The SD-R framework, derived from the well-established JD-R model in the work context (Demerouti et al., 2001; Bakker and Demerouti, 2014), includes two central paths. The health-impairment path posits that study demands are positively related to burnout, while the motivational path posits that study resources are positively related to engagement and negatively related to burnout (Lesener et al., 2020). The validity of these two paths is supported by broad empirical evidence from cross-sectional (e.g., Gusy et al., 2016), longitudinal (e.g., Gusy et al., 2021), and diary studies (e.g., Körner et al., 2021).

Study demands occur at an organizational, physical, social, or psychological level. They require high levels of physical or mental effort, and are therefore associated with physiological or psychological costs (Lesener et al., 2020). Overload and time pressure are among the most common study demands and key predictors of burnout within the SD-R framework (e.g., Lesener et al., 2020; Gusy et al., 2021). Consequently, our study examined the psychological demands of studying. This study demand encompasses aspects such as time pressure, haste, or competing tasks (Schmidt et al., 2019).

Study resources also occur at organizational, physical, social, or psychological levels and can help achieve goals, promote personal development, and reduce study demands (Lesener et al., 2020). An important study resource, and a strong predictor of engagement within the SD-R framework even during the COVID-19 pandemic, is social support from lecturers (Salmela-Aro et al., 2022). In line, Reichel et al. (2023) assume that social support from lecturers plays an especially important role in times of numerous stressors, such as during a pandemic. We therefore examined this study resource in our study. Social support from lecturers includes the extent to which lecturers/professors take an interest, help students in their studies and support them through good organization.

The JD-R model has been extended to incorporate personal resources alongside job resources (Bakker and Demerouti, 2017). Personal resources are self-aspects associated with resilience and a sense of being able to successfully control and influence the environment, even in difficult situations (Hobfoll et al., 2003). These resources help to achieve goals, grow personally, and protect against threats (Xanthopoulou et al., 2009). A key personal resource receiving increased attention in recent years is mindfulness, defined as the “enhanced attention to and awareness of current experience or present reality” (Brown and Ryan, 2003, p. 822). Grover et al. (2017) integrated mindfulness into the JD-R model, highlighting its relevance as a personal resource within this model. In the academic context, a positive relationship between mindfulness and engagement as well as a negative relationship between mindfulness and burnout was found (Robins et al., 2015), demonstrating its impact similar to that of study resources within the SD-R framework. Consequently, our study examined the personal resource of mindfulness.

According to the SD-R framework, engagement is fostered by high levels of study and personal resources (Ouweneel et al., 2011; Lesener et al., 2020). Engagement is a fulfilling, positive state that encompasses the three dimensions of vigor, dedication, and absorption. Vigor includes high levels of energy and perseverance and a willingness to try hard even when difficulties arise. Dedication includes feelings of enthusiasm, inspiration, and pride. Absorption is a state of concentration and flow (Schaufeli et al., 2002). In the academic context, engagement is of great importance due to its positive relationship with academic performance (Salanova et al., 2010) and life satisfaction (Lesener et al., 2020).

Conversely, the SD-R framework posits that burnout results from high study demands and a lack of study resources (Lesener et al., 2020). Burnout encompasses the three dimensions of exhaustion, cynicism, and professional inefficacy. Exhaustion describes a feeling of fatigue due to high study demands and represents the core dimension of burnout. Cynicism refers to a detached attitude toward one’s studies, and professional inefficacy refers to a feeling of incompetence as a student (Schaufeli et al., 2002).

The JD-R model has been further expanded to incorporate two behavioral variables: job crafting within the motivational path (Tims and Bakker, 2010) and self-undermining within the health-impairment path (Bakker and Wang, 2020). Recently, the concept of job crafting has been adapted to the academic context as study crafting and incorporated into the SD-R framework (Körner et al., 2021). Study crafting refers to the proactive adjustments students make to their study environment, aligning their studies with their personal skills and preferences (Körner et al., 2021). This concept draws from the job crafting strategies distinguished by Tims et al. (2013), and includes four analogous study crafting strategies: Increasing structural resources, increasing social resources, increasing challenging demands, and decreasing hindering demands. Increasing structural resources involves activities that contribute to personal growth, such as skill development. Increasing social resources includes behaviors such as seeking feedback or advice. Increasing challenging demands involves activities such as taking on additional projects or attending extra lectures. Decreasing hindering demands involves, for example, trying to make study less demanding (Tims et al., 2012; Körner et al., 2021).

