Adolescence and young adulthood are periods of life associated with a number of challenges, including entering the world of work and increasing independence from parents. Addictive behaviors such as consumption of alcohol, tobacco, and cannabis [1] but also problematic internet use [2] are particularly widespread in this phase of life.

Many adolescents and young adults in Germany attend vocational schools to start an apprenticeship or attend vocational grammar school or vocational preparation courses. Vocational schools are promising settings for prevention efforts [3], especially as there is evidence that students in vocational schools report even higher rates of substance use than other adolescents and young adults. For example, in a survey of 5688 German vocational school students, the reported smoking rate was more than twice as high as that of their peers in the general population [4]. Representative data from German vocational school students in Mecklenburg-Western Pomerania also show particularly high rates of problematic alcohol consumption (62%), smoking (62%), and cannabis use (26%). The study [5] also provided a first insight into the prevalence of media-related problems among German vocational school students. Problematic internet use was reported by 17% of vocational school students, problematic gaming was reported by 10%, and potential smartphone addiction was reported by 17% [5]. This is supported by findings from an intervention trial to reduce problematic use of the internet in vocational school students [6].

In vocational school settings, achieving prevention goals typically requires a shift away from primary prevention strategies. Instead, indicated prevention programs are considered more appropriate as the vast majority of vocational school students have already initiated addictive behavior. A high proportion of vocational school students also engage in multiple health risk behaviors, especially multiple addictive behaviors [5,7]. However, they differ widely in their risk profiles (ie, in the combinations and number of addictive behaviors). Vocational school students also differ in other factors that may contribute to addictive behaviors, such as low social competence or high work stress. This means that complex interventions are needed to address different risk profiles.

Mobile interventions are best suited to address various risk profiles as they can be easily adapted to the needs of the users and allow for timely and anonymous participation. Mobile interventions have been shown to be effective in changing multiple addictive behaviors in vocational school students [8,9].

Mobile interventions have the advantage of automated delivery of highly individualized content through computer tailoring. Tailoring is a key component in ensuring the effectiveness of an intervention [10]. With tailoring, feedback that is relevant to the individual is derived from an individual assessment based on prespecified rules. Examples include content or wording tailored to users’ gender or motivation to change. It has been shown that tailored feedback compared to generic feedback is more likely to be processed and remembered [11]. In addition to automatic computer tailoring, some interventions include forms of self-tailoring, such as allowing participants to choose the medium [12], dose [13], or content (eg, specific behavioral targets or modules [14-17]). In line with theoretical frameworks such as self-determination theory, self-tailoring is assumed to increase autonomy [18].

Although mobile health interventions are effective, adherence is a major challenge [19], including approaches among vocational school students [20]. Low adherence may partly explain the lack of long-term effects and small effect sizes in mobile health interventions [21]. Allowing participants the autonomy to choose which behavioral modules to engage with could positively influence engagement and efficacy [22,23]. Within the vocational context, offering a choice of intervention modules could be valuable, particularly as many vocational school students exhibit multiple addictive behaviors [5,7]. Attempting to address all these behaviors simultaneously might be overwhelming [24], especially considering the decreasing intention to change all behaviors as their number increases [25].

While some research has explored participants’ module choices in multibehavior interventions [14,15,26-31], there is limited understanding of the underlying choice mechanisms, such as which personal characteristics are related to choice. Investigating this is crucial as diverse subgroups may use interventions differently and some may not use it as recommended (eg, not choosing the recommended modules). Such insights can inform tailored strategies to align interventions with participant needs, reducing nonadherence and enhancing intervention exposure and effectiveness.

Most of the evidence on module choice comes from multiple-behavior interventions in adults. These studies have shown that certain modules are preferred over others, for example, physical activity over diet [14,15,26]. There is also consistency that health-promoting behaviors such as physical activity and diet are preferred over substance-related modules [14,15,26-31]. In total, 2 studies among vocational school students reported on preferences within multiple–addictive behavior interventions. They showed that stress and non–substance-related behaviors such as social media use were chosen more often than substance-related behaviors such as smoking, alcohol or cannabis use [8,17].

With regard to potential determinants of module choice, the results are less consistent [14-16,27] and are likely to vary depending on the number and topics of available behavior modules to choose from, whether module recommendations were given, and whether any restrictions were placed on module choice (eg, how many modules could be chosen or whether modules that were not recommended could be chosen). Providing feedback or recommending modules may enhance the alignment between participants’ module choice and their prevention needs [30]. Studies show that people are more inclined to choose modules that were recommended to them based on their demonstrated risk profile [17,29]. Consequently, most previous studies have used feedback or recommendations to guide module choice [14-17,26,29,32], with only a few exceptions [27,28,30,33].

