In 1991, legislation was introduced into the Kansas state legislature that would have required underground shelters for mobile home parks. During the course of the committee debate on that legislation, I was approached by a state senator for information to counter some claims made by representatives of the Manufactured Housing Association that mobile homes were no more dangerous than other types of housing in tornadoes and, in particular, that they were no more dangerous than apartment buildings without underground shelters (see the media brief by the Manufactured Housing Institute, which contains the demonstrably false statement "A direct hit from a tornado will bring about severe damage or destruction of any home in its path."). I went back through the records in Storm Data to find where fatalities occurred in order to check those claims. Fatalities in mobile homes have been singled out in the record for a number of years, but discrimination between kinds of "permanent" dwellings is not a part of the record. I gathered what information I could and passed it along to the senator, but the legislation failed anyway.
In early 1994, representatives of the Manufactured Housing Association of Oklahoma visited the National Weather Service Forecast Office (NWSFO) in Norman in an effort to convince the office to change its statements about the safety of mobile homes in tornadoes. They claimed that the NWS is "unfairly" singling out mobile homes in warnings. Their position was that new manufacturing procedures and tie-down regulations have resulted in improved safety. While it is possible that the improvements in construction and installation may lead to a reduction in fatalities in mobile homes, such a reduction has not occurred yet. Since it appears that this visit is part of a larger, national campaign to change warning statements and safety advisories about tornadoes, we wanted to share some information about the current state of mobile home safety in tornadoes.
Since 1975, fatalities in mobile homes have accounted for one-third of all tornado deaths in the United States (Table 1). Since 1985, the fraction has been slightly higher, with just over one
fourth in permanent homes, and 12.6% in vehicles, the third highest value of any identifiable location (Table 2). This final number may be the beginning of a respsonse to the question of Schmidlin and King ("Cars and tornadoes: Where is the research?", Bull. Amer. Meteor. Soc., 77, 963-4), who have raised the possibility that people may be safer in vehicles than outdoors. The problems in using these data to come to a conclusion on the question is the difficulty in identifying the underlying population at risk and the question of the 34 deaths that do not have a location reported in the Storm Data reports (6.6%)Nevertheless, the observation that three times more people have been killed in vehicles than outside is a smoking gun that supports the notion that vehicles are dangerous places in tornadoes.
To normalize the risk by population in mobile homes and other kinds of residences, we've looked at residential fatalities since 1985. From 1985-1995 there have been 191 mobile home deaths and 130 in other kinds of residences (Table 3). Since (according to the US Census Bureau) only 6.1% of the United States population lives in mobile homes, this represents a much greater risk than for residents of "permanent" housing. During that time period, the average number of annual deaths per 10 million mobile home residents was approximately 11.4, while it was only 0.5 in other housing, implying that mobile home residents die at a rate 22.6 times greater than non-mobile home residents. Even assuming that tornadoes primarily occur in areas where the percentage of the population living in mobile homes is greater (say, double it to 12.2%, the approximate value for Butler County, Kansas, where Andover is located), the death rate is 10.6 times greater in mobile homes. As far as the statements about risks in apartments compared to mobile homes, assuming that all of the "unknown" permanent housing deaths occurred in multiple family dwellings, and using the percentage of US residents living in multiple family dwellings (28.5%, according to the US Census Bureau), the annual death rate per 10 million multiple family dwelling residents was 0.7.
It is possible that, in the future, as a result of modern building techniques, mobile homes may become safer in severe convective weather. However, that day has not arrived yet and it would be imprudent for the NWS or its local offices to deemphasize the dangers of mobile homes in tornado and severe thunderstorm warnings and in its preparedness work.