On the other hand, the concept of self-undermining, defined as “behavior that creates obstacles that may undermine performance” (Bakker and Costa, 2014, p. 115), has been included in the JD-R model. Self-undermining involves behaviors such as making mistakes or provoking conflicts (Bakker and Wang, 2020). This concept has received little attention within the academic context so far. However, an initial study investigated self-undermining within the SD-R framework and confirmed that the health-impairment path can be extended to include this concept (Körner et al., 2021).

Recent versions of the JD-R model specify a gain cycle within the motivational path and a loss cycle within the health-impairment path, further clarifying the interplay among JD-R variables (Bakker et al., 2023a). The model assumes that engaged employees want to maintain their engagement and therefore attempt to build new resources through job crafting. The resources built through job crafting, in turn, foster engagement, resulting in a gain cycle with reciprocal relationships between resources, engagement, and job crafting (Bakker et al., 2023a). Empirical evidence supports this assumption. Systematic reviews and meta-analyses in the work context have affirmed positive relationships between job crafting and various job resources (e.g., autonomy, social support), between job crafting and engagement, as well as between job resources and engagement (Rudolph et al., 2017; Lesener et al., 2019; Lichtenthaler and Fischbach, 2019; Zhang and Parker, 2019). In the academic context, similar positive correlations have been observed between study and personal resources and engagement (Ouweneel et al., 2011; Lesener et al., 2020). Furthermore, a weekly diary study found a positive relationship between study resources and study crafting, mediated by engagement (Körner et al., 2021).

Conversely, the JD-R model suggests that exhausted employees tend to engage in self-undermining. Through self-undermining, they create new demands and obstacles, which further increase exhaustion, resulting in a loss cycle (Bakker et al., 2023a). This assumption is empirically supported by research indicating a positive reciprocal relationship between job demands and burnout and a positive relationship between exhaustion and self-undermining as well as between job demands and self-undermining (Lesener et al., 2019; Rațiu and Dobre, 2020; Bakker et al., 2023b). Within the academic context, the concept of self-undermining is relatively underexplored. However, preliminary research indicated a positive relationship between study demands and self-undermining via exhaustion at the weekly within-person level (Körner et al., 2021).

In the work context, the JD-R model has been effectively utilized as a theoretical foundation for interventions aimed at enhancing employee well-being (Bakker and Demerouti, 2017; Bakker et al., 2023a). This model provides various starting points for such interventions, including optimizing job demands, increasing job and personal resources, and promoting job crafting (Bakker and Demerouti, 2014; Bakker et al., 2023a). Research has validated the effectiveness of these interventions, demonstrating significant increases in personal resources, engagement, and job crafting, as well as reductions in exhaustion (Bakker, 2017; van Wingerden et al., 2017; Gordon et al., 2018; Bakker and van Wingerden, 2021). To our knowledge, only one intervention has employed the SD-R framework as a theoretical foundation for an intervention within the academic context so far. This intervention, a study crafting intervention, was adapted from job crafting interventions in the work context. It successfully increased study crafting while concurrently fostering engagement and reducing exhaustion (Körner et al., 2022).

