Findings Problem Outcomes in Adolescence and Adulthood: Summary of Prevalence and Odds Table 1 shows the prevalence and odds of each problem outcome during adolescence and adulthood. Prevalence refers to the probability of experiencing a particular outcome. Odds compare the probability of experiencing an outcome with the probability of not experiencing it; mathematically, odds are equal to p/(1p) (prevalence divided by 1 minus the prevalence). For example, the prevalence and odds of violent victimization during adolescence are 0.683 and 2.155 (0.683/ [10.683]), respectively. This means that almost 7 out of 10 sample members were victims of a violent crime when they were adolescents and that the odds of being a victim of a violent crime during adolescence are a little more than 2 to 1. In contrast, the prevalence and odds of violent victimization during adulthood are much loweronly 0.353 and 0.546, respectively. Table 1 shows that, except for marijuana and polydrug use, the prevalence of the problem outcomes is lower during adulthood than during adolescence. Physical and Financial Consequences Table 2 presents information on the physical and financial consequences of the most recent victimization incidents reported by NYS respondents in 1989 and 1992, when they were 24 to 33 years old. (Data from the 1989 and 1992 waves were selected for analysis because these two waves provide the most extensive information about respondents' victimization experiences.) Table 2 includes information for specific offenses about the presence of firearms or other weapons, the extent of physical injury (for violent offenses), and the dollar value of lost or damaged property (for property offenses). Even though NYS data (like NCVS data) include both completed offenses and unsuccessful attempts, the offenses listed in table 2 generally are nontrivial events that would result in police intervention if detected by or reported to the police. Table 2 shows that in most instances of sexual assault (17, or about 90 percent of the 19 incidents reported), no weapon was used; however, a firearm was used in 1 incident and a weapon other than a firearm was used in another. Sexual assault victims suffered some injury in about two-thirds (12) of the 18 incidents with followup data on injury; the injury was serious in 2 incidents. For aggravated assault, a firearm was used in one of every five incidents, and other weapons were involved in nearly all the rest. Most assaults were unsuccessful, in the sense that no injury resulted; however, in about 1520 percent of the incidents, the victim was seriously injured. In addition, the distribution of injury outcomes for aggravated assault appears to be statistically "bimodal" (symmetrical, with two predominant peaks)the outcome tends to be either no injury at all or serious injury. For battery, the rate of serious injury is similar to that of aggravated assault, but the percentage of incidents with no injury is lower, and the percentage of incidents with some injury is more similar to the percentage with serious injury. In most incidents of battery (about 90 percent), no weapon was present. Robberies typically involved no weapon and no injury but a property loss of more than $100. Most incidents of motor vehicle or bicycle theft and vandalism involved property loss of more than $100 (one-third of motor vehicle thefts involved property loss of more than $1,000), whereas most pickpocket and purse-snatching offenses involved losses of less than $50, and most thefts from a vehicle or public place involved losses of more than $50. Comparisons With NCVS. Overall, the rates of injury reported by NYS respondents are somewhat higher than those reported by NCVS respondents, but the rates of serious injury for NYS respondents, particularly for victims of aggravated assault and battery, are comparable to the percentage of victims needing medical care, as reported for the NCVS by Klaus (1994). NYS estimates of financial loss per offense are consistent with NCVS estimates, with variations that one might expect (e.g., more than half of the NYS pickpocket or purse-snatching victimizations, but fewer than 10 percent of the NYS motor vehicle or bicycle theft victimizations, involved losses of less than $50). Relationship of Victimization to Other Problems in Adolescence and Adulthood Table 3 presents correlations between being a victim of violent crime and property crime and being involved in other problems, shown separately for adolescents and adults. Both prevalence (whether a respondent had a particular problem) and frequency (the number of times the respondent had the problem) are considered.13 The table shows that, for the most part, the prevalence and frequency of violent and property crime victimization are positively and significantly correlated with the prevalence and frequency of the other problems. The major exception is the prevalence of violent victimization in adulthood, which has no significant correlation with most of the other problem measures in adulthood. Most illegal behaviors are correlated more strongly with violent victimization than with property victimization. The exceptions are polydrug use and problem drug use in adulthood, both of which are more strongly correlated with property victimization than with violent victimization. Mental health problems in adolescence are clearly more closely correlated with violent victimization than with property victimization. For adults, the only statistically significant correlation with anxiety is that between frequency of violent victimization and ever-prevalence of anxiety. Depression in adults is more closely correlated with violent victimization than with property victimization (except for the nonsignificant correlation between prevalence of violent victimization and past-year prevalence of anxiety). PTSD has no statistically significant correlation with prevalence of violent victimization; however, its strongest correlations are with frequency of violent victimization. In general, in both adolescence and adulthood, the prevalence correlations between