Adolescents with intellectual disability (ID) engage in risky behavior and offending. However, little is known on the impact school-related predictors have on engagement in risky behaviors for adolescents with ID. This study analyzed secondary data from the National Longitudinal Transition Study-2 (NLTS2) to determine levels of engagement in risky behaviors and offending for adolescents with mild and moderate/severe ID. School-related predictors of engagement for adolescents with mild ID were also explored. Results indicated adolescents with mild ID engage in risky behaviors and offending at significantly higher rates as compared to adolescents with moderate/severe ID. Participation in a social skills or life skills class was a significant predictor of less engagement in risky behaviors for individuals with mild ID.
Adolescent engagement in risky behaviors increased in recent years (Huang, Lanza, Murphy, & Hser, 2012). While minor engagement in risky behaviors (e.g., smoking, alcohol consumption, sexual activity) can be considered typical behavior for adolescents (McNamara & Willougby, 2010), continued involvement could result in negative health outcomes (e.g., sexually transmitted infections; Wang et al., 2014) or offending (i.e., committing an illegal act; Lindsay et al., 2010). In addition, adolescents who engage in one risky behavior are more likely to engage in multiple risky behaviors, which places them at a higher risk for long-lasting negative health outcomes (Wang et al., 2014).
Adolescents with disabilities face additional challenges when deciding whether or not to engage in risky behaviors as compared to adolescents without disabilities. Such challenges include increased difficulty in understanding social conventions (e.g., not understanding the difference between a peer liking them and a peer making fun of them) as well as trouble understanding criminal law (Asscher, van der Put, & Stams, 2012). With increased inclusion in general education settings and the community, adolescents with disabilities can be more vulnerable to engagement in risky behaviors (Slayter, 2010). These concerns are especially true for adolescents with intellectual disability (ID). Individuals with ID may have limited understanding of the health detriments of smoking, drug use, and alcohol abuse (Kerr, Lawrence, Darbyshire, Middleton, & Fitzsimmons, 2013). Individuals with ID may also have limited understanding of sexuality, sexual abuse, and legal rights such as age of consent (O'Callaghan & Murphy, 2007). In addition to limited understanding of risky behaviors, results of a study indicated adolescents with ID have higher rates of starting fights, threatening others, trouble with police, and use of weapons compared to adolescents without ID (Dickson, Emerson, & Hatton, 2005).
According to Gray, Fitzgerald, Taylor, MacCulloch, and Snowden (2007), “individuals with intellectual disability represent a sizable minority of the offender population” (p. 474). Out of the numerous ways individuals can become an offender (e.g., drug dealing, stealing, murder, sexual assault), contact sexual offenses (e.g., rape), noncontact sexual offenses (e.g., public indecency), and property damage are most commonly reported as the offenses of individuals with ID (Lindsay et al., 2010). A recent study also indicated rates of offending against persons (e.g., assault, rape) were higher for adolescents with ID compared to adolescents without ID (Asscher et al., 2012).
While there is some research available on adolescents with ID and engagement in risky behaviors and offending, recent research examining the differences between adolescents engaging in risky behaviors or offending by their level of functioning (i.e., mild vs. moderate/severe) is minimal. Findings from one study indicated substance use was significantly more likely in individuals with mild ID compared to individuals with moderate or severe ID (Chaplin, Gilvarry, & Tsakanikos, 2011). Researchers acknowledged a need for large sample studies to establish base rates of engagement in risky behaviors and offending for subgroups of individuals with ID (e.g., mild ID, Chapman & Wu, 2012). In Asscher et al. (2012), researchers concluded if levels of functioning were taken into account, different results may have emerged when comparing adolescents with and without ID on levels of offending.
In addition to determining the level of engagement in risky behaviors and offending for subgroups of individuals with ID, exploring predictors of such engagement is needed. A vulnerable time for individuals with ID who engage in risky behavior (e.g., substance use) is during early to late adolescence (i.e., during school years; Chapman & Wu, 2012). If predictors of engagement in risky behaviors are known, educators and others can address the associated risk factors through prevention and intervention (Katsiyannis, Thompson, Barrett, & Kingree, 2012). Although multiple risk factors contribute to engagement in risky behavior, school-related factors are consistently identified as key predictors of engagement in risky behaviors and offending for adolescents with and without disabilities (Katsiyannis et al., 2012; Krezmien, Mulcahy, & Leone, 2008). In particular, lack of preventative measures (e.g., violence prevention, social skills training; Martinez, 2009), disciplinary action history (e.g., school suspensions; Heitzeg, 2009), and lack of involvement in extracurricular activities (Eccles, Barber, Stone, & Hunt, 2003) are linked to engagement in risky behaviors.
