Adolescent Substance Use and Other Illegal Behaviors and Racial Disparities in Criminal Justice System Involvement: Findings From a US National Survey

Abstract

We used data from a national survey to examine arrest rate disparities between African American and White adolescents (aged 12–17 years; n = 6725) in relation to drug-related and other illegal behaviors. African American adolescents were less likely than Whites to have engaged in drug use or drug selling, but were more likely to have been arrested. Racial disparities in adolescent arrest appear to result from differential treatment of minority youths and to have long-term negative effects on the lives of affected African American youths.

Racial disparities in the juvenile and criminal justice systems are important in both criminology and public health. African American youths have higher rates of arrest and detention than White youths.^1–4 The US governmental policies known collectively as the War on Drugs have contributed significantly to increases in rates of arrest and incarceration, especially of African Americans,^5–13 although rates of substance use and abuse among African Americans are either similar to^14,15 or even lower than those of Whites.¹⁶

Two main hypotheses address the overrepresentation of racial/ethnic minorities in the juvenile and criminal justice systems: (1) the “differential offending” hypothesis (that this overrepresentation generally reflects racial and ethnic differences in the incidence, seriousness, and persistence of engagement in delinquent and criminal behavior) and (2) the “differential treatment” hypothesis (that this overrepresentation is attributable to inequities—intended or unintended—in justice system practices as they affect particular populations).^2,17–20

We used data from a nationally representative survey of youths to examine the relationship of substance use and other illegal behaviors with arrest among African American and White youths to better understand racial disparities in arrest by testing whether the differential offending hypothesis or the differential treatment hypothesis best explains observed disparities, and by examining the longitudinal impact of arrest in adolescence on educational attainment.

METHODS

Data for the analyses came from the National Longitudinal Survey of Youth 1997 (NLSY97).²¹ Our baseline sample included 6725 youths (aged 12–17 years), of whom 4400 were White and 2325 were African American. We used a subsample of 5796 youths reinterviewed in 2003 in analyses examining high-school or general equivalency diploma (GED) completion by that year. This subsample was not found to differ significantly, other than on age, from those lost to follow-up.

We assessed arrest history at baseline (in 1997), by using a 3-category variable: no arrests, a single arrest, or multiple arrests. Respondents were asked if they had ever used alcohol or any illicit drug, and if they had ever sold drugs. A scale variable based on 6 other behavioral items represented non–substance-related illegal behaviors. We obtained information on high-school completion in follow-up interviews. Other variables included child age, gender, race/ethnicity, family income, urbanicity, and 2 county-level variables: high-crime area and high county unemployment rate.²¹

We compared the 2 racial groups on major sociodemographic factors and on baseline rates of arrest, substance use, and non–drug-related illegal behaviors, separately for the group aged 12 to 14 years and the group aged 15 to 17 years. We applied multinomial logistic modeling hierarchically to the 3-category arrest history outcome variable for each age group.

To examine the long-term impact of adolescent arrest on later educational attainment, we divided the 5796 White and African American respondents who remained in the study in 2003 into 4 groups, according to race and baseline arrest history, and compared the high-school (high-school diploma or GED) completion rates of the 4 groups. We used logistic regression models for high-school completion to examine the group differences. We weighted all analyses with the NLSY sample weights.

RESULTS

At the time of the baseline interview, among the 6725 youths from the 2 racial groups, 543 had been arrested at least once. African American youths were more likely than Whites to have been arrested multiple times, with rates of 3.1% versus 1.3% within the younger age group, and 6.5% versus 4.1% in the older age group (Table 1). White youths, however, had higher rates of substance use than did African American youths. Reported drug-selling activity was also more common among White than among African American youths; this racial difference was statistically significant for the older age group.