In online multiple-behavior interventions for adults addressing a range of health-promoting behaviors and substance use, choice of the physical activity– or diet-related modules was higher among women [14] and those with healthier lifestyles [15]. In the study by Schulz et al [15], choice of the alcohol module was linked to being a man, lower income, and higher psychosocial distress, whereas medium (vs high) educational level, lower quality of life, and lower psychosocial distress were linked to choice of the smoking module. Reinwand et al [34] showed that adherence to module recommendations in online multibehavior interventions (ie, starting all recommended modules vs fewer) also depended on individual characteristics. Individuals were more likely to start all the recommended modules if they were older, a woman, unemployed, ill, and in a relationship.

The main gap in the existing research is that no study has reported on the determinants of module preferences among vocational school students or statistically analyzed the alignment between prevention needs and module preferences in this group. This is particularly important given that a high proportion of vocational school students engage in multiple addictive behaviors. Furthermore, although there have been previous studies on the prevalence of addictive behaviors among German vocational school students [5,35], these studies did not analyze prevention needs in terms of life skills and their determinants. This has only been investigated in a Swiss sample [17].

Given the fact that lower educational attainment among vocational school students is associated with higher prevalence [35] and clustering [5] of addictive behaviors, as well as lower engagement with mobile interventions [36], it is particularly important to analyze educational track as a potential determinant of module choice.

This study examined the prevention needs of German vocational school students as well as their module choices within the app-based addiction prevention intervention ready4life. The ready4life intervention offers 6 behavior modules: alcohol, tobacco, cannabis, social media and gaming, stress, and social competencies. After receiving feedback on their individual risks and competencies in the form of a traffic light system to guide choice, students could freely choose 2 out of the 6 modules to be coached in. In terms of module choice, we aimed to examine (1) how often each intervention module was chosen, (2) how module choice aligned with the traffic light feedback received, and (3) whether individual characteristics (age, gender, educational track, and prevention needs) were associated with module choice. In addition, the relationship between sociodemographic factors and module choice is likely to be influenced by the prevention needs of the students in this sample. Therefore, we also analyzed how sociodemographics (age, gender, and educational track) were associated with prevention needs. This would provide a clearer understanding of how these factors may shape students’ intervention preferences.

The prevention needs of vocational school students in terms of addictive behaviors were expected to be high [5]. On the basis of previous literature on module choice [8,14,15,17,26-31] and the generally low intentions to change addictive behaviors [5], it was expected that life skill modules (stress or social competencies) would be preferred over modules related to addictive behaviors. It was also expected that module choice would show a high alignment with the traffic light feedback received [8,17,29]. Given the differences in prevalence of substance use between the genders [37], it was expected that men would choose modules related to substance use more often than women.

The data came from a cluster randomized controlled trial among German vocational school students testing the efficacy of the app-based addiction prevention program ready4life [38]. The trial was registered in the German Clinical Trials Register (DRKS00022328).

Ethics approval was granted by the ethics committees of the University of Lübeck (19-419) and the University Medicine Greifswald (BB 024/20). Digital informed consent was obtained from vocational school students through the app, with no additional parental consent necessary in accordance with the European Union General Data Protection Regulation. Underaged participants were encouraged to inform their parents or guardians about study participation. Students completed the app-based screening anonymously. No names were collected; instead, students chose nicknames. Data protection was ensured through encrypted data transmission and storage within the app. For those who provided informed consent, research data were stored in a pseudonymized format, and contact information (email or phone number) was securely stored on a separate server. Participants were not offered any compensation other than the chance to win prizes.

The flow of the participants is depicted in Figure 2. Study participants are described in Table S4 in Multimedia Appendix 1. Among the total sample, the mean age was 19.7 (SD 3.7) years, and 45.6% (1170/2568) were women. Most study participants (1648/2568, 64.2%) were in vocational training, followed by vocational grammar school (582/2568, 22.7%) and vocational preparation (282/2568, 11%). Within those students in vocational training, the most prevalent occupations were technicians and associate professionals (511/1648, 31%), as well as craft-rated trade workers (422/1648, 25.6%). Most students (1785/2568, 69.5%) were in their first or second year of education.

