Prevalence and Descriptive Results
Weighted results indicated that 15.57% of young adults dropped out of high school (). Design-adjusted chi-square results showed significant differences in the distribution of dropout by racial/ethnic group (p < 0.001). Asians had the lowest rate of dropout (5.58%), followed by Whites (8.37%), African Americans (17.19%) and Afro-Caribbeans (19.02%), whereas Latinos had the highest rate of high school dropout (38.89%). The Asian and Latino groups were also broken down by larger sub-ethnicities and dropout rates were compared using survey design-adjusted chi-square tests, revealing significant differences between Asian sub-ethnicities (p < 0.047), with the highest dropout rate for Vietnamese (18.68%) compared to Filipinos (5.16%), Chinese (4.02%) and all other Asians (4.41%), and significant differences between Latino sub-ethnicities (p < 0.001) with the lowest rate for Cubans (13.46%) compared to Puerto Ricans (27.50%), Mexicans (45.99%) and all other Latinos (29.17%). When compared against each other there was no statistical difference between dropout rates of African Americans and Afro-Caribbeans.
Sample characteristics for U.S. youth ages 21–29 (n=2532).
There were no significant differences by gender with the male dropout rate at 16.18% and females at 14.98% (p=0.625). Dropout rates were significantly different between the nativity groups (p<0.001). Rates were lowest for U.S. born respondents (12.02%), but more than doubled for respondents who had immigrated to the U.S. at age 12 or younger (25.40%), more than four times as high for immigrants who arrived in the U.S. as adolescents (51.52%), and almost three times the rate for those who immigrated as adults (32.89%). We also looked at dropout rates for Latinos, Asians, and Afro-Caribbeans by immigration age categories. We found that Latinos had the highest dropout rate across all age immigration categories (for age 12 or younger, age 13 to 17, and 18 and older, the dropout rates were 36%, 63%, and 53% respectively). Afro-Caribbeans have fairly steady dropout rates at the younger (19%) and older (21%) immigration age categories but a lower rate for those who immigrated between age 13–17 (12%). We also observed that Asians had the lowest dropout rates compared to other immigrant groups. Asian dropout rates for immigration category at age 12 or younger was 9%; 5% dropout rate for immigration at ages 13 to 17; and 6% for those who immigrated as young adults. Respondents who dropped out of high school reported significantly lower parental education levels compared to non-dropouts: 53% reported that their mothers had also dropped out of school (p<0.001), and 62% reported that their fathers had also dropped out of school (p<0.001). Dropouts rates were marginally higher among girls from the NLAAS sample who had experienced early unwanted pregnancy (43.93%) compared to girls who did not report that experience (p=0.050). The dropout rate in youth who had entered the workforce at an early age (16.98%) was not significantly different from the dropout rate in youth who had not entered the workforce at an early age (p=0.478).
Psychiatric Characteristics of Dropouts
Thirty-eight percent of the CPES young adult subsample reported experiencing a major childhood trauma occurring at age 16 or younger (). Of those respondents who reported experience of any major childhood trauma, 19.79% dropped out of school, a significantly higher rate than those who did not report childhood trauma (12.97%, p < .001). Review of specific trauma experiences found significantly higher dropout rates were for those who had experienced child physical abuse (31.13%), witnessed domestic violence (26.01%), experienced rape (25.34%), were beaten (24.82%), or experienced a natural disaster (22.43%) compared to those who had not experienced those same specific traumas. Similarly, the dropout rate for respondents with childhood onset of substance disorder or conduct disorder was much higher (24.22% and 28.51% respectively) compared to respondents without those childhood disorders. Almost one-third of the sample (32.05%) reported symptoms which indicated childhood onset of one or more DSM-IV diagnoses, and the dropout rate for those respondents was significantly higher than for those without a childhood onset diagnosis (19.75% vs. 13.60%, p < .01). However a lower percentage (17.02%) reported use of any mental health services during that same age range (16 or younger).
Rates of trauma experience, psychiatric disorder and service use for U.S. youth ages 21–29 (n=2532).