Table 1: Annual mobile home tornado fatalities (MH), non-mobile home fatalities (Non-MH), total tornado fatalities and percentage of fatalities in mobile homes for United States from 1975-1993.
| Year | MH | Non-MH | Total | %Mobile |
| 1975 | 13 | 47 | 60 | 21.7 |
| 1976 | 18 | 26 | 44 | 40.9 |
| 1977 | 7 | 36 | 43 | 16.3 |
| 1978 | 15 | 38 | 53 | 28.3 |
| 1979 | 9 | 75 | 84 | 10.7 |
| 1980 | 11 | 17 | 28 | 39.3 |
| 1981 | 14 | 10 | 24 | 58.3 |
| 1982 | 20 | 44 | 64 | 31.3 |
| 1983 | 20 | 14 | 34 | 58.8 |
| 1984 | 43 | 79 | 122 | 35.2 |
| 1985 | 28 | 66 | 94 | 29.8 |
| 1986 | 9 | 6 | 15 | 60.0 |
| 1987 | 25 | 34 | 59 | 42.4 |
| 1988 | 15 | 17 | 32 | 46.9 |
| 1989 | 8 | 42 | 50 | 16.0 |
| 1990 | 9 | 44 | 53 | 17.0 |
| 1991 | 23 | 16 | 39 | 59.0 |
| 1992 | 21 | 18 | 39 | 53.8 |
| 1993 | 13 | 20 | 33 | 39.4 |
| 1994 | 24 | 45 | 69 | 34.8 |
| 1995 | 11 | 19 | 30 | 36.7 |
| Total | 356 | 713 | 1069 | 33.3 |
Table 2: List of locations of tornado fatalities (from Storm Data and National Disaster Survey Reports) from 1985-1995. Bottom row is percentage of total deaths for period.
| Outside | Vehicle | Permanent Home | Mobile Home | Other/ Unknown | Total | |
| 1985 | 2 | 7 | 40 | 28 | 13 | 90 |
| 1986 | 0 | 3 | 3 | 7 | 2 | 15 |
| 1987 | 3 | 3 | 7 | 24 | 22 | 59 |
| 1988 | 0 | 3 | 6 | 21 | 2 | 32 |
| 1989 | 0 | 16 | 8 | 12 | 14 | 50 |
| 1990 | 1 | 14 | 25 | 7 | 6 | 53 |
| 1991 | 9 | 4 | 3 | 23 | 0 | 39 |
| 1992 | 0 | 0 | 12 | 21 | 6 | 39 |
| 1993 | 3 | 7 | 6 | 13 | 4 | 33 |
| 1994 | 1 | 3 | 12 | 24 | 29 | 69 |
| 1995 | 2 | 4 | 8 | 11 | 5 | 30 |
| Total | 21 | 64 | 130 | 191 | 103 | 509 |
| % | 4.1 | 12.6 | 25.5 | 37.5 | 20.2 | 100 |
Table 3: Locations of tornado fatalities in residences from 1985-1995. "Single", "Multiple", and "Unknown" all refer to fatalities in permanent (i.e., non-mobile) dwellings. "Total permanent" is sum of all permanent residence fatalities.
| Single | Multiple | Unknown | Total Permanent | Mobile | Total Residence | |
| 1985 | 7 | 2 | 31 | 40 | 28 | 68 |
| 1986 | 2 | 1 | 0 | 3 | 7 | 10 |
| 1987 | 1 | 0 | 6 | 7 | 24 | 31 |
| 1988 | 4 | 0 | 2 | 6 | 21 | 27 |
| 1989 | 4 | 4 | 0 | 8 | 12 | 20 |
| 1990 | 17 | 8 | 0 | 25 | 7 | 32 |
| 1991 | 2 | 1 | 0 | 3 | 23 | 26 |
| 1992 | 12 | 0 | 0 | 12 | 21 | 33 |
| 1993 | 6 | 0 | 0 | 6 | 13 | 19 |
| 1994 | 12 | 0 | 0 | 12 | 24 | 36 |
| 1995 | 8 | 0 | 0 | 8 | 11 | 19 |
| Total | 75 | 16 | 39 | 130 | 191 | 321 |
| % | 23.4 | 5.0 | 12.1 | 40.5 | 59.5 | 100 |