letting them in, or stopping mail delivery when away for extended periods. Such precautions, of course, require little financial investment or time. More expensive and time-consuming precautions, however, are not rare. Although estimates vary, roughly 25-40 percent of American households have invested in such measures as window bars or grates, improved locks, property engraving, alarm systems, improved lighting, or theft insurance (see generally DuBow et al., 1979; Research and Forecasts, 1980; Skogan and Maxfield, 1981). The most reliable national data come from the 1983 Victim Risk Supplement to the National Crime Survey (U.S. Department of Justice, 1987), in which household informants were questioned about the steps they had taken "to make [their home] safer from crime." Burglary alarms were present in 7 percent of U.S. households; 34 percent of households had engraved or marked valuables; 42 percent had a gun or firearm "for protection"; and 58 percent had a dog.
Among household security measures, the most controversial is gun ownership. According to repeated GSS surveys, approximately one-half of U.S. households contain one or more firearms (National Opinion Research Center, 1988), but the purpose of those firearms has been the subject of vigorous debate. In a review of the literature on fear of crime and gun ownership, Wright et al. (1983:101) conclude that "there is no credible study anywhere in the literature that shows, clearly and unmistakably, a fear … effect in the weapons trend," and argue that firearms are primarily purchased for hunting and recreational purposes. Two subsequent studies, however, cast some doubt on this conclusion. Using survey data from 59 neighborhoods in three standard metropolitan statistical areas (SMSAs), Smith and Uchida (1988) found that the probability of purchasing "a gun or weapon for your protection" was significantly related to respondents' perceived risk of victimization, prior victimization experiences, and perceptions of neighborhood crime trends. Like McDowall and Loftin (1983), they found that the probability of purchasing a weapon increased when the police were perceived to be ineffective. (For related research, see Smith and Uchida, 1988).
Apart from their frequency, perhaps the most striking feature of public responses to fear is their age and sex distribution. As Skogan and Maxfield (1981:195) have observed, "Every analysis of crime-related behavior indicates that women and the elderly are more likely to avoid exposure to risk and to take numerous measures to reduce their chances of being victimized." These differences appear to be most pronounced with respect to avoidance