Adolescence, particularly mid- to late-adolescence (15–20 years old), is considered to be the period of the lifespan with the highest risk-taking and sensation-seeking behaviors [1,2]. While likely adaptive (e.g., promoting independence from caregivers) [1], this lifespan peak in risk-taking confers vulnerability to accidental fatalities and substance use disorders [1,2]. Prior large-scale epidemiological research has examined population patterns and social determinants of adolescent substance use [3]. However, most prior research on dispositional measures of adolescent risk-taking propensity and sensation-seeking, precursors to substance use and thought to be more generally relevant to all adolescents [2,4], has relied on relatively small university-based convenience samples lacking participant diversity in social, environmental, and psychological experiences. Thus, the generalizability and inclusivity of current theories of adolescent risk-taking and its associated vulnerabilities remain unknown. Clarity on the real-world generalizability of this theory is essential to current efforts to address youth mental health [5], through for example, identifying broadly applicable developmental periods for intervention and prevention efforts and the increasing number of targeted studies of adolescent risk-taking and substance use (e.g., Adolescent Brain Cognitive Development Study, ABCD [6]). Toward current field-wide goals of 1) explicitly testing the generalizability of mechanisms of health and disease across diverse populations [7] and 2) establishing inclusive theories of mental health risk in youth [8], the current project used national survey data to test the generalizability of adolescent peaks in self-reported risk-taking propensity and its link to substance use across a comprehensive set of participant-level social, environmental, and psychological factors (19 total subgroups defined via levels of biological sex at birth, race/ethnicity, socioeconomic status, population density, religious affiliation, and mental health).

Methods

Public data (N = 1,005,421) were drawn from the 2002–2019 National Survey on Drug Use and Health (NSDUH), an annual survey representative of the noninstitutionalized civilian population (12-years-old and older) of the United States. NSDUH procedures were approved by the RTI International Institutional Review Board, and participants provided informed oral consent.

Linear and nonlinear (spline) models were fit to a self-reported risk-taking propensity measure, consistent with prior work with NSDUH [9] and also supporting inferences from related work in adults [10] (sum score of two items: “[I] get a kick out of doing things that are dangerous”, [I] like to test [myself] by doing risky things”; association among scores: φc = 0.587; see Supplementary Material). Leveraging the very large size and participant diversity of the NSDUH data and with the goal of informing future smaller targeted studies of adolescence with expanded risk-taking and sensation-seeking metrics, variation in this risk-taking propensity measure was independently examined across a comprehensive set of participant-level factors (biological sex at birth, race/ethnicity, socioeconomic status, population density, religious affiliation, and mental health; 19 total population subgroups). Through this framework, primary study aims were to characterize the generalizability across social, environmental, and psychological factors of (1) lifespan peaks in self-reported risk-taking propensity during adolescence (ages at peak risk-taking determined via nonlinear lifespan trajectories) and (2) the association between adolescent risk-taking and regular (at least weekly; see Supplementary Material) alcohol and cannabis use. To ensure reproducibility and robustness, models were run on individual NSDUH study years, and inferences were drawn on results that reproduced across independently assessed NSDUH years (2002–2019). NSDUH guidelines for inclusion by study year are reflected in population density and depression-defined subgroups only using a subset of NSDUH years. All other analyses included all 18 years of NSDUH data. See Table S1 for a complete list of analysis variables, corresponding NSDUH variable names, and frequencies. All statistical models were run in R version 4.2 with the survey package version 4.1, which used analysis weights, replicates, and stratum distributed with public access data from the 2002–2019 NSDUH to account for the complex survey design.

