(most notably age). The exposure/fear association disappeared when these confounding variables were held constant.
The mass media, of course, are not the sole sources of information on crime in the United States. Individuals may learn about crime by talking with friends, neighbors, family members, or coworkers. Crime is in fact a rather frequent topic of conversation among Americans. In surveys of three U.S. cities, Skogan and Maxfield (1981) found that 43 percent of respondents had talked with someone else about crime "in the past week or two." A survey of Seattle residents by the author (see note 1) found that 38 percent of respondents discussed crime at least once a week, and 74 percent at least once a month. Interviews with victims of street robberies suggest that crime victims frequently discuss their experiences with others (Lejeune and Alex, 1973).
Somewhat surprisingly, the propensity to talk about crime is quite general, meaning that it is not associated with such personal characteristics as age, sex, or race (Skogan and Maxfield, 1981). On the other hand, neighborhood characteristics strongly affect conversation about crime. Those who live in neighborhoods where crime is perceived to be a problem are more apt to talk about crime, and conversations among neighbors are more common in socially integrated neighborhoods (Gubrium, 1974; Skogan and Maxfield, 1981).
Everyday conversations about crime may cover a variety of topics, of course, but one of their functions is to transmit information about victimization experiences. As in the homicide example in the introduction, the multiplying effect of interpersonal networks in spreading news of crime can be stunning. Table 2 presents some additional evidence on the multiplying effect of interpersonal networks. The data come from a 1982 survey of 2,464 Americans conducted for ABC News (Inter-University Consortium for Political and Social Research, 1982). In the survey, respondents were asked to indicate whether they had been a victim of each of several offenses during the previous year, and whether these offenses had happened during the previous year to "anyone you personally know in your neighborhood." The first column in Table 2 shows the proportion of respondents who had been victimized by each offense, and the second column shows the proportion who knew neighborhood victims. As we would expect, respondents were much more likely to know a victim of each