victimization and other problems are less than or equal to the frequency correlations. This indicates that within each of the two life stages, frequency (rather than prevalence) of victimization generally is a better predictor of both prevalence and frequency of other problems. Adolescent Risk Factors for Adult Problems Table 4 shows the relationship between being a victim of violent crime and property crime as an adolescent and experiencing various problem outcomes as an adulti.e., the "developmental comorbidity" of adolescent victimization and adult problems. The table analyzes the prevalence of the following adult problem outcomes: victimization by violent crime and property crime, domestic violence victimization and perpetration, violent offending (felony assault perpetration), property offending (felony theft perpetration), marijuana use, polydrug use, and problem drug use. Table 5 presents a similar analysis for adult mental health problems, including anxiety, depression, and PTSD. The analysis focuses on prevalence rather than frequency, in part because prevalence data are available for all of the problems but frequency data are not (for some problems, only Likert-type scales are available instead of frequency; see endnote 13). In addition, as noted earlier, the present research focuses on whether (rather than how often) adolescent victims of crime experience various problem outcomes in adulthood. Violent victimization as a risk factor. As shown in table 4, violent victimization during adolescence appears to be a risk factor forand, given the combination of association, time ordering, and control for spuriousness (as discussed above), a cause ofmost of the adult problem outcomes measured: violent crime victimization, domestic violence perpetration and victimization, violent and property crime perpetration, and problem drug use. Only two of the adult outcomes in table 4marijuana use and polydrug usedo not appear to be affected by violent victimization during adolescence. Risk factors for adult violent victimization. Victims of violence in adolescence also tend to be victims of violence in adulthood. The odds of adult violent victimization are more than twice as high for respondents who were victims of violence in adolescence than for those who were not. The odds of adult violent victimization are highest for the youngest age cohort and lowest for the oldest age cohort, probably reflecting the well-known pattern of decreasing victimization with age. The relationship between violent victimization in adulthood and victimization-age interaction in adolescence indicates that the pattern of higher adult victimization for adolescent victims is even more evident for respondents whose adolescent victimization occurred at an older age than for those who were victimized at a younger age. Again, this is consistent with what might be expected, given past research on the relationship between age and victimization. Risk factors for adult property victimization. Predictably, property victimization (rather than violent victimization) during adolescence affects property victimization in adulthood. This finding offers evidence of continuity in victimization from adolescence to adulthood. Less predictably, the other variables in the analysis do not appear to affect property victimization in adulthood. Risk factors for adult domestic violence victimization and perpetration. As shown in table 4, the odds of being a victim of domestic violence as an adult are increased by a factor of about 1.7 by being a victim of violent crime as an adolescent and by a factor of about 1.5 by being a perpetrator of such a crime as an adolescent. Similarly, the odds of being a perpetrator of domestic violence as an adult are increased by a factor of 1.7 by being a victim of violent crime as an adolescent and are doubled by being a perpetrator of violent crime as an adolescent. Being both a perpetrator and a victim of violent crime during adolescence more than doubles the odds of being a victim of domestic violence in adulthood and more than triples the odds of being a perpetrator of domestic violence in adulthood. Table 4 also shows that being male nearly doubles the odds of being a victim of domestic violence as an adult but decreases the odds of perpetrating domestic violence. This finding, although contrary to public perception, is consistent with findings of national surveys in the United States (Fagan and Browne, 1994; Morse, 1995) and New Zealand (Magdol et al., 1997). As Morse (1995) explains, females hit more often than males, but males do more damage when they hit. Morse's observation, together with the wide range of incidents included in the NYS definition of serious domestic violence (see sidebar), provides important context for interpreting the gender-related findings in this paragraph. Table 4 also shows that having parents of lower socioeconomic status is associated with a slightly higher probability of being a victim of domestic violence as an adult. This association, although weak, is consistent with other research findings about domestic violence. Risk factors for adult violent offending. As shown in table 4, being either a perpetrator or a victim of violent crime as an adolescent increases the odds of being a perpetrator of violent crime as an adult by a factor of about 3.5. Being both a perpetrator and a victim of violent crime as an adolescent increases the odds of perpetrating a violent crime as an adult by a factor of 13 (calculated by multiplying the odds ratio for the two risk factors). Being male doubles the odds of violent offending as an adult, and having parents of higher socioeconomic status slightly (but statistically significantly) decreases the odds of violent offending as an adult. Risk factors for adult property offending. Table 4 shows that the odds of property offending as an adult are more than doubled by being a property offender as an adolescent and nearly tripled by being a victim of violent crime as an adolescent. Being male triples the odds of adult property offending. Being nonwhite doubles the