In terms of prevention needs, students on average perceived a moderate level of stress (mean 3.29, SD 1.20 on a scale from 1 to 5) and reported a moderate level of social competencies (mean 29.50, SD 4.69 on a scale from 8 to 40). The mean score on the Short CIUS was 9.14 (SD 4.18), which is above the threshold for problematic use of the internet. The median number of standard alcoholic drinks consumed per day was 0.20 (IQR 0.00-1.00). A proportion of 31.7% (814/2568) and 18.1% (464/2568) had used tobacco and cannabis in the previous month, respectively.

Of the total sample, 45.1% (1158/2568) of the students used the app on Android, and 54.9% (1410/2568) used it on iOS. Notably, iOS users reported significantly higher prevention needs across all behavioral domains except social competencies, where they had lower needs compared to Android users. In addition, iOS users were more likely to select the tobacco module (data not shown).

When analyzing prevention needs as outcomes, ICCs were highest for alcohol consumption, indicating that 13.5% of the variance in the number of standard alcoholic drinks per day was attributable to class membership (Table S5 in Multimedia Appendix 1). Being a woman was associated with significantly higher stress (regression coefficient=0.70, 95% CI 0.60-0.80; P<.001), higher problematic internet use (regression coefficient=0.74, 95% CI 0.39-1.09; P<.001), and lower social competencies (regression coefficient=−2.27, 95% CI −2.67 to −1.88; P<.001). In contrast, being a woman was also associated with lower substance use (lower alcohol consumption: regression coefficient=−0.40, 95% CI −0.55 to −0.24 and P<.001; lower odds of tobacco use: odds ratio [OR] 0.79, 95% CI 0.65-0.97 and P=.02; lower odds of cannabis use: OR 0.60, 95% CI 0.48-0.75 and P<.001) holding age and educational track constant.

Higher problematic internet use was significantly associated with younger age (regression coefficient=−0.10, 95% CI −0.15 to −0.06; P<.001), whereas higher odds of tobacco use were associated with older age (OR 1.04, 95% CI 1.01-1.06; P=.006). Stress (regression coefficient=0.01, 95% CI −0.001 to 0.03; P=.06), social competencies (regression coefficient=0.05, 95% CI −0.005 to 0.10; P=.07), odds of cannabis use (OR 1.02, 95% CI 0.99-1.05; P=.18), and alcohol consumption (regression coefficient=0.01, 95% CI −0.004 to 0.03; P=.12) were not significantly associated with age.

Wald tests indicated that social competencies (P=.02), problematic internet use (P=.002), and the odds of tobacco use (P<.001) were significantly associated with educational track after adjusting for gender and age, whereas perceived stress (P=.26), odds of cannabis use (P=.68), and alcohol consumption (P=.54) did not differ significantly between educational tracks. Clerical support workers (regression coefficient=1.10, 95% CI 0.36-1.84; P=.004), service and sales workers (regression coefficient=0.79, 95% CI 0.05-1.52; P=.04), students in vocational grammar school (regression coefficient=0.89, 95% CI 0.27-1.52; P=.005), and students in vocational preparation (regression coefficient=1.02, 95% CI 0.26-1.79; P=.009) reported significantly higher social competencies than the reference group of professionals, technicians, and associate professionals. The other educational tracks did not differ significantly in their social competencies from each other (data not shown).

Problematic internet use was significantly lower among craft-related trade workers (regression coefficient=−0.76, 95% CI −1.35 to −0.17; P=.01), and the odds of tobacco use were significantly higher among service and sales workers (OR 1.61, 95% CI 1.10-2.37; P=.02) than in the reference group.

Comparisons between the other educational tracks revealed that service and sales workers (regression coefficient=−0.81, 95% CI −1.57 to −0.06; P=.03), as well as craft-related trade workers, plant and machine operators and assemblers, and elementary occupations (regression coefficient=−1.26, 95% CI −1.93 to −0.58; P<.001), reported significantly lower problematic internet use than clerical support workers, whereas students in vocational grammar school (regression coefficient=0.95, 95% CI 0.38-1.52; P=.001) and vocational preparation (regression coefficient=1.00, 95% CI 0.30-1.69; P=.005) reported significantly higher problematic internet use than craft-related trade workers, plant and machine operators and assemblers, and elementary occupations. Vocational grammar school students also had lower odds of tobacco use than service and sales workers (OR 0.46, 95% CI 0.31-0.68; P<.001); craft-related trade workers, plant and machine operators and assemblers, and elementary occupations (OR 0.52, 95% CI 0.37-0.74; P<.001); and students in vocational preparation (OR 0.64, 95% CI 0.43-0.95; P=.03). Finally, service and sales workers had higher odds of tobacco use than clerical support workers (OR 1.61, 95% CI 1.03-2.49; P=.04).