Logistic Regression and Mediation Analysis
In , we present the results of our logistic regression models examining correlates of dropout status. In Model 1, Asians had significantly lower odds of dropping out (OR=0.41, 95% CI=[0.20,0.84]) compared to non-Latino White youth adjusting for demographic variables, early work experiences, and childhood trauma experience. In contrast, Afro-Caribbeans (OR=1.98[1.14,3.44]), African Americans (OR=2.18[1.40,3.40]), and Latinos (OR=2.88[1.88,4.44]) were more likely to report dropping out of high school as compared to non-Latino Whites. Those who immigrated as a young adult (age 18 and older) were marginally more likely (OR=1.54[1.01,2.34]) to have failed to complete a secondary education, in either their home country or the U.S., as compared to U.S. born respondents. With each year of maternal education completed, youth had decreased odds of dropping out of high school (OR=0.83[0.78, 0.87]).
Series of weighted logistic regressions predicting dropout for combined CPES sample (n=2532). Odds ratios and 95% confidence intervals presented
Next, we examined possible mediation of the relationship between childhood trauma and high school dropout status by psychiatric disorders and mental health services use by assessing the four criteria for establishing mediation (see ). We found that childhood trauma is significantly associated with dropping out ( Model 1; OR=1.65[1.18, 2.32]) and childhood trauma is also significantly associated with psychiatric diagnosis and services use variables (results not shown), thus satisfying the first two criteria. In Model 2 we added psychiatric diagnosis and services use variables to Model 1 and found that some of these variables were significantly associated with high school dropout status, thus fulfilling the third criteria. Particularly, the odds ratio of dropping out were 2.48[1.30, 4.74] for those with a childhood onset of substance disorder compared to those without substance disorder, and 2.38[1.43, 3.96] for those with childhood onset conduct disorder compared to those without conduct disorder. In contrast, mental health services use was not significantly associated with dropout after adjusting for childhood trauma and other covariates (OR=0.95[0.57,1.59]), and thus failed to satisfy the third criteria for mediation. Therefore, mental health services use was not considered a mediator of the relationship between childhood trauma and school dropout. After adjusting for the psychiatric diagnoses variables, the odds ratio of dropping out associated with childhood trauma were reduced to 1.36[0.98, 1.89], thus satisfying the last criteria. Since the odds ratio associated with childhood trauma was no longer statistically significant in Model 2, it suggests that the set of multiple mediators completely mediated the relationship between childhood trauma and dropout. The indirect effect of childhood trauma on dropout via the psychiatric diagnosis variables was estimated as a reduction in the odds ratio of dropout by 1.21[1.11,1.41] times.
Figures 1A and 1B Total effect of childhood trauma on dropout (A) and indirect effect of childhood trauma on dropout via multiple mediators: DSM-IV disorders and mental health services use (B). Odds ratios presented
Exploratory Moderated Mediation Analysis
As an additional exploratory analysis, we examined possible moderated mediation of the psychiatric diagnosis and services use variables moderated by race/ethnicity (MacKinnon, Fairchild, & Fritz, 2007
). For this analysis we used the same model as Model 2 of with additional variables that are interactions between each of the psychiatric diagnoses and services use variables and each of the major race/ethnicity groups, adjusting for the same sociodemographic variables as in Model 2 of . When we added the psychiatric diagnosis and services use variables and the interaction between these variables and race/ethnicity to Model 1, the odds ratio of dropping out associated with childhood trauma decreased from 1.65[1.18,2.32] to 1.35[0.97,1.88] (results not shown). Results suggest that the relationship between childhood trauma and dropping out of high school is mediated by psychiatric disorders and their interactions with race/ethnicity. More specifically, Afro-Caribbeans with childhood onset depressive disorders (OR=8.64[2.25, 33.11] and Latinos with childhood onset of anxiety disorders (OR=5.81[2.26,14.90]), and African Americans with childhood onset of conduct disorders (OR=2.70[1.00,7.27]) were more likely to report dropping out of high school compared to their non-Latino White counterparts. In contrast, Asians with childhood onset of depressive disorders, childhood onset of anxiety disorders, or those who have received mental health services as children were much less likely to report dropping out of high school compared to their non-Latino White counterparts in the same circumstance (p<0.0001).