Which particular rooms or areas were flooded was a significant factor in evacuation. There were significant correlations between the depth of flooding in particular rooms and the propensity to evacuate, and flooding was significantly deeper in the rooms of those who evacuated compared with those who did not (Table 5.8).
The extent of flooding in the home as measured by the number of main parts of the home affected by flooding (main parts being living room, bedroom, bathroom and kitchen) was a significant factor in evacuation. Where none of these main parts of the dwelling or only one were affected only a quarter reported leaving home; where two parts were affected the proportion was 68%, for three parts, 74% and where all the main parts of the home were flooded very few households, only 11%, managed to stay together in the home during and after the flood.
Table 5.8: Type of rooms flooded and evacuation : Intangibles Survey
% with room flooded
% who evacuated:
If room flooded
% who evacuated:
If room not flooded
Correlation between depth of flooding in room and evacuation
Other flood characteristics were also significant in the Intangibles study as drivers of evacuation behaviour. These include: speed of onset, the presence of contaminants, receipt of flood warnings, social characteristics and material resources
Speed of onset
Evacuation was less likely where flood waters were reported to have risen quickly (63% evacuated) rather than slowly (78%), or in-between (71%). This may be because residents did not have time to organise an evacuation in advance but it is surprising since most evacuations take place after flooding anyway. It is possible that in areas where flood waters rose fast, they also retreated quickly and caused less damage.
There are health fears associated with living in property believed to have been flooded by contaminated waters. Respondents were asked whether the floodwater contained sewage or other pollution. Those 68% who said yes were more likely to evacuate (compared with the 57% who said no and those who did not know, 54%). One of the main concerns for flood victims raised in qualitative studies following the 1998 and June 2000 floods was the fact that the floodwaters contained sewage and other contaminants (Tapsell et al., 1999; Tapsell and Tunstall, 2001). The perceived threat of diseases and risk to human health from the contaminants had caused anxiety, particularly where young children were living in the properties. In addition there was fear of contaminated drinking water, rat infestations and possible bacteria left in the materials of properties following the flooding. Many people spoke of the unpleasantness of having to clean up after the floodwaters had receded, and the smell which had remained in the properties for months afterwards.
Respondents who received a flood warning of some kind before the flood were significantly more likely to evacuate than those who did not (76% compared with 62%). It may be that the warning gave residents a better chance to arrange to stay elsewhere and to get people out of the house.
Various factors appear to have some relevance as influences on evacuation in the Intangibles Survey, these include: institutional factors, the performance of the formal warning system, social factors, and the social networks operating informal warnings. However, literature on warning response across different kinds of hazards reviewed by Fisher et al (1995) indicates that the source of a warning is important to the warning being taken seriously and acted on. However, the issue of the potential influence of the source of warning, whether official or unofficial, has not been explored in this analysis. Other factors may include: clarity and consistency of the message, frequency of the warnings, the type of authority giving the message, accuracy of past warnings, and frequency of disaster.
5.4.5 Social characteristics and material resources
The literature on evacuations in a range of hazard situations indicates that the presence of children in the home is a key factor (Heath et al., 2001; Fisher et al., 1995; Drabek, 2000; Van Duin and Bezuyen, 2000). The bi-variate analyses of the Intangibles Survey confirmed this. Households containing children under ten were more likely to evacuate than households without young children (75% with such children evacuated compared with 62% without).
The Intangibles data did not contain any information on the presence of pets in the home. Having pets has been found to act as a disincentive to evacuation because households are often not allowed to take pets to evacuation rest centres and will not abandon them (New Orleans anecdotal evidence). Heath et al. (2001) found that ownership of dogs or cats appeared to be the most important reason why households without children failed to evacuate and that people without children were prepared to put themselves in danger in order to stay with their pets. It was not possible to investigate this factor for this report.
Vulnerable households might be expected to have a higher incidence of household members leaving home. This was the case for those who lived alone (271). They had a significantly higher propensity to evacuate than those living with others (72% compared with 61% left home). This was also true of those aged 65 and over living alone (235) as compared with all other households. However, households that included an ill or disabled person and respondents who described their health prior to flooding as only fair or poor were no more likely to report household members leaving home than others without any such health problems in the home.
The kind of property lived in made a difference. Not surprisingly, people living in vulnerable property (bungalows, ground floor or basement flat and mobile homes i.e. property without an upstairs to retreat to, only 83 households) were significantly more likely to evacuate (89% compared to 63% of other in less vulnerable housing). Owner-occupiers were no more likely to evacuate than those with other forms of tenure.