Figure 3 shows the percentage of study participants who received red, yellow, or green traffic light feedback for each behavior. Approximately two-thirds of the study participants received either red or yellow traffic light feedback for stress (1858/2568, 72.4%), social media and gaming (1828/2568, 71.2%), or alcohol (1672/2568, 65.1%). For tobacco, less than half (1133/2568, 44.1%) of the study participants received red or yellow traffic light feedback, whereas for cannabis (673/2568, 26.2%) and social competencies (778/2568, 30.3%), only approximately one-third received red or yellow traffic light feedback.

On average, students received red or yellow traffic light feedback for 3.09 (SD 1.42) of the 6 behaviors and red traffic light feedback only for 1.87 (SD 1.17) of the 6 behaviors. Only 2.7% (70/2568) of the study participants received neither red nor yellow traffic light feedback, and 11.1% (286/2568) received no red traffic light feedback for any of the 6 behaviors. The number of participants receiving red or yellow traffic light feedback for 1, 2, 3, 4, 5, and 6 behaviors were 11.1% (285/2568), 21.42% (550/2568), 24.8% (637/2568), 22.6% (580/2568), 13.6% (350/2568), and 3.7% (96/2568), respectively. The number of participants receiving red traffic light feedback for 1, 2, 3, 4, 5, and 6 behaviors were 30% (769/2568), 30.5% (783/2568), 19% (488/2568), 7.9% (204/2568), 1.4% (37/2568), and 0.04% (1/2568), respectively.

Among students who made a module choice, the most frequently chosen coaching topic was stress (818/1236, 66.2%), followed by social media and gaming (625/1236, 50.6%), alcohol (360/1236, 29.1%), social competencies (306/1236, 24.8%), tobacco (232/1236, 18.8%), and cannabis (131/1236, 10.6%). The frequency of module choice was higher among those students who had received red or yellow traffic light feedback for the corresponding behavior than among the total group (Table 1). This trend was highest for stress, with 80.3% (727/905) of students with yellow or red traffic light feedback and 87.7% (541/617) of students with red traffic light feedback only choosing the stress module. The lowest choice rate was found for alcohol consumption. Even within those with red traffic light feedback, only 46.6% (274/588) chose the alcohol module.

Table S6 in Multimedia Appendix 1 shows that, among those students with only 1 or 2 instances of yellow or red traffic light feedback, the choice rates for the recommended modules ranged from 68% (13/19) to 98% (40/41). For those with >2 instances of yellow or red traffic light feedback, the choice rates were much lower as students had to choose between several recommended modules. For example, choice rate for social media and gaming among those with >2 instances of yellow or red traffic light feedback (193/206, 93.7%) was approximately 30% lower than among those with 1 or 2 instances of yellow or red traffic light feedback (542/884, 61.3%). However, the number for some of the subsamples was rather small.

Most students chose modules that were congruent with their presented risks. For example, 88.9% (727/818) of the students who chose the stress module had yellow or red traffic light feedback in this behavior area. The congruence rates for the other behaviors were 61.4% (188/306) for social competencies, 86.6% (541/625) for social media and gaming, 95.6% (344/360) for alcohol, 97.4% (226/232) for tobacco, and 89.3% (117/131) for cannabis.

Table S7 in Multimedia Appendix 1 shows the association between gender, age, educational track, and prevention needs and module choice. The ICCs for the modules ranged from 2.1% to 11%, indicating a modest influence of class membership on module choice.

Women were significantly more likely to select the stress module than men (OR 2.38, 95% CI 1.69-3.33; P<.001), who were more likely to select social media and gaming (OR 0.52, 95% CI 0.40-0.69; P<.001), alcohol (OR 0.50, 95% CI 0.37-0.67; P<.001), and cannabis (OR 0.37, 95% CI 0.21-0.63; P<.001) than women when holding age, educational track, and prevention need for the corresponding behavior constant. Men and women did not differ significantly with respect to their frequency of module choice for the tobacco (OR 1.01, 95% CI 0.70-1.46; P=.97) and social competencies (OR 1.13, 95% CI 0.83-1.54; P=.44) modules. Younger students were significantly more likely to choose the cannabis module (OR 0.81, 95% CI 0.74-0.90; P<.001), whereas there were no significant differences in module choice by age for the other 5 modules.