Our present study expands this research by again utilizing the SD-R framework for developing interventions, but with different starting points within the SD-R framework compared to the previous study crafting intervention. Whereas the study crafting intervention primarily focused on enhancing study crafting, our current interventions aim to balance demands and resources. Besides the impact on demands and resources, we also investigate the influence of the interventions on the other variables of the SD-R framework. Drawing from the theoretical assumptions and empirical findings from both the JD-R model and the SD-R framework, and building on results from job crafting and study crafting interventions, we hypothesize:

We also explore the underlying mechanisms that influence the outcomes of our intervention. Research into these mediating processes of interventions based on the JD-R model or the SD-R framework remains relatively scarce. However, in the work context, there is evidence suggesting that job crafting serves as a mediator in the relationship between job crafting interventions and JD-R outcomes like engagement (Mukherjee and Dhar, 2023). Similarly, an initial study in the academic context has found comparable results, with study crafting acting as a mediator in the effects of a study crafting intervention on engagement and exhaustion (Körner et al., 2022). Our SD-R-based interventions are targeted at balancing demands and resources. Therefore, we assume that the interventions will primarily influence these variables, which in turn, will affect other variables within the SD-R framework. Thus, we hypothesize:

The study was conducted at Aalen University. All students received an email invitation to participate in the study, we presented the study in selected lectures, promoted it at a trade fair, and through social media. Inclusion criteria were a minimum age of 18 years and enrollment at Aalen University. The exclusion criterion was a diagnosed mental disorder. After registration, students received detailed participant information. The study was conducted in accordance with the ethical guidelines of the American Psychological Association: Participation was voluntary and could be terminated at any time without giving a reason. There was no monetary compensation, but students received credit for their participation and got access to all modules of the online intervention upon completion of the study. Informed consent was obtained from all participants before the study began.

The study was a randomized controlled trial with two IGs and one WLC. Students were randomly assigned to one of the three groups after enrollment. In the week prior to the start of the intervention, students received information about their group assignment, access information to the online platform, and the link to the T1 questionnaire by e-mail. In the first intervention week, the Introductory Module and Module 1 (“Understanding Stress”) were unlocked for both IGs. In the second intervention week, IG1 received access to Module 2 (“Recognizing Stressors”) and IG2 received access to Module 3 (“Awakening Resources”). During the third week, students were asked to apply what they had learned in the two modules to their daily routines. Students then completed the T2 questionnaire and 5 weeks later the T3 questionnaire. It is important to note that due to the COVID-19 pandemic and the increasing number of infections, most of the lectures changed from face-to-face to online from T1 to T2. The T3 questionnaire was completed 2 weeks prior to the start of the exam period and lectures continued to be delivered online. Figure 1 shows the study design.

We conducted an a priori power analysis for a repeated measures analysis of variance in a 3×3 factorial design to estimate the required sample size. We assumed a small effect size of f = 0.1, based on findings of previous job crafting interventions, and targeted a power of 0.80. This resulted in a sample size of 204. A total of 253 students enrolled in the study of whom 242 completed the T1 questionnaire (95.7% response rate), 220 completed the T2 questionnaire (9.1% dropout from T1 to T2), and 208 completed the T3 questionnaire (5.5% dropout from T2 to T3). The overall dropout rate was 18.7%, which is relatively low compared to other online interventions, which report adherence rates of about 50% (Kelders et al., 2012) and attrition rates of 40–50% (Bennett and Glasgow, 2009; Kuster et al., 2017). Previous studies have pinpointed various factors contributing to dropout in online interventions, such as dwindling interest over time (Bennett and Glasgow, 2009), the burden of high time demands, and uncertainties related to the intervention’s content and instructions (Fasthoff et al., 2023). Given that our intervention took place just before the Christmas period and exam season, it is plausible to suggest that the observed dropout could especially be linked to heightened demands and the resultant time constraints.

Our final sample consisted of 205 students, distributed across the two IGs and the WLC, who completed the three questionnaires. Of these, 141 were female, 62 were male, 1 was diverse, and 1 did not provide gender information. The participants studied in five faculties (chemistry: n = 12, electronics and informatics: n = 12, mechanical and materials engineering: n = 19, optics and mechatronics: n = 30, economics: n = 132). They were on average 21.82 years old (SD = 3.08) and the majority of participants (n = 199) were in a bachelor’s degree program (master’s degree program: n = 6). On average, the students were in their third study semester (M = 3.25, SD = 2.61). IG1 consisted of 64 students. Of these, 44 were female and 20 were male with a mean age of 21.94 years (SD = 3.57). IG2 consisted of 70 students. Of these, 47 were female and 21 male (diverse: n = 1, not specified: n = 1) with a mean age of 21.50 years (SD = 2.60). The WLC consisted of 71 students. Of these, 50 were female and 21 were male with a mean age of 22.04 years (SD = 3.07). Figure 2 shows the study procedure.

