recent event: as such it was a census rather than a sample survey. The addresses in the areas selected were pre-screened to establish their eligibility. Of 283 addresses provided, 215 were found to be eligible for inclusion in the survey and 168 interviews were achieved, a response rate of 78%. Phase One of the Warnings Survey covered households that had experienced flooding inside the home but also those who only had other parts of their property (including gardens and drives) flooded and even some properties that were not affected by the flooding in any way.
As both this Phase One of the Warnings Survey and the Intangibles Survey had covered the areas where substantial numbers of properties had been flooded recently, the Second Phase of the Warnings Survey had to draw on a large number of areas where only a small number of properties had been affected. All 26 Environment Agency areas were approached by the researchers and asked for property details and maps of locations affected by flooding since September 2000. Eleven areas provided information on recent property flooding in over 40 different locations. This information was used by interviewers to identify properties affected by flooding. In Phase Two of the Warnings Survey, because the focus of the survey was on damage reduction, interviewers were asked to obtain 80% of their interviews at residential properties that were flooded inside the dwelling. The remaining 20% were to have experienced flooding to ‘built property’ i.e. flooding to their garage or outbuildings. Those with only gardens or driveways flooded were excluded.
In the Intangibles Survey and in Phase Two of the Warnings Survey, interviewers were given target quotas of the number of interviews to achieve within an area. The number of addresses given to the interviewers was restricted to ensure that the interviewers could exercise little choice over the respondents they interviewed. Given this sampling approach, it was not possible to calculate response rates for these studies.
The Lower Thames Survey focuses on a reach of the Thames extending from Datchet to Walton Bridge within the Lower Thames study area where there are a number of flood prone properties along the river banks and on islands in the Thames. For the survey, from an initial list of 39 possible localities, 13 specific areas were chosen to represent the reach in the survey. The choice of areas to include was informed by the residents’ level of flood risk (a 1 in 50 return period or less), the possible feasibility of providing community based flood risk reduction, the focus of the survey within each of the areas and a range of possible combinations of such approaches, a spread of localities along the river to represent the reach, and areas providing a sufficient numbers of households in order to maximise the chances of recruiting the target number of respondents within the limited time available for the fieldwork. Each locality was tightly defined by street and house address and the 13 areas provided nearly target 500 addresses from which the interviewers were to recruit respondents. Thus, as interviews were undertaken in approximately half the households available, it is likely that the respondents responses are a valid reflection of the residents across the localities chosen. Interviewers were instructed to recruit one adult member per household with no restrictions placed on gender, age or social grade.
In the Intangibles survey, the targets were to achieve 1,000 interviews with flooded households and nearly that number was achieved (983) The target for those at risk, 500 was exceeded with 527 interviews achieved. In the Warnings Survey, the target was to achieve at least 400 interviews, with a target of 300 in Phase Two. The number of interviews achieved in that Phase fell a little short of the target. In the Lower Thames Survey, the target number of 200 was achieved with 206 interviews completed. In all three surveys, only one interview was conducted with an adult aged 18 and over in each household to avoid clustering effects.
In the Intangibles Survey, different questionnaires with a core of common questions were used for the flooded sample and the at risk sample. The flooded sample were asked detailed questions about the health and social impacts of the flooding which were not relevant to those unaffected by flooding.
In the two Phases of the Warnings Survey, slightly different versions of the questionnaire were employed. The Phase One questionnaire contained more detailed questions about the flood warning service required by the Environment Agency for the evaluation of its service. These questions were not included in Phase 2. This allowed additional more detailed questions about flood prevention actions to be included. Thus, although there was a large core of common questions asked in the two phases, a few different questions were introduced in each phase.
Fieldwork was undertaken by two different market research companies, Market Opinion Research International (MORI) and BMRB. BMRB has had a contract with the Environment Agency to undertake regular annual public attitude surveys including post event surveys since 1997 and has therefore substantial experience of designing questionnaires and undertaking fieldwork with flood affected and at risk populations. MORI has undertaken surveys at coastal and river sites for FHRC over a number of years. The market research companies assisted in finalising the questionnaires and briefed and supervised their trained interviewers through their regional supervisors.
