National multi-cohort time trends in adolescent risk preference and the relation with substance use and problem behavior from 1976 to 2011

Abstract

Aims: Preference for risky activities is an important developmentally graded predictor of substance use. Population-level trends in adolescent risk preference, as well as the way in which risk preference may be a conduit to risk behavior, have never been documented. The present study examines population-level trends in risk preference among U.S. high school seniors for the 36 years from 1976 to 2011, as well as trends in the association between risk preference and substance use and other problem behaviors.

Methods: Data were drawn from yearly nationally representative cross-sectional surveys of US high school seniors (N=91,860). Risk preference was measured consistently with two items. Marijuana and cocaine use, binge drinking, and conduct problems were assessed. Trends were tested using JoinPoint software.

Results: The mean level of reported risk preference among US 12th graders has increased over time, especially in the 1980s. For example, the proportion of high school females who reported enjoying activities that were "a little dangerous" more than doubled, from 4.9% in 1976 to 10.8% in 1988. While risk preference reports among adolescent males leveled off in 1992, risk preference reports among females show a continued positive overall slope through 2011. The magnitude of the association between risk preference and marijuana use has increased over time.

Conclusions: Reported preference for risky activities has increased among adolescents in the US, especially among young women. Reported risk preference is increasingly associated with a higher use of marijuana. Our findings argue for the importance of placing risk preference within a multi-level framework that attends to historical variation.

Keywords: Adolescence; Alcohol; Cocaine; Conduct disorder; Marijuana; Risk preference; Sensation seeking; Time trends.

Figure 1

Mean risk preference (scale of 2 to 10) among high school 12^th grade females (red) and males (blue) in the US from 1976 through 2011 *Best fitting model for mean risk preference across years included three slopes with two points of permutation. For males, points of permutation were 1978 (95% C.I. 1978–1984) and 1992 (95% C.I. 1988–1995). Slope 1: B=−0.02, Se=0.02, p=0.35; Slope 2: B=0.01, Se=0.0009, p<0.01; Slope 3: B=−0.0003, SE=0.0005 p=0.51. For females, points of permutation were 1983 (95% C.I. 1980–1985), and 1987 (95% C.I. 1985–1988). Slope 1: B=0.002, SE=0.003 p=0.57; Slope 2: B=0.04, SE=0.01, p=0.001; Slope 3: B=0.003, SE=0.0004, p<0.001.

Figure 2

The relation* between substance use, conduct problems** and risk preference from 1976 through 2011 among high school seniors in the US by sex * Based on linear regression with the continuous risk preference variable as the outcome, and controlled for all substance use simultaneously, conduct problems, age, sex, race/ethnicity, presence of father in the home, and grade point average. Parental education? ** Conduct problems based on a 7-question scale with total possible score ranging from 7 to 35. The scale was also dichotomized to those >9 versus 9 or less.