Our study utilized two interventions based on the “Einfach weniger Stress [Simply less stress]” (EWS) concept (Paulsen and Kortsch, 2020), which has been tested and certified by the Zentrale Prüfstelle Prävention [Central Examination Office for Prevention] (ZPP; www.zentrale-pruefstelle-praevention.de).¹ The concept has already been successfully applied in the work context (Fasthoff et al., 2023). Initially designed as a face-to-face training, EWS was adapted into an online format to enhance accessibility and increase its reach. The web-based version of the intervention was implemented as a standalone Wordpress-based solution. The EWS intervention includes five modules: (1) understanding Stress, (2) recognizing stressors, (3) awakening resources, (4) planning implementation, (5) acting calmly. The modules build on each other but are carefully crafted to be self-sufficient, offering a complete package of content. Every module starts with clearly defined learning objectives and concludes with a comprehensive review, coupled with an observation task designed to integrate the learned content into daily life. In each module, exercises address at least one Behavior Change Technique, as outlined by Michie et al. (2015) from one of four main categories: goals and planning (main category 1), observing behavior and giving feedback (main category 2), building knowledge (main category 4) or repeating and generalizing (main category 8). This thoughtful design ensures that even standalone modules can deliver significant intervention effects.

To address students’ preference for concise interventions that seamlessly fit into their academic routines, we have deliberately chosen to limit the selection to two modules per group, as supported by the findings of Seidl et al. (2018). Our selection prioritized modules 1–3, which are fundamentally based on the SD-R framework. This choice was made to ensure that our interventions are in line with the aims of SD-R-based strategies, following the guidance provided by research on JD-R-based interventions (Bakker et al., 2014, 2023a).

Module 1 provides an in-depth explanation of the SD-R framework, highlighting the interplay between demands and resources. Module 2 encourages participants to identify and reflect on their demands, while Module 3 focuses on recognizing, enhancing, and activating their resources. Both IGs completed Module 1, providing all participants with foundational knowledge about the SD-R framework. IG1 then primarily focused on adapting demands, while IG2 then emphasized increasing resources. This design enabled us to compare the effectiveness of these two intervention approaches.

Each module consisted of theoretical input, exercises, reflection activities, and transfer tasks. Participants could download and print worksheets or complete them digitally. Two fictional characters guided participants through the online course. An additional third fictional student character was created especially for this study. Participants were allocated 1 week to complete each module, with the option to stop at any time and continue later. The completion time of module 1 was approximately 90 min, and that of module 2 and 3 were approximately 110 min each. Table 1 provides an overview of the intervention modules.

The questionnaires were identical at the three measurement time points. In the T1 questionnaire, we also collected sociodemographic data.

Social support from lecturers was measured with five items (e.g., “My lecturers/professors support me through good organization.”) and psychological demands were measured with seven items (e.g., “My studies are hectic.”) of the questionnaire on Structural Study Conditions (Schmidt et al., 2019). Items were rated on a 4-point Likert-type scale (1 = does not apply, 4 = does apply).

Mindfulness was measured using the short version of the Freiburg Mindfulness Questionnaire (Walach et al., 2004). The questionnaire contains 14 items (e.g., “I am in touch with my experiences, here and now.”), which were rated on a 4-point Likert-type scale (1 = almost never, 4 = almost always).

Study crafting was measured with a shortened version of the German Job Crafting Scale (Lichtenthaler and Fischbach, 2016) adapted to the academic context. The sub dimensions increasing structural resources (e.g., “I try to develop my capabilities.”), increasing social resources (e.g., “I ask others for feedback on my performance in my studies.”), increasing challenging demands (e.g., “When there is not much to do in my studies, I see it as a chance to start new projects.”), and decreasing hindering demands (e.g., “I make sure that my studies are mentally less intense.”) were measured with four items each. Items were rated on a 5-point Likert-type scale (1 = not at all true, 5 = completely true).