FHRC has, for some surveys recruited, briefed and supervised its own team of fieldworkers. This has the advantage of keeping the researchers in close contact with the fieldwork areas and respondents. However, for the surveys included in this analysis, because of the large number of interviews involved, the wide distribution of the fieldwork sites, and the very short time available for the fieldwork in the Lower Thames area, this approach was not considered feasible. The researchers had gained insights into some of the areas included in the samples through screening activities in the Intangibles Survey and through prior qualitative research, focus group discussions and pilot interviews undertaken at many of the survey locations included in the studies.
2.5.4 Data processing and checking
The market research companies were responsible for data checking and data entry. They provided FHRC with a disk with the data entered into an SPSS database. In the Intangibles Survey, the researchers were given limited access to the questionnaires but were able to read them and check the data entry directly with them. In Phase 2 of the Warnings Survey, FHRC were given the questionnaires to read and check. In all other cases, this was not possible because of the exact nature of the confidentiality agreement included in the survey.
Data from the three surveys were analysed by FHRC researchers using SPSS. This report mainly provides insights from new analyses of these data. However, there are points at which it is useful to include data that have already been presented elsewhere, either for comparison with new analyses or in support of the argument being made in this report. It was decided to analyse and present the findings by themes across the data from the three surveys where available. This results in some de-contextualising of the data, particularly that from the Lower Thames which is focused on one particular locality. For the other national surveys which were undertaken in many areas covering many different events, it has not been possible to take the detail of the local context into account.
The surveys, particularly the large Intangibles data sets, provide a rich source of data. However, it was not possible to analyse the data as fully as we would have liked across some themes because of time restrictions. Therefore, some themes have been analysed more thoroughly than others. In some cases, only bivariate analyses have been undertaken where it would otherwise have been advantageous to also have undertaken multi-variate analyses. The Lower Thames Survey offers more limited opportunities for analysis because of the size of the sample.
Generally, as a guide in the analysis, sub-group analysis has been restricted to those sub-groups containing 40 or more cases. Where exceptionally smaller groups have been included in the analysis, this is normally indicated in the text. Where reference is made to significant differences or associations in the text, this indicates that appropriate statistical tests have been applied and have been found to be statistically significant at at least the p<0.05 level and these are normally reported in the text. Where differences or associations are not statistically significant but are potentially interesting or important, this is pointed out.
Table 2.2 summarises the main independent variables available for analysis in some, but not in all cases, of the three surveys. Most of these variables are self-explanatory or are described in detail at the point where they are first used in the analysis. Background information and a more extensive explanation is provided in section 3 for a few of the variables.
2.6 Summary The data offered in the three FHRC data sets are very different from the case study data available to our German and Italian partners. The FHRC data were originally collected and analysed for other purposes, based on particular theoretical frameworks. These data sets have been further analysed to augment our understanding of flood event experiences, preparedness and response to elaborate our understanding of the social and health impacts of flooding in the UK. The main independent variables used in the analysis for this report, are the most appropriate available from the earlier studies which address the aims of Task 11.
The first two UK data sets (from the ‘Intangibles’ and ‘Warnings’ surveys) cover a range of up to 30 locations and many different flood events in England and Wales between 1998 and 2003. The third data set (‘Lower Thames’ survey) focused on a particular location along the River Thames and a single key flood event in 2003. All surveys were preceded by a qualitative research stage involving focus group discussions. The Intangibles Survey involved two separate samples: flooded and ‘at risk’.
The Warnings study was carried out in two phases in areas where flooding had most recently occurred. The surveys mainly covered households that had experienced flooding inside the home but also some who only had other parts of their property (including gardens and drives) flooded and even a few properties that were not affected by the flooding in any way.
The Lower Thames study was almost exclusively of those at risk since very few of those in the survey had been flooded in the event of January 2003 that affected their area.
In the Intangibles Survey, different questionnaires with a core of common questions were used for the flooded sample and the at risk sample. In the Warnings Survey, slightly different versions of the questionnaire were employed in two Phases of the survey although there was a large core of common questions. The Phase One questionnaire contained more detailed questions about the flood warning service while Phase 2 contained detailed questions about flood prevention actions.
Fieldwork was undertaken by two different market research companies, Market Opinion Research International (MORI) and British Market Research Bureau (BMRB), who were also responsible for data checking and data entry. Data from the surveys were analysed by FHRC researchers using SPSS, sub-group analysis was generally restricted to those sub-groups containing 40 or more cases. For the Intangibles and Warnings data it has not been possible to take the detail of the local context into account due to the large number of survey locations.
Table 2.2: Main independent variables used in the analysis