This study involved a secondary analysis of the National Longitudinal Transition Study-2 (NLTS2) database relative to adolescents with mild ID and moderate/severe ID; school-related predictors; and post-school outcomes related to risky behaviors and offending. The NLTS2 was a 10–year project gathering data on both in school and post-school experiences of adolescents with disabilities through parent and/or student surveys, direct student assessments, teacher surveys, school program and characteristic surveys, and student transcripts (SRI International, 2000a). Data collection for the NLTS2 began during the 2000–2001 school year and occurred in five waves, each equaling a 2-year period of data collection, and ending in 2010. This particular secondary analysis attempted to answer the following questions: (a) To what extent do adolescents with mild ID and moderate/severe ID engage in risky behaviors and offending? (b) What is the relationship between engagement in risky behaviors and offending between adolescents with mild ID and moderate/severe ID? And (c) what school-related factors predict the likelihood to engage in risky behaviors or offending for adolescents with mild ID? The last research question initially included an analysis for individuals for moderate/severe ID as well. However, due to the lack of engagement in risky behavior and offending reported for this group of individuals (refer to results from research question one), further analysis wasn't warranted at this time.
The NLTS2 used a two-stage sampling procedure (SRI International, 2000b). First, through a stratified random sample (i.e., geographic region, size, community wealth), Local Educational Agencies (LEAs) and state-supported schools were selected who served students between the ages of 13 and 16. Next, the SRI researchers randomly selected students from each school until they received a sufficient sample from each disability category. Using the weighted design of the study, a total of 19,899,621 students receiving special education services from 12,435 LEA were represented in the NLTS2 study (SRI International, 2000b).
This particular analysis focused on students with mild ID and moderate/severe ID. To be included in this analysis, students had a primary disability on their Individualized Education Program (IEP) of mild ID or moderate/severe ID as reported on the School Program survey; students who were identified as having both were excluded. Next, students who did not respond to any of the dependent variables of interest were excluded. There were 940 students with ID in this study (i.e., the unweighted number of participants). Based on weighted data, the analysis represented 205,899 students with ID. All results and demographic information were reported using weighted data. Weighting produced estimates regarding the population and were computed by accounting for youth and school characteristics used in stratifying variables in the sampling as well as nonresponses in those strata (Newman, Madaus, & Javitz, 2016; see Newman et al., 2012 for more details on the weighting strategy used in the NLTS2).
Of the 205,899 students, 65.1% had a primary disability of mild ID and 34.9% a primary disability of moderate/severe ID. There were more male (59.8%) as compared to female (40.2%) students with ID. Participants were predominantly Caucasian (61.2%), followed by African American (29.5%), Hispanic (6.4%), and other ethnicities (3.0%). At the time of the school interview (i.e., wave 1 or 2), the majority of students were 17–18 years old (38.6%), followed by 15–16 (29.1%), 19–20 (26.4%), and then 13–14 years old (5.9%). The most frequently identified parental income was equal to or under $25,000 (43%), followed by greater than $50,000 (28.9%), and then between $25,001 and $50,000 (28.2%).
To answer the research questions, we worked with the Parent/Youth survey (i.e., a phone survey completed by the student, when appropriate, or parents) and the School Program survey (i.e., a mail survey completed by the teacher most familiar with the student). The variables used for analysis included ones representing in-school experiences or information (i.e., demographic variables, predictor variables, and weight variables) and variables representing outcomes. All outcome variables were reported within two years of the reported in-school variables (i.e., in the subsequent wave of data collection). For the predictor variables, we focused on ones that represented school-related protective or risk factors across three areas including: (a) preventative instruction which encompassed receipt of a social skills or life skills class, substance abuse prevention, reproductive health education, and conflict resolution prevention; (b) history of disciplinary actions which comprised of any history of suspension or expulsion incidents; and (c) involvement in extracurricular activities provided by the school (i.e., involvement in in-school extracurricular activities). For the outcomes variables, we chose risky behaviors similar to risky behavior variables used in Wang et al. (2014): smoking, alcohol consumption, drug use, sexual activity, and physical violence. Students' arrest record was included as an outcome variable to represent offending. We also used demographic variables such as age, gender, ethnicity, and family income to provide a description of our participants.