odds of adult property offending (but does not affect the odds of adult violent offending). Risk factors for adult drug use. As shown in table 4, marijuana use and property victimization during adolescence appear to be the only statistically significant risk factors for adult marijuana use. (Why property victimization should predict marijuana use is unclear; this relationship deserves further exploration.) Both marijuana use and polydrug use during adolescence are risk factors for both polydrug use and problem drug use in adulthood. Because polydrug use very rarely occurs without prior marijuana use (Elliott, Huizinga, and Menard, 1989), this relationship is to be expected. Although violent victimization during adolescence does not affect simple use of illicit drugs in adulthood, it nearly doubles the odds of problem use in adulthood. Being male increases the odds of problem drug use in adulthood by a factor of about 1.5. Socioeconomic status and race/ethnicity as risk factors for adult offending. Given the findings regarding socioeconomic status (which affects adult violent offending but not adult property offending) and race/ethnicity (which affects adult property offending but not adult violent offending), it may be useful to explore these relationships in more detail in future research. The findings do reveal, however, at least a slight tendency for adult offenders to be drawn disproportionately from respondents who have suffered socioeconomic disadvantage or who are members of racial/ethnic minority groups. This conclusion contrasts with other findings that show no direct relationship between adolescent offending and either socioeconomic status or race/ethnicity (Elliott, Huizinga, and Menard, 1989). The full impact of socioeconomic disadvantage or minority group membership may not be realized until the adult years. Adolescent risk factors for mental health problems. Table 5 shows that adult respondents who have been victims of violence during adolescence are twice as likely as others to report ever having had symptoms of PTSD. Violent victimization during adolescence is not, however, predictive of adult anxiety, depression, or past-year PTSD. This finding is consistent with the literature suggesting that PTSD is usually a short-term effect of violent victimization. Table 5 also shows that adolescent mental health problems are predictive of all the adult mental health problems studied, as would be expected. In addition, age cohort appears to be predictive of adult anxiety and PTSD, gender is predictive of adult depression and PTSD (both are less common in males), and socioeconomic status is weakly predictive of PTSD (higher socioeconomic status was associated with slightly lower odds of adult PTSD). Continuity of problems. Tables 4 and 5 show that, as expected, problem behaviors and experiences evidence continuity from adolescence to adulthood. For all of the adult problem outcomes analyzed in table 4, having a similar type of problem during adolescence is a risk factor. For example, violent offending during adolescence is related to both felony assault and domestic violence perpetration during adulthood, and both marijuana and polydrug use in adolescence are related to problem drug use in adulthood.14 As shown in table 5, adolescent mental health problems, as assessed by respondents and by their parents, triple the odds of adult anxiety, double the odds of adult PTSD, and also increase the odds of adult depression. A summary of risk. The "explained variation" (RL2) numbers in tables 4 and 5 all indicate only weak to moderate relationships between the independent variables (the predictors from adolescence) and the dependent variables (the adult problem outcomes). In other words, although the predictors matter, many individual respondents do not reflect the "average" pattern suggested by the logistic regression equations that are the basis of the tables. Even though considerable variation remains unexplained, individual predictors do have a substantial overall effect on the adult behaviors they predict. The figure below illustrates these effects by showing the percentage of adolescent victims of violence and nonvictims of violence who can expect to experience each of seven problem outcomes as adults, based on the results from tables 4 and 5. ![]()
As shown in the figure, no more than 25 percent of nonvictims can expect to have any given adult problem outcome. The highest risk outcomes for nonvictims are violent crime victimization and domestic violence victimization. The lowest risk outcomes for nonvictims (about 5-6 percent) are PTSD, serious violent offending, and property offending. For adolescent victims of violence, the risk is at least 10 percent for each adult problem outcome, more than 40 percent for violent victimization, and more than 25 percent for problem drug use and domestic violence victimization and perpetration. The figure indicates that, all other things being equal, adolescent victims of violence (compared with nonvictims) can be expected as adults to be 50 percent more likely to be victims of violent crimes and domestic violence, perpetrators of domestic violence, and problem drug users; twice as likely to experience PTSD; more than 2.5 times as likely to be serious property offenders; and 3 times as likely to be serious violent offenders. Adolescent Predictors of Adult Success A more general indication of how violent victimization and other experiences during adolescence affect adult outcomes can be found by considering how these experiences are related to an index of successful transition to adulthood used in the MacArthur Chicago-Denver Neighborhood Project (Elliott et al., 1996; Elliott et al., forthcoming). The index defines adult success in terms of four criteria: employment or financial stability, conventional aspirations and beliefs, involvement in a support network of friends or family, and abstinence from serious criminal behavior and problem substance use. (For details of the criteria, see sidebar.) To be classified as a success, a respondent must meet all four criteria.