Results

Replicating basic science studies of adolescent development [1], self-reported risk-taking propensity followed a nonlinear trajectory across the lifespan, with a robust peak during mid-adolescence (∼16-years-old) for the full sample (Figure 1A). Importantly, however, expanding this foundational work to consider a wide range of participant-level social, environmental, and psychological factors, independent models assessing the 19 population subgroups demonstrate a high degree of generalizability of this mid-adolescent peak, with 18 of 19 subgroups having peaks (median across NSDUH years) in risk-taking propensity occurring between 15 and 18 years old (Figure 1B; 87.3% of all NSDUH year by subgroup combinations). In contrast, the magnitude of the peak (average value from 15–18 years old) in self-reported risk-taking propensity widely and significantly varied (p's < .05 corrected [for 6 comparison categories], uncorrected p < .008) by population subgroup (Figure 1C). For example, adolescent males reported higher risk-taking propensity than females, and white adolescent participants reported higher risk-taking propensity than other racial/ethnic groups (Figure 1C). However, across all population subgroups, higher levels of self-reported risk-taking propensity between 15–18 years old significantly (p's < .05 corrected, uncorrected p < .003) disambiguated adolescents who regularly (weekly; see Supplementary Material) use cannabis or alcohol from non-using adolescents, with a high degree of consistency and generalizability in the magnitude of these differences (Figure 1D). A broader, graded pattern of self-reported risk-taking propensity differences was observed when comparing both adolescents specifically, as well as the full sample of participants, among those with regular substance use, nonregular use, and nonuse (see Supplementary Material).

Figure 1 Self-reported adolescent risk-taking propensity across population subgroups in the National Survey on Drug Use and Health. (A) Nonlinear lifespan models averaged across 2002–2019 NSDUH of self-reported risk-taking for the full sample (all participants; dark line) and for 19 population subgroups, analyzed individually (light lines). Vertical tick marks on x-axis indicate coded ages available in public NSDUH dataset (Supplementary Material). (B) Age of peak risk-taking (maximum value from nonlinear fit) from individual subgroup models from (A). Each dot indicates one NSDUH study year (2002–2019). Eighteen out of 19 population subgroups have an average peak (median across NSDUH years) between 15–18 years old, 87.3% of all year by subgroup combinations. Dashed line indicates range of year-to-year variability from the full sample. (C) Magnitude of adolescent (15–18 years old) self-reported risk-taking by subgroup. Dashed line indicates range of year-to-year variability. Apart from population density, omnibus differences among all population subgroups within a given category (sex, race/ethnicity, income, religious affiliation, depression) were statistically significant (p < .05, corrected [for six comparison categories]; uncorrected p < .008) across all NSDUH study years. (D) Standardized differences (Cohen's [d]) on self-reported risk-taking between adolescents (15–18 years old) who report weekly (or more) alcohol (dark blue) or cannabis (light blue) use compared to non-using adolescents for each population subgroup, adjusted for all remaining variables (see Supplementary Material). Circles display NSDUH year by subgroup combinations with corrected significance for users greater than nonusers (91.1% of all combinations for cannabis, 94.6% for alcohol; p < .05 corrected [for 19 population subgroups], uncorrected p < .003). Triangles display effects that do not exceed this significance threshold. Dashed line indicates 0 (no effect).

Discussion

We demonstrate that a lifespan peak in self-reported risk-taking propensity during adolescence and its link to substance use is generalizable and independently observed across a comprehensive set of social, environmental, and psychological participant-level characteristics, including subgroups based on biological sex at birth, race/ethnicity, socioeconomic status, population density, religious affiliation, and mental health. This work thus supports the potential broad applicability and real-world utility of intervention and prevention efforts that build from central theories emphasizing risk-taking as a general vulnerability factor (e.g., for substance use) for adolescents [1–3,11]. In parallel with models that highlight ongoing development of cognitive processes [12], the generalizability of observed lifespan peaks in risk-taking propensity during adolescence may help clarify broadly applicable but developmentally specific behavioral features through which to tailor intervention and prevention efforts. Importantly, however, unlike the developmental timing, the magnitude of adolescent risk-taking notably varies across individuals according to relevant social, environmental, and psychological factors, with patterns that mirror results seen in adults [10,13]. Such response patterns and engagement in self-reported risk-taking for an individual adolescent that vary according to contextual factors highlights the value of personalized approaches to assessment and intervention [14]. Future work expanding these efforts towards personalized approaches should also consider intersectional perspectives (multiple, overlapping identities within an individual that may be related to historical patterns of inequity, marginalization, and racism) that are particularly relevant for culturally and equity-informed assessment, intervention, and policy related to adolescent health [15].