TABLE 1—

Adolescent Race by Sociodemographics, Arrest History, Substance Use, and Other Illegal Behaviors, and by Age Group: National Longitudinal Survey of Youth 1997

	Ages 12–14 Years (n = 3566)			Ages 15–17 Years (n = 3159)
Characteristic	African American (n = 1208), % or Mean (SD)	White (n = 2358), % or Mean (SD)	Pa	African American (n = 1117), % or Mean (SD)	White (n = 2042), % or Mean (SD)	Pa
Gender
Male	50.5	51.3	.709	51.1	50.7	.87
Female	49.5	48.7		48.9	49.3
Income, $
<20 000	30.5	11.1	<.001	26.1	9.5	<.001
20 001–40 000	21.5	18.5		21.3	18.4
40 001–80 000	17.7	35.3		13.9	33.1
>80 000	3.7	15.6		4.9	15.3
Income data missing	28.6	19.5		33.8	4.4
Urbanicity
Urban resident	78.3	62.7	<.001	76.8	62.2	<.001
Rural resident	17.9	32.3		18.2	33.4
Urbanicity data missing	3.8	5.0		5.0	4.4
High-crime areab
Yes	67.9	42.3	<.001	68.4	40.9	<.001
No	32.1	57.7		31.6	59.1
High county unemployment rateb
Yes	56.7	49.8	<.001	57.9	48.7	<.001
No	43.3	50.2		42.1	51.3
Arrest history
No arrests	92.9	96.0	<.001	87.2	89.2	.049
Single arrest	4.1	2.7		6.3	6.6
Multiple arrests	3.1	1.3		6.5	4.1
Substance use—lifetime
Any illicit drug(s)	10.1	12.7	.066	27.5	35.6	<.001
Alcohol	23.2	32.7	<.001	48.4	66.1	<.001
Drug selling and other illegal behaviors—lifetime
Sold drugs	2.8	4.0	.13	6.6	11.9	<.001
Non–drug-related illegal behaviors scorec	0.78 (0.82)	0.65 (1.22)	.008	0.94 (0.98)	0.88 (1.39)	.31

^aFor categorical variables, P value shown is from χ² test; for continuous variables, P value is from t test.

^bWe assessed county crime rate and county unemployment rate for each study participant’s place of residence by using county-level geocoded crime rates and unemployment rates.²¹ We used these 2 continuous variables to create a dichotomous “high-crime area” variable, and a “high county unemployment rate” variable, in each case using the 50th percentile as the cut-off point.

^cThe respondents were asked whether they had ever (1) carried a hand gun, (2) belonged to a gang, (3) purposely damaged property, (4) stolen something worth more than $50, (5) committed other property crimes, or (6) attacked someone. The 6 items were summed to produce youths’ scores on the “other illegal behaviors” scale.

The results of the multinomial logistic regression analyses showed that, within the younger age group, when we controlled for age and gender, African American adolescents were 2.5 times as likely as Whites to have been arrested multiple times, and 1.6 times as likely to have been arrested once (Table 2, model 1). In model 2, with substance use, drug selling, and other illegal behaviors also taken into account, race remained a significant predictor of both single and multiple arrest(s). For older youths, our results similarly indicated racial disparities in arrest that were not explainable on the basis of behavioral differentials (data not shown). Table 2, model 3, shows that household income level, urbanicity, and county unemployment rate do partially, but do not fully, explain racial disparities in arrest among younger adolescents.

TABLE 2—

Multinomial Logistic Regression Analyses Predicting Arrest History Among Adolescents Aged 12 to 14 Years From the National Longitudinal Survey of Youth 1997