After identifying relevant variables, we created a database of predictor variables and outcomes variables. To focus on immediate outcomes, we used subsequent waves. In other words, data from predictor variables came from Waves 1 and 2 and data from outcome variables from Waves 2 and 3, respectively. All variables used in this analysis were categorical or dichotomous, including demographic variables. Although we used some of the original NLTS2 variables, we also recoded some categorical variables by merging categories. Age categories were coded 1–4 for students who were 13–14 years, 15–16 years, 17–18 years, and 19–20 years respectively; gender was coded as male (1) and female (2); ethnicity was coded 1–4 for Caucasian, African American; Hispanic; and other respectively; and income was coded 1–3 for less than or equal to $25,000, $25,001–50,000, and $50,001 and greater respectively. Preventative instruction variables and extracurricular activity involvement were already represented with dichotomous responses in the original survey (0 = no, 1 = yes). The variables on disciplinary actions including in-school suspensions, out-of-school suspensions, and expulsions were combined to form one dichotomous variable with 0 representing the student was never suspended or expelled or 1 representing students who were suspended and/or expelled during the current school year. The outcome variables of smoking and alcohol use were recoded to a represent a dichotomous response with youth who did not smoke or consume an alcoholic beverage in the past 30 days represented with a 0 and youth who smoked or drank an alcoholic beverage in the past 30 days represented with a 1. Drug use, sexual activity, physical aggression, and arrest record were already represented with dichotomous responses in the original survey (0 = no, 1 = yes).
Not all students responded to every question in the original NLTS2 data collection (e.g., students may have skipped questions, attrition may have occurred). Actual response rates for each question/variable were not collected. The frequencies reported for all variables were based out of the number who responded from the population and not the overall number of students with mild ID or moderate/severe ID reflected in the study.
To answer research question one (i.e., the extent to which students with mild ID and moderate/severe ID engage in risky behaviors and offending), we conducted descriptive analyses. Frequency distributions were created for all variables of interest (i.e., smoking, alcohol consumption, drug use, sexual activity, physical aggression, arrest record) and results were summarized for all students with mild ID and students with moderate/severe ID separately, contingent on students responding yes or no to the variables of interest. Please note, to account for the weighted data in the all analyses, the researchers used SPSS Complex Samples. For the second research question (i.e., the relationship between engagement in risky behaviors and offending for students with mild ID compared to students with moderate/severe ID), both groups were compared with a significance test provided with the NLTS2 to indicate differences between disability groups on each dependent variable of interest; the significance test was an F-test. The formulas for the significance test were provided on an Excel spreadsheet with the NLTS2 restricted-use database. We inputted the frequencies and standard errors for the two groups into the spreadsheet to identify significant differences between the two. Refer to the Institute of Education Sciences (IES; n.d.) and Newman, Wagner, Cameto, & Knokey (2009) for more information regarding this test.
To answer research question three, logistic regression analyses were conducted for each of the six binary dependent variables of interest: smoking, alcohol consumption, drug use, sexual activity, physical aggression, offending (i.e., arrest record). Models included preventative measures, disciplinary action history, and involvement in extracurricular activities which have been linked to decreased engagement in risky behaviors (Eccles et al., 2003; Heitzeg, 2009, Martinez, 2009). Specifically, each of the six models included the following independent variables: participation in a social skills or life skills class, history of disciplinary actions including suspensions and expulsions, and extracurricular activity involvement in school. Per research and theory, the appropriate preventative education variable was also included in each model. Substance abuse education was added as an independent variable in the models for smoking, alcohol consumption, and drug use. Likewise, reproductive health education was added as an independent variable in the model for sexual activity and conflict resolution education in the model for physical aggression. The model with the dependent variable of offending included substance abuse prevention, reproductive health, and conflict resolution education as independent variables. Finally, student characteristics (i.e., age, gender, ethnicity, and family income) were included in each logistic regression model as potential covariates if they showed a significant association with the dependent variable. This association was determined for each model with a Goodness of Fit test. The following covariates had significant associations: age with physical aggression, gender with alcohol consumption and physical aggression, ethnicity with sexual activity, and income with sexual activity and offending. The McFadden's R-squared test was used to determine how well each model fit the data (McFadden, 1974).