Limitations of this study include the use of brief self-report data. Such brief risk-taking items embedded within very large, population-representative surveys allowed the current project to test the generalizability of adolescent risk-taking propensity across a comprehensive set of participant-level social, environmental, and psychological factors that have not been possible in prior smaller targeted studies of adolescent risk-taking. The insights from the current work are supported by recapitulating prior observations of adolescent peaks in risk-taking propensity, an association with substance use, and population patterns from adults. Nevertheless, the insights from the current work should be expanded with comprehensive self-report and objective measures in future targeted studies. We suggest that complementary approaches in targeted studies of adolescence and the large-scale, representative national survey data presented here provide a well-suited means towards addressing growing concerns regarding reproducibility and generalizability in developmental sciences [16,17], as well as to foster additional studies on risk-taking measurement. For example, in line with previous work, we note that dispositional measures of risk-taking propensity and sensation-seeking appear only moderately associated with behavioral manifestations of risk-taking [18]. Expanding the current results with multiple levels of analysis (e.g., neuroimaging, simultaneous behavioral and clinical assessments [19]) and timescales (e.g., ecological momentary assessment [20]) will likewise support improved rigor and reproducibility.

In summary, the current project leveraged large-scale (N > 1 million) national survey data to test the generalizability of a foundational theory of the adolescent period. Results demonstrated that across 19 independent subgroups spanning levels of key social, environmental, and psychological factors, mid- to late-adolescence is consistently the period of the lifespan with the highest self-reported risk-taking behaviors, and these may confer risk to substance use. Future work with complementary, targeted cohort studies of adolescence and large-scale, representative data can further refine these results and accelerate generalizable and inclusive translation of developmentally informed intervention and prevention efforts.

Summary

This study aimed to investigate the generalizability of adolescent peaks in self-reported risk-taking propensity and its link to substance use across diverse populations. Using data from the National Survey on Drug Use and Health (NSDUH) from 2002 to 2019, the researchers examined the relationship between age and risk-taking propensity, as well as the association between risk-taking and substance use, across 19 subgroups defined by factors such as sex, race/ethnicity, socioeconomic status, population density, religious affiliation, and mental health.

Methods

The study utilized a large-scale, population-representative dataset from the NSDUH, which included over 1 million participants aged 12 years and older. Participants self-reported their risk-taking propensity using two items: "[I] get a kick out of doing things that are dangerous" and "[I] like to test [myself] by doing risky things." The researchers fitted linear and nonlinear models to the risk-taking propensity data, examining variation across the 19 subgroups. They also investigated the association between risk-taking propensity and regular alcohol and cannabis use.

Results

The study found that across all 19 subgroups, there was a consistent peak in self-reported risk-taking propensity during mid-adolescence (around 16 years old). However, the magnitude of this peak varied significantly across subgroups. For example, adolescent males reported higher risk-taking propensity than females, and white adolescent participants reported higher risk-taking propensity than other racial/ethnic groups. Regardless of subgroup, higher levels of risk-taking propensity between 15 and 18 years old were significantly associated with regular cannabis and alcohol use.

Discussion

The findings of this study suggest that the adolescent peak in risk-taking propensity is generalizable across a wide range of social, environmental, and psychological factors. This information supports the potential for broad applicability of interventions and prevention efforts targeting adolescents, particularly those focusing on risk-taking as a vulnerability factor for substance use. However, the magnitude of risk-taking propensity varies considerably among individuals, highlighting the need for personalized approaches to assessment and intervention. Future research should explore the intersectional aspects of risk-taking, incorporate more comprehensive and objective measures, and utilize multiple levels of analysis, including neuroimaging and ecological momentary assessment, to enhance rigor and reproducibility.

Summary

This study examined the generalizability of adolescent risk-taking propensity and its association with substance use across various social, environmental, and psychological factors. Utilizing data from the National Survey on Drug Use and Health (NSDUH), the researchers analyzed a large sample (N = 1,005,421) to explore patterns in self-reported risk-taking across the lifespan.