One counter-intuitive finding emerged in the analysis. We would expect those with higher income and probably greater resources to be more likely to evacuate than those with less income since paying to stay in a hotel or to rent an alternative property to stay in while repairing their home would present them with fewer financial problems. However the reverse was the case (Table 5.9). Those in the lowest household income group are likely to be living alone, and to be elderly people living on pensions. These factors may explain the high evacuation rate in this income group.
When evacuation by social grade was considered, those in the highest social grade group (AB) like the highest income group were surprisingly the least likely to evacuate with only 54% doing so. It is possible that the higher income and social grade groups had large enough homes to make living on the premises while rehabilitation works were carried out more feasible and the resources to ensure that their homes were made habitable more speedily.
Table 5.9: Evacuation by gross monthly household income
Gross monthly income
£4,000 or more
Chi-square; p < 0.05
Having some form of insurance cover was a significant factor in evacuation. Significantly more of those with building and structure insurance (85% of homes were insured in this way) evacuated (67% evacuated compared with 56% of the uninsured). The same was true of ‘new for old’ contents insurance (80% had this cover). Two thirds (67%) with this cover evacuated compared with 59% without it. However, those with other forms of contents insurance were less likely to evacuate. Only just over half (53%) those with this cover evacuated, compared with 67% without it. Findings from the qualitative research show that having insurance which covers buildings or contents does not guarantee that the insurers will pay for alternative temporary accommodation (Tapsell et al., 1999; Tapsell and Tunstall, 2001). Varying levels of service are offered by different insurers. Failure or reluctance to pay for temporary accommodation was cited by some respondents. Although included in people’s policies, insurance companies often did not point this out unless people asked for it. Other common complaints include: slowness in dealing with claims and no up-front payments to cover immediate financial needs (Tapsell and Tunstall, 2001).
There were significant correlations between length of evacuation and many of the variables that affected whether or not residents left their homes. Length of evacuation was measured as a continuous variable and it correlated significantly with the following variables:
Maximum depth of flooding, correlation = 0.206, p< 0.001
Receiving a flood warning, correlation = 0.144, p < 0.001
Insurance: building structure, correlation = 0.146, p < 0.001
Insurance: contents, new for old, correlation = 0.072, p <0.05
Vulnerable property, correlation = 0.117, p <0.001
There was no significant correlation with children aged under 10 in the household.
5.5 Worry about future flooding The Intangibles Survey and the Lower Thames Surveys contained some form of questions on worry about future flooding. Worry is perhaps best considered as an affective response to a perceived risk. It may, along with flood risk perception, be a driver of behaviour after the flood where a flood event has occurred, or before an event when there has not been recent flooding. Not surprisingly, ‘flooded’ respondents in the Intangibles Survey were more worried about flooding, even over the relatively short time period of twelve months, than the ‘at risk’ (Figure 5.11).
Twenty seven percent (261) of flooded respondents said they were ‘very worried’ compared with only 9 % (49) of the ‘at risk’ sample. Half of the at risk respondents were either ‘not worried at all’ or ‘not very worried’ compared with 28% of flooded respondents despite a majority of them being aware of the flood risk in the area. Generally, however, the levels of worry were fairly low considering that all those surveyed lived in areas that had been affected by flooding in recent years.
Figure 5.11: Worry about future flooding: Intangibles Survey
Respondents in the Intangibles survey were also asked: ‘When thinking about your own home, which one (of various listed characteristics of a house flooding) worries you most?’ Those who had experienced flooding above ground floor and those at risk were similar in their concerns (Figure 5.12).
There were also significant differences in the degree of worry about the possibility of flooding in the next 12 months between flooded and non-flooded Lower Thames respondents, although the number who had some experience of flooding in this study was very small (Table 5.10). In the Intangibles Survey, there were significant differences between flooded men and women, with 31% of female respondents ‘very worried’ versus 20% of male respondents. A similar significant difference was found between men and women in the at risk sample (Table 5.11).
Figure 5.12: Concerns about flooding: Intangibles Survey
Table 5.10: Worry about flooding by those flooded and not flooded:
Lower Thames Survey
Worry about the possibility of being flooded in the next 12 months
Table 5.11: Worry about flooding by gender: Intangibles Survey
How worried are you about the possibility of your property being flooded during the next 12 months?