The extent students with mild ID engaged in risky behaviors and offending ranged from 2.4% (engaged in drug use) to 39.4% (sexually active). For students with moderate/severe ID, engagement ranged from 0% (engaged in smoking; engaged in drug use) to 17.9% (engaged in a physical fight). Results indicated students with mild ID engaged in higher rates of risky behavior and offending compared to students with moderate/severe ID. Significant differences between groups were found for smoking (p < .01), sexual activity (p < .001), and offending (p < .05). Refer to Table 1 for specific results for outcomes on each risky behavior and offending for students with mild ID and moderate/severe ID.
Results from the logistic regression analyses indicated receipt of a social skills or life skills class and receipt of a substance abuse preventative education were statistically significant predictors of not engaging in smoking (p < .001). For students who did not receive a social skills or life skills class, the chance of smoking were 49 times higher. For students who did not receive a substance abuse instruction, the chance of smoking were 8.9 times higher. History of suspension or expulsion and involvement in extracurricular activities were not significant predictors of engagement in smoking.
In terms of alcohol consumption, receipt of substance abuse preventative education, incidents of suspension and/or expulsion, and gender were statistically significant predictors at the p < .01 level of not engaging in alcohol consumption. Students who did not receive substance abuse prevention education were 6.1 times more likely to consume alcohol as compared to those who did receive such instruction. In contrast, students with zero suspensions and/or expulsions were 11.9 times more likely to consume alcohol as compared to those with at least one suspension or expulsion. Being male was also a predicator of alcohol consumption. Receipt of a social skills class and involvement in extracurricular activities were not significant predictors of engagement in alcohol consumption. For drug use, no significant predictors of engagement in drug use or improbable results were reported.
Findings indicated receipt of a social skills or life skills class and income were statistically significant predictors of non-engagement in sexual activity for students with mild ID (p < .05). Students who did not receive a social skills or life skills class were 3.6 times more likely to be sexually active, and students from lower income households were more likely to be sexually active than students with higher socioeconomic status. Receipt of a reproductive health class, history of suspension or expulsion, involvement in extracurricular activities, and ethnicity were not significant predictors of engagement in sexual activity.
Engagement in physical aggression (i.e. fighting) was predicted by receipt of a social skills or life skills class (p < .01), receipt of conflict resolution instruction (p < .01), incidents of suspension and/or expulsion (p < .001), age (p < .05), and gender (p < .001). Males were more likely to engage in physical altercations, as were younger adolescents. Students who did not receive a social skills or life skills class in school were 119.8 times more likely to engage in physical aggression as compared to those who received such instruction. In contrast, students who received conflict resolution instruction were more likely to engage in physical fighting (15.9 times). Finally, students who received a suspension and/or expulsion while in school were 50 times more likely to engage in physical aggression. The only predictor that was not a significant in this model was involvement in extracurricular activity.
In terms of being arrested, the only statistically significant predictor was conflict resolution instruction (p < .05). Like with physical aggression, adolescents with mild ID who received conflict regression in school were 4.5 times more likely to be arrested. Refer to Table 2 for results of all logistic regression analyses.
Adolescents with ID put themselves at risk by engaging in risky behaviors and offending (Wang et al., 2014). It was important to determine the level at which this engagement occurs as well as understand the predictors that increase—or decrease—the likelihood of engagement. For practitioners, specifically, understanding school-related predictors can help them provide more appropriate supports to students with mild ID to lessen the probability of engagement in risky behaviors and offending. The major findings of this study are: (a) adolescents with mild ID engage in some risky behaviors and offending at significantly higher rates compared to adolescents with moderate/severe ID, (b) participation in a social skills or life skills class significantly decreases the likelihood of engagement in some risky behaviors and offending for adolescents with mild ID, (c) receiving instruction in specific prevention education as well as students' disciplinary action history produces mixed results for adolescents with mild ID in relation to engaging in risky behavior and offending, (d) involvement in extracurricular activities does not have a significant effect on engagement in the risky behaviors or offending measured in this study, and (e) certain demographics significantly increased the likelihood of engagement in risky behaviors.
The results extend the current literature by comparing subgroups of adolescents with ID using a national sample (Chapman & Wu, 2012). For adolescents with mild ID, engagement in risky behavior and offending occurs at significantly higher rates compared to adolescents with moderate/severe ID. These results are not surprising, as individuals with higher levels of intellectual functioning tend to live more independent in the community and experience greater access to engage in risky behaviors (Slayter, 2010). When compared to adolescents in the general population, students with mild ID engage in risky behaviors at similar or lower rates (CDC, 1991-2013). However, individuals with ID are less likely to seek treatment or engage in treatment retention for risky behaviors (e.g., substance abuse) making prevention efforts more imperative (Slayter, 2010).