Although Wald tests did not reach significance, there were individual educational tracks that differed in their odds of choosing the social competencies, social media and gaming, alcohol, and cannabis modules when compared to the reference category. Craft-related trade workers, plant and machine operators and assemblers, and elementary occupations had a significantly lower likelihood of choosing social media and gaming when compared to the reference group of professionals, technicians, and associate professionals (OR 0.61, 95% CI 0.39-0.97; P=.04), whereas the other tracks did not differ significantly from the reference group in their likelihood of choosing social media and gaming. Clerical support workers (OR 1.85, 95% CI 1.02-3.38; P=.04); service and sales workers (OR 1.77, 95% CI 1.00-3.15; P=.049); and craft-related trade workers, plant and machine operators and assemblers, and elementary occupations (OR 1.98, 95% CI 1.16-3.38; P=.01) were significantly more likely to choose alcohol compared to the reference group, whereas students in vocational grammar school (OR 1.23, 95% CI 0.73-2.09; P=.43) and vocational preparation (OR 1.57, 95% CI 0.86-2.87; P=.14) did not differ significantly from the reference group in their likelihood of choosing alcohol. Clerical support workers were most likely to select the cannabis module (OR 3.66, 95% CI 1.33-10.03; P=.01), and service and sales workers were least likely to select the social competencies module (OR 0.51, 95% CI 0.30-0.88; P=.02) when professionals, technicians, and associate professionals were used as the reference group.

Prevention needs for each behavior were significantly related to the corresponding module choice (eg, the higher the alcohol consumption, the more likely it was that the alcohol module would be chosen; OR 1.31, 95% CI 1.21-1.43; P<.001).

The prevalence of the occupational tracks based on the German Classification of Occupations is shown in Table S8 in Multimedia Appendix 1.

Wald tests indicate that only problematic internet use (P=.002) and the odds of tobacco use (P=.002) were significantly associated with educational track after adjusting for gender and age (Table S9 in Multimedia Appendix 1). All educational tracks showed numerically higher problematic internet use than the reference category of production and construction. However, these differences did not reach significance for the tracks of natural science, geography, and computer science (regression coefficient=0.44, 95% CI −0.43 to 1.31; P=.32) and trading and hotel and tourism (regression coefficient=0.65, 95% CI −0.04 to 1.34; P=.06).

The odds of tobacco use were significantly lower in natural science, geography, and computer science (OR 0.58, 95% CI 0.34-0.97; P=.04); organization and accounting and media (OR 0.59, 95% CI 0.38-0.92; P=.02); and vocational grammar school (OR 0.52, 95% CI 0.37-0.72; P<.001) than in the reference category of production and construction.

Comparisons between the other educational tracks revealed that the odds of tobacco use were lower in vocational grammar school than in traffic, logistics, and security and safety (OR 0.51, 95% CI 0.30-0.89; P=.02); trading and hotel and tourism (OR 0.51, 95% CI 0.34-0.76; P<.001); health, social affairs, teaching, and education (OR 0.65, 95% CI 0.44-0.97; P=.03); and vocational preparation (OR 0.64, 95% CI 0.43-0.95; P=.03). Similarly, the odds were also lower in organization and accounting and media than in trading and hotel and tourism (OR 0.58, 95% CI 0.35-0.95; P=.03).

Choice of the social competencies module could not be modeled using multilevel logistic regression as the inclusion of the independent variables led to nonconvergence of the model. Instead, multiple logistic regression was used. Wald tests indicated that only the odds of choosing the social competencies (P<.001) and alcohol (P=.01) modules were significantly associated with educational track after adjusting for gender, age, and the corresponding prevention needs (Table S10 in Multimedia Appendix 1). The social competencies module was chosen significantly more often by students from the natural science, geography, and computer science tracks (OR 3.40, 95% CI 1.71-6.73; P<.001) and significantly less often by students from the traffic, logistics, and security and safety tracks (OR 0.32, 95% CI 0.11-0.99; P=.048) compared to the reference group. Comparisons between the other educational tracks revealed that all educational tracks were significantly less likely to choose the social competencies module compared to students in natural science, geography, and computer science. Students in organization, accounting and media, and vocational grammar school were significantly more likely to choose the social competencies module than students in traffic, logistics, and security and safety, as well as students in trading and hotel and tourism, whereas the other educational tracks did not differ significantly from each other in their odds of choosing the social competencies module (data not shown).