	Model 1, AOR (95% CI)		Model 2, AOR (95% CI)		Model 3, AOR (95% CI)
	Single Arrest vs None	Multiple Arrests vs None	Single Arrest vs None	Multiple Arrests vs None	Single Arrest vs None	Multiple Arrests vs None
Race
African American	1.61* (1.03, 2.53)	2.53*** (1.46, 4.38)	2.14*** (1.32, 3.48)	3.20*** (1.72, 5.96)	2.18** (1.31, 3.63)	2.20* (1.23, 4.29)
White (Ref)	1.00	1.00	1.00	1.00	1.00	1.00
Agea	1.68*** (1.28, 2.20)	1.73** (1.20, 2.49)	1.23 (0.95, 1.70)	1.23 (0.85, 1.85)	1.28 (0.95, 1.71)	1.31 (0.88, 1.96)
Gender
Male	2.52*** (1.64, 3.88)	4.43*** (2.29, 8.57)	1.63* (1.02, 2.60)	2.27* (1.11, 4.61)	1.65* (1.03, 2.64)	2.37* (1.15, 4.89)
Female (Ref)	1.00	1.00	1.00	1.00	1.00	1.00
Any drug use
Yes			1.17 (0.67, 2.03)	3.62*** (1.78, 7.36)	1.12 (0.62, 1.94)	3.15** (1.53, 6.49)
No (Ref)			1.00	1.00	1.00	1.00
Any alcohol use
Yes			2.74*** (1.65, 4.55)	1.12 (0.55, 2.26)	2.71*** (1.63, 4.51)	1.11 (0.53, 2.28)
No (Ref)			1.00	1.00	1.00	1.00
Ever sold drugs
Yes			2.12* (1.11, 4.03)	1.27 (0.58, 2.82)	2.31* (1.20, 4.46)	1.80 (0.79, 4.13)
No (Ref)			1.00	1.00	1.00	1.00
Non–drug-related illegal behaviors scorea			1.72*** (1.47, 2.01)	2.01*** (1.65, 2.46)	1.75*** (1.49, 2.05)	2.08*** (1.68, 2.68)
Low household income (<$20 000)b
Yes					1.25 (0.71, 2.19)	2.72** (1.43, 5.19)
No (Ref)					1.00	1.00
Urbanicityb
Urban resident					1.54 (0.93, 2.58)	2.51* (1.10, 5.72)
Rural resident (Ref)					1.00	1.00
High-crime area
Yes					0.70 (0.45, 1.10)	1.34 (0.70, 2.55)
No (Ref)					1.00	1.00
County unemployment rate
High					1.01 (0.66, 1.53)	2.46** (1.29, 4.67)
Low (Ref)					1.00	1.00

Note. AOR = adjusted odds ratio; CI = confidence interval.

^aContinuous variable.

^bCases with missing income or urbanicity data were also included in the analyses; the AORs for these “missing” categories, however, are not shown.

*P < .05; **P < .01; ***P < .001.

We divided the 5796 participants who remained in the sample in 2003, 6 years after the baseline interview, into 4 subgroups according to baseline arrest history and race. African Americans who had had arrest histories at baseline were more likely than their White counterparts to have failed to complete high school as of 2003 (adjusted odds ratio = 2.43; P < .001).

DISCUSSION

We examined adolescent substance use and other illegal behaviors in relation to racial disparities both in criminal justice system involvement and in educational attainment subsequent to an arrest, and tested the differential offending and differential treatment hypotheses, using data from a nationally representative community sample of youths. Consistent with previous research,^22,23 we found that, overall, African American youths were more likely to have been arrested, and to have been arrested multiple times, than were White youths. This disparity could not be explained by differences in the 2 groups’ rates of drug-related and other illegal behaviors. The fact that the magnitude of the race effect diminished after we adjusted for some socioeconomic factors suggests that racial disparities overlap with social class disparities. We also found that racial disparities in early juvenile or criminal justice system involvement were negatively associated with later educational attainment, indicating some long-term effects that these disparities may have on African American youths.

The study was limited by the lack of data on policing behaviors or attitudes and by its reliance on adolescents’ self-reports for information on their illegal behaviors. No causal relationship can be established on the basis of these findings. However, the findings do provide policymakers, both in public health and in the justice system, with important information regarding racial disparities in arrests and their possible consequences, calling for further action aimed at reducing these disparities.