School Predictors of Engagement in Risky Behaviors and Offending
Participation in a social skills or life skills class was the only school-related predictor in this analysis that did not produce mixed results for risky behaviors and offending for students with mild ID. Participation in a social skills or life skills class had significant effects on three of the six outcome variables. These results indicate students with mild ID should receive more explicit instruction when learning about topics such as decision-making and safety to make more informed decisions related to engagement in risky behaviors and offending (Sitlington, Frank, & Carson, 1993).
In regards to other specific preventative instruction (e.g., substance abuse, conflict resolution, reproductive health), participation produced mixed results. While receipt of substance abuse prevention significantly decreased the likelihood of engagement in risky behaviors such as smoking and alcohol consumption, receipt of conflict resolution significantly increased the likelihood of engaging in physical aggression and being arrested. The study design does not allow us to investigate the quality or components of the preventative education or services received, so we are not able to determine why preventative education decreased the likelihood of engagement in risky behaviors and offending in some circumstances and increased the likelihood in others. For example, some students may have received an information packet about reproductive health without practitioner-directed instruction and some students may have received comprehensive reproductive health education training. The comprehensiveness, duration, and intensity of a program as well as the practitioner's level of training influences the effectiveness of the prevention program (Robin et al., 2004). Likewise, in considering the positive relationship between receipt of conflict resolution and increased likelihood of fighting, it may be that conflict resolution was used as an intervention for a previous fight, abet ineffective intervention, rather than a preventative measure. Perhaps students may need instruction in conflict resolution skills prior to engaging in an altercation.
Similar to specific preventative intervention, a history of suspension and expulsion also produced mixed results. Students with no history of suspension and/or expulsion were more likely to consume alcohol than students who had a history of suspension and/or expulsion. This finding was surprising because disciplinary problems in school are routinely linked to risk-taking and violent behaviors (Katsiyannis et al., 2012). Perhaps, since a majority of adolescents consume alcohol for social facilitation (e.g., to improve social gatherings, get into party mood; Kuntsche, Knibbe, Gmel, & Engels, 2005), adolescents with a suspension and/or expulsion history did not have the opportunity to spend as much time with their peers in social settings (e.g., parties) compared to adolescents with no suspensions or expulsions (e.g., they were grounded). The validity of parent-report of this behavior should be considered as well. However, consistent with previous findings, students who were suspended and/or expelled were significantly more likely to get into a physical fight after the suspension and/or expulsion (Katsiyannis et al., 2012). Students who are suspended are likely to continue to display the same or more severe behaviors after suspension (Christie, Nelson, & Jolivette, 2004). Sending students out of a supervised environment (e.g., expulsion) without providing needed services can intensify the problem (Christie et al., 2004). Alternatives to exclusionary disciplinary measures should be explored for students with ID as they may improve future outcomes (e.g., community service around the school, peer support groups).
Involvement in extracurricular activities did not have a significant impact on increasing—or decreasing—the likelihood of engaging in risky behaviors or offending. This differs from previous studies represented in a literature review where researchers found positive associations between extracurricular involvement and pro-social behavior (Feldman & Matjasko, 2005). Overall, the results in the current study suggest other variables not measured in this study (e.g., peer influence in extracurricular activities, intensity or duration of involvement in extracurricular activities, type of extracurricular activity) may present a profound impact on engagement in risky behaviors and offending compared to the dichotomous variable of involvement or no involvement used in this secondary analysis.
Demographics and Engagement in Risky Behaviors and Offending
Demographics were influential on the impact of engagement in particular behaviors. Being male was a significant predictor of engagement in alcohol consumption. This is consistent with previous findings from studies examining alcohol consumption for individuals with ID (Chaplin et al., 2011). For engagement in sexual activity, students from a lower socioeconomic status household were more likely to be sexually active. Previous researchers linked neighborhood poverty with adolescent risk-taking behavior, including early initiation of sexual activity (DiClemente, Salazar, Crosby, & Rosenthal, 2005; Wang et al., 2014). Adolescents who choose to engage in sexual activity may lack knowledge of safe practices or have difficulty obtaining or using contraception, putting themselves at an increased risk for unplanned pregnancy or sexually transmitted infections. Finally, males and younger adolescents were more likely to engage in physical aggression. These results are consistent with previous literature with more males engaging in a physical fight compared to females and younger adolescents engaging in physical fights at a higher rate than older adolescents (Wang et al., 2014). As students age, they may be less impulsive and consider other alternatives to physical aggression.