The alcohol module was selected significantly less often by students from the natural science, geography, and computer science tracks (OR 0.22, 95% CI 0.09-0.59; P=.002) compared to the reference group. The other educational tracks did not differ significantly from each other in their odds of choosing the alcohol module.

Although Wald tests did not reach significance, there were individual educational tracks that differed in their odds of choosing the stress, social media and gaming, and cannabis modules when compared to the reference category. Stress was chosen significantly more often by students from the health, social affairs, teaching, and education tracks (OR 2.10, 95% CI 1.11-3.98; P=.02), and social media and gaming was chosen significantly more often by students from the organization and accounting and media tracks (OR 2.08, 95% CI 1.16-3.72; P=.01). The cannabis module was chosen significantly more often by students from the traffic, logistics, and security and safety tracks (OR 4.41, 95% CI 1.54-11.13; P=.005). The results for the other potential determinants in the multiple regression models are consistent with those of the main analysis.

This study examined the prevention needs of German vocational school students and their module choices within the app-based addiction prevention intervention ready4life. Key findings included, first, that prevention needs among vocational school students were high. Almost 80% reported 2 or more risks according to yellow or red traffic light feedback. The highest prevention needs were identified for stress, social media and gaming, and alcohol. Second, in line with this, the most frequently chosen modules were stress and social media and gaming. Third, alignment of module choice with the risk feedback received was high, especially for those with 1 or 2 risks. Fourth, gender, age, educational background, and the corresponding prevention needs were associated with module choice.

The strengths of this study include (1) the focus on vocational school students, an underserved group in mobile intervention research; (2) the use of a novel multibehavioral approach that combined life skill training with addressing multiple addictive behaviors; (3) the use of a facilitated access method that ensured a more representative sample compared to relying on online or media recruitment [40]; and (4) being the first study to statistically analyze the determinants of module choice among vocational school students.

The following limitations of this study should be considered when interpreting the results. First, the data may not be fully representative of German vocational school students due to selection bias (eg, additional analyses showed that men and students from nonprofessional backgrounds were less likely to participate in this study). Furthermore, not all educational tracks were represented in the sample (eg, the agriculture, forestry, or military tracks were not included). Second, data collection during the COVID-19 pandemic may have influenced participation willingness, addictive behaviors, or module selection. For example, gaming and social media use were elevated during the pandemic in adolescents [71]. Third, all assessments were based on self-report measures. Thus, there is the potential for over- or underreporting, which could have affected the accuracy of the risk and competence feedback. Fourth, this study focused on module choice rather than module completion or frequency of app use. However, completion rates are likely to differ between modules [15,31]. Fifth, we cannot exclude the possibility that the high interest in the stress module was partly triggered by the classes’ introductions, which were largely focused on stress. Finally, the number of potential determinants assessed was limited. Other variables such as students’ personal values and goals, subjective importance of health, perceptions of social norms, or migration background could also be related to their prevention needs or module choices.

Our study supports the existing evidence that vocational school students have high prevention needs in relation to addictive behaviors. An important finding was that the students’ module choices were highly congruent with their identified prevention needs. This is important because intervention effects are likely to be higher when modules that are more relevant to the participant are chosen (eg, when smokers choose to work on the smoking module [60]). Our findings also highlight that incorporating self-tailoring options into mobile health interventions works well to address multiple risks and different combinations of risks for vocational school students. This is particularly important given that a high proportion of vocational school students engage in multiple addictive behaviors and may not be able to address all their issues at once. In terms of educational track, our mobile intervention approach was able to achieve a positive equity impact (ie, those in tracks typically associated with a lower socioeconomic status were more likely to choose modules related to alcohol and cannabis). The identified module preferences may inform the further development of mobile interventions to improve adherence and effectiveness within specific subgroups. To promote more gender-equitable engagement, mobile interventions need to address disparities in module selection. Future interventions for vocational school students could tailor module content or descriptions to gender to ensure that both genders engage with modules that meet their prevention needs. For example, framing stress management around work performance for men or linking substance use prevention to emotional well-being for women may increase engagement. Alternatively, embedding cross-links within modules, such as mentioning substance use in life skill modules, could promote more holistic behavior change across diverse groups. Future research should also investigate whether the choice of a particular module influences change in the behaviors addressed in other unchosen modules (eg, the effect of choosing the stress module on addictive behaviors).