Success as defined by these criteria was measured for the full NYS sample at wave 8 (ages 24-30) and wave 9 (ages 27-33). Approximately three-fourths of respondents had the same success classification in both waves; for the remaining one-fourth whose classification changed, the flows from failure to success and success to failure were roughly equal. The present analysis combines success measures for waves 8 and 9 into a single dependent variable (outcome measure) with three values: nonsuccess (not meeting the success criteria in either wave), unstable success (meeting the criteria in one wave but not the other), and stable success (meeting the criteria in both waves).15 The predictors in this analysis are the cumulative frequency (in adolescence) of violent victimization, property victimization, violent offending (felony assault), serious property offending (felony theft), alcohol use, marijuana use, and polydrug use; the cumulative prevalence of mental health problems as reported by the respondents; the prevalence of adolescent mental health problems as assessed by parents; sociodemographic characteristics, including gender, race/ethnicity, and socioeconomic status (based on parents' socioeconomic status during respondent's adolescence); age cohort (again, the analysis is limited to the three youngest cohorts); plus most recent grade-point average (wave 5 for most respondents) and the cumulative prevalence of adolescent employment.16 Cumulative frequency (as opposed to cumulative prevalence) was used where possible because it resulted in better prediction of adult success.17 Table 6 shows two comparisons: nonsuccess versus stable success and unstable success versus stable success. For each comparison, odds ratios indicate the extent to which each independent variable (such as cumulative frequency of adolescent victimization) predicts either nonsuccess (as opposed to stable success) or unstable success (as opposed to stable success). Victimization. As shown in table 6, violent victimization during adolescence predicts nonsuccess in adulthood, at a marginally significant (p < .10) level. Violent victimization is not quite a marginally significant predictor of success overall. Illegal behaviors. Of the adolescent illegal behaviors measured, frequency of felony assault offending and marijuana use are the best (and the only statistically significant) predictors of adult success. Both are significantly associated with higher odds of nonsuccess; marijuana use is also significantly associated with higher odds of unstable success. This finding may result in part from the continuity of illegal behavior from adolescence to adulthood, as described earlier. Mental health problems. Adolescent mental health problems are not statistically significant predictors of adult success. Sociodemographic characteristics. Being male is a marginally significant predictor of unstable success. Males are about 1.5 times more likely than females to experience unstable success. Being of minority racial/ethnic status and being from lower socioeconomic status both are marginally significant predictors of nonsuccess, a finding consistent with a long history of research on social stratification (e.g., Blau and Duncan, 1967; Wilson, 1987). None of the sociodemographic measures was a significant predictor of success overall. Grade-point average and employment. Having a lower grade-point average is a statistically significant predictor of nonsuccess. Adolescent employment is the only variable with a marginally significant effect on success overall but not on either nonsuccess or unstable success. (It nonsignificantly increases the odds of nonsuccess but decreases the odds of unstable success.) Summary. Nonsuccess as an adult is more likely for individuals who as adolescents were frequent victims of violence, perpetrators of violent offenses, and marijuana users. Nonsuccess is also more likely for minorities and persons with a lower socioeconomic background. Less stable patterns of adult success are found for males and marijuana users. Employment in adolescence appears to be more of a disadvantage than an advantage, although it is an advantage for some. In table 6, as in tables 4 and 5, the level of explained variation (RL2 = 0.127) is fairly modest. This indicates considerable variation from the "average" pattern with regard to the predictors of success.
|