At risk **
Not worried at all
Not very worried
* Chi-square, p<0.001,** Chi-square, p<0.01
Many researchers have argued that gender has been ignored in the study of the impacts of disasters in general and flooding in particular (Enarson and Morrow, 1968; Fordham, 1998). Natural disasters such as floods have often been shown to have more adverse impacts on women than men (Morrow, 1999; Fordham, 1998; Ketteridge and Fordham, 1997, Enarson and Fordham, 2001; Tapsell and Tunstall, 2001; Tapsell et al., 2003). It is well recognised that men and women experience and respond to flooding differently. Fordham (1998) suggests that women are often invisible in disasters as they are often confined to the ‘feminine space’ and private domain of the home, and therefore may suffer more inconvenience when their routine in the home is disrupted. Women tend to be the chief homemakers and carers, looking after children and other dependants who may be upset by flooding, and often may have a greater emotional investment in the home; they also usually have to bear the greater part of the burden of getting the home back to normal after flooding. Women therefore may be more prepared to admit to worrying or they may feel more concerned about the prospect of flooding. For men, disasters may lead to changes in self-perception away from the traditional identity as provider and protector of their families to one of helplessness.
In contrast to the Intangibles Survey findings, there were no significant differences identified in worry by gender in the Lower Thames Survey. Furthermore, those whose property was immediately on the river bank (70 respondents, 34% of the sample) were no more worried than those at a greater distance from the river.
The degree of worry varied by area in the Intangibles Survey (RPA/FHRC, 2004). Local circumstances, in particular whether or not there is a flood alleviation scheme or any developments perceived to have potential to exacerbate the flood risk (such as run off from a new building) may affect the degree of worry experienced in a particular area (Tapsell et al., 1999, 2003).
There are also other factors such as the extent and depth of flooding experienced and other social and demographic characteristics, that may affect worry among the flooded and at risk that have not been examined in the analysis.
5.6 Summary People’s behaviour before, during and after a flood was analysed along with a number of driving factors such as individual or household characteristics. These factors help to highlight the levels of social vulnerability relating to capacity to cope with and recover from a flood event.
Registering on the AVM system can be one preparedness measure for flooding. Only 28% of Warnings Survey residents in Phase 2 had taken this precaution before the recent/worst flood (this question was not asked in the other surveys), while 24% signed up for the service after the event. Those with prior experience of more than one flood event and prior awareness of flood risk were more likely to take this preparedness measure. Taking out insurance was a common form of preparedness measure by residents in the flood affected areas, although for many flood insurance may have come automatically as part of their general household cover. Age was a significant factor relating to insurance take up in the Intangibles and Warnings Surveys, with older age groups less likely to have ‘new for old’ insurance and more likely to have other forms of contents insurance. The key factor important in insurance take-up was social grade, with those in the lowest social groups significantly less likely to have all kinds of insurance. Tenure was also important in affecting insurance take-up, with those not owning or buying their property less likely to have insurance of all kinds. Those people living in ‘vulnerable’ housing were generally also less likely to have insurance cover. Therefore, we can see very different drivers of human behaviour before a flood at work across forms of preparedness action such as registering onto the flood warning system and taking out insurance. Flood awareness and experience were important for the former and of no significance for the latter. Instead, taking out insurance appears to be related to socio-economic factors and institutional arrangements affecting tenure.
Few Warnings Survey residents had taken pre-flood measures to protect the structure of their property; the only preparatory action taken by more than half of the respondents was taking out insurance. Only minorities had adapted their behaviour and the way they lived in their homes in preparation for flooding before the recent or worst flood. The number of floods experienced was significantly correlated with the number of flood preparedness actions undertaken before the flood across all three surveys. Those who were aware of the risk of flooding in an area had undertaken a significantly higher mean number of preparedness actions prior to the flood than those unaware. Similarly, longer-term residents and owner occupiers had undertaken on average a higher number of preparedness actions before the last/worst flood. Vulnerable households had on average undertaken fewer preparedness actions before the flood. Readiness to take actions like signing up for the AVM appeared to be influenced by flood awareness and experience rather than by socio-economic factors and resources.
The most common post-flood measure taken by respondents in the Warnings Survey was obtaining sandbags or sand to prevent flood waters entering the property. Behavioural and other adaptations to the home to reduce the potential for flood damage were more common than structural changes post-flooding. In the Warnings Survey, even after the flood over a fifth of the respondents had taken none of the 16 actions to prepare for flooding. Respondents were somewhat more active and took more actions after the flood as compared with before but the number of flood preparedness actions taken after the flood event remained small. In the Lower Thames Study there were some significant differences in behaviour before or after flooding according to views on the likelihood of flooding, with those considering flooding likely mainly reporting themselves to be more watchful for the possibility of flooding but not much else.
Therefore, the only and most common action to protect property prior to flooding taken by more than a minority of residents was taking out flood insurance. Flood awareness and experience were clearly important in preparedness. Those who had taken preparedness actions tended to be those with the highest levels of flood risk awareness and previous experience of flooding, such as longer-term residents. Home owners/buyers had also taken more preparedness actions than renters. Following the last flood event, unsurprisingly flooded respondents were more likely to take actions than non-flooded to make their homes more flood-resilient, although levels of actions were fairly low overall.