Limitations and Future Directions
There were several limitations to this study related to using the NLTS2. Results from a secondary analysis are subject to limitations of the original data collection and lack of control over questions and designs. For one, the NLTS2 lumps receipt of social skills with receipt of life skills training. While we understand receipt of social skills training and receipt of life skills training are not the same, it was represented as such in the questionnaire and we were unable to separate receipt of a social skills class from a life skills class. Another limitation involves the self-reported data of the NLTS2; self-reported data have potential for biases, especially with more sensitive questions or information. For example, parents or youth may underreport drug use to appear more pleasing. In addition, when youth were unable to respond to questions, parents responded in their place. It is possible that the informant may make a difference in the answers provided. Missing data are also a limitation. Although the analysis was limited to students with ID who responded to at least one of the dependent measures, not all of these students responded to each of the six outcome variables of interest. In addition, this study collected data from multiple instruments, across multiple time points. Use of a cross-wave or cross-instrument weight may have been more appropriate for this analysis.
One limitation directly related to the secondary analysis involves the focus on only immediate outcomes. However, we felt school-related predictors were more relevant to immediate outcomes as opposed to long-term outcomes (i.e., being out of school longer). The longer one is out of school, the more likely that out-of-school experiences (e.g., living arrangement, employment, friendships) may predict risky behavior and outcomes. Finally, there were limitations with the logistic regression analyses conducted. In regards to statistical significance, readers should be cautious not to infer the meaningfulness of differences (Newman et al., 2009). With large samples, a small p-value can occur even though the sample estimate is very close to the parameter value in the null hypothesis. Also, some of the results of the logistic regression produced improbable results into the millions. The predictor variable of discipline for youth smoking, social skills or life skills class for youth drug use, and substance abuse instruction for youth drug use were among the improbable results in this analysis. Causes of these results can be produced by multicollinearity, having a zero cell count (i.e., a frequency of zero in a contingency table), and complete separation (i.e., two groups are perfectly separated by the scores on one or more independent variable; Hosmer, Lemeshow, & Sturdivant, 2013). While multicollinearity levels were all well into acceptable ranges, our analyses suggest zero cell counts or complete separation on one or more independent variables.
Additional research is needed to understand students with mild ID and their engagement in risky behaviors and offending. In this study, we explored some school-related predictors using data from the NLTS2. However, our investigation is exploratory in nature and more variables are needed to capture the complex nature of the relationship between students with mild ID and making decision to engage in risky behaviors or decisions resulting in offending. For example, this analysis did not take into account functional capabilities of participants. Individual differences (e.g., challenges with speech or hearing, physical impairment) may have impacted engagement or non-engagement in risky behaviors and offending.
The same school-related predictors could be explored across other disability categories to determine if participation in a social skills or life skills class, preventative education participation, history of disciplinary actions, or extracurricular activity involvement produce the same effect on engagement in risky behaviors and offending as findings in this secondary analysis. In regards to the preventative effect of extracurricular activities, researchers should consider exploring extracurricular involvement with involvement in risk factors during the same data collection wave for students with ID rather than in the subsequent wave used in this analysis. For preventative education specifically, a more in-depth look into programs across school districts (e.g., components, duration and intensity of instruction) and their effectiveness, including effects on future engagement in risky behaviors and offending is warranted for students with mild ID.
The results of this study suggest a need for prevention and, if appropriate, intervention for risky behaviors. While many prevention program materials are designed for school-based use, there is little research addressing the inclusion of adolescents with ID in these programs nor the appropriateness or effectiveness of these programs for adolescents with ID. Based on the results that show participation in a social skills or life skills class can decrease the likelihood of future engagement in risky behaviors and offending, educators need to think about their students' curriculum and how they can incorporate such skills into their learning. Also, educators should determine the appropriate duration and intensity, based on their students' needs, to dedicate to teaching about the dangers associated with engagement in risky behaviors as these factors may play a role in future engagement (Robin et al., 2004). Other factors influencing decisions to engage or not engage in risky behaviors (e.g., peers, supervision at home) should also be explored when developing and implementing curricula.