Methods

The study utilized data from the NSDUH, a representative survey of the noninstitutionalized civilian population in the United States. The researchers analyzed self-reported risk-taking propensity using linear and nonlinear models, examining variation across 19 population subgroups defined by factors such as biological sex, race/ethnicity, socioeconomic status, population density, religious affiliation, and mental health.

Results

The study found that self-reported risk-taking propensity exhibited a nonlinear trajectory across the lifespan, with a robust peak during mid-adolescence (∼16 years old) for the full sample. Notably, this peak in risk-taking propensity was observed across 18 out of 19 population subgroups, indicating its generalizability across diverse populations. While the timing of the peak was consistent, the magnitude of risk-taking varied significantly across subgroups. For example, adolescent males reported higher risk-taking propensity than females, and white adolescent participants reported higher risk-taking propensity than other racial/ethnic groups. Importantly, higher levels of self-reported risk-taking propensity between ages 15-18 significantly distinguished adolescents who regularly used cannabis or alcohol from non-using adolescents.

Discussion

The findings suggest that the lifespan peak in self-reported risk-taking propensity during adolescence and its link to substance use are generalizable across diverse social, environmental, and psychological factors. This provides support for the potential broad applicability of intervention and prevention efforts that target risk-taking as a vulnerability factor for adolescents. The generalizability of the observed peaks in risk-taking propensity highlights the importance of developmentally specific behavioral features for tailoring interventions. However, the variability in the magnitude of risk-taking across individuals underscores the need for personalized assessment and intervention approaches.

The study acknowledges limitations such as the use of brief self-report data. Future research is recommended to further refine these results using comprehensive measures, neuroimaging, and longitudinal assessments. The researchers emphasize the value of integrating large-scale national survey data with targeted cohort studies to improve the rigor and generalizability of developmental science findings.

Summary

This study used a large dataset from the National Survey on Drug Use and Health (NSDUH) to examine if the peak in risk-taking behavior during adolescence is consistent across various groups of people.

Methods

Researchers analyzed data from over 1 million individuals aged 12 and older, focusing on a self-reported risk-taking propensity measure. They investigated how this measure varied across 19 subgroups defined by factors such as sex, race/ethnicity, socioeconomic status, population density, religious affiliation, and mental health. The study aimed to determine if the peak in risk-taking during adolescence was generalizable and if there was a link between risk-taking and substance use.

Results

The findings showed that a peak in self-reported risk-taking propensity occurred during mid-adolescence (around 16 years old) across almost all subgroups. However, the magnitude of this peak varied significantly depending on the subgroup. For example, adolescent males and white adolescents reported higher risk-taking propensity than their counterparts. Importantly, regardless of the subgroup, adolescents with higher levels of risk-taking between ages 15 and 18 were more likely to regularly use cannabis or alcohol.

Discussion

These findings suggest that the peak in risk-taking during adolescence is a generalizable phenomenon across diverse populations, highlighting the importance of targeted interventions and prevention efforts focused on this period. While the timing of the peak is consistent, its magnitude varies significantly depending on individual factors, emphasizing the need for personalized approaches to assessment and intervention. The researchers acknowledge the limitations of using self-reported data and recommend future research using multiple levels of analysis, including neuroimaging and behavioral assessments.

Summary

This study looked at a large group of people to see if the idea that teenagers are more likely to take risks is true. They found that it is! Teens are more likely to say they like to do risky things, and this is true for teens from all kinds of backgrounds.

Methods

The study used data from a survey of people in the United States. The survey asks people about their risk-taking habits and if they use alcohol or marijuana.

Results

The study found that teens are more likely to say they like to do risky things, even when looking at people with different backgrounds. They also found that teens who say they like risky things are more likely to use alcohol or marijuana.

Discussion

This study shows that teens are more likely to take risks, and this is true for most teens. This means that programs to help teens avoid risky behavior might be helpful for many teens. However, the study also found that some teens are more likely to take risks than others, based on things like their race, gender, or family income. This means that programs to help teens should consider these differences and tailor their programs to different groups.