Self-undermining was measured using the self-undermining scale (Bakker and Wang, 2020). We adapted the six items (e.g., “I make mistakes.”) to the academic context and participants rated them on a 7-point Likert-type scale (1 = never, 7 = always).

Engagement was measured using the Utrecht Work Engagement Scale – Student Form, which includes three items each for vigor, dedication, and absorption (Schaufeli et al., 2002). Items (e.g., “I am immersed in my studies.”) were rated on a 7-point Likert-type scale (1 = never, 7 = always).

Exhaustion was measured using the short German Maslach Burnout Inventory – Student Survey (Wörfel et al., 2015). The scale consists of three items (e.g., “I feel drained by my studies.”) which were rated on a 7-point Likert-type scale (1 = never, 7 = always).

Data were analyzed using IBM SPSS 28 and R 4.1.0 with package welchADF (Villacorta, 2017). First, we used chi-square tests and t-tests to test for sociodemographic differences among the three groups and multivariate analysis of variance (ANOVA) to test for significant T1 differences in all research variables. We tested hypotheses 1–4 with 3×3 repeated measures (RM) ANOVA. Measurement time (T1, T2, and T3) was the within-subject factor and group (IG1, IG2, and WLC) was the between-subject factor. The requirements for RM ANOVA are homogeneity of error variances (tested with Levene’s test) and sphericity (tested with Mauchly’s test). If the sphericity assumption was violated, the Greenhouse–Geisser correction (for ε < 0.75) or the Huynh-Feldt correction (for ε > 0.75) was used. If there was a significant time x group interaction effect, the time effect was examined separately for each group. If there was a significant time effect, the differences between each of the two measurement time points (T1–T2, T1–T3, and T2–T3) were further examined. If there was no significant time × group interaction effect, the main time effect (also separately for the three groups) was examined. Hypotheses 5 and 6 were tested using Model 4 of the Hayes Process macro for SPSS (Hayes, 2017). Group membership was included as the independent variable. Demands and resources at T2 were each included as the mediator. Engagement, exhaustion, study crafting, or self-undermining at T3 were each included as the dependent variable. T1 scores of the mediator and dependent variable were included as control variables. Bootstrapping with 5,000 samples was used to calculate 95% confidence intervals.

Our study contributes to the literature in three significant ways: First, it extends the literature on the SD-R framework by demonstrating that SD-R-based interventions can actively influence the postulated paths. Since we confirmed intervention effects on mindfulness, study crafting, and self-undermining in addition to exhaustion, our study also contributes to the validation of an extended SD-R framework. Additionally, our mediation analyses underscore the importance of personal resources within the SD-R framework, highlighting their role in influencing both behavioral (i.e., study crafting, self-undermining) and well-being outcomes (i.e., engagement).

Second, our study contributes to the literature on interventions in higher education settings. While existing reviews indicate heterogeneous effects of online stress management or mindfulness interventions (e.g., Harrer et al., 2019; Dawson et al., 2020), our findings indicate that the SD-R framework represents a novel effective intervention approach. Thereby, our study also addresses the critical need for further evidence-based support programs for students (Mülder et al., 2022).

Third, our research adds to the understanding of proactive and dysfunctional student behavior. A study crafting intervention only increased the study crafting strategy of decreasing hindering demands (Körner et al., 2022), while our intervention had an impact on a broader range of study crafting strategies (i.e., increasing structural resources and increasing challenging demands). This suggests that the intervention focus on balancing resources and demands can achieve different effects than a study crafting intervention. Moreover, our study is pioneering in demonstrating that interventions can positively affect self-undermining, which, to our knowledge, has not been previously investigated in an intervention in either the academic context or work context.

Although our study was carefully planned and conducted, it has some limitations that should be considered in future research: First, we focused on a limited set of study demands, study resources, and personal resources, with some showing no significant intervention effects. Future studies could explore the impact of SD-R-based interventions on a broader range of study resources (e.g., qualification potential) and study demands (e.g., incompatibility of study and private life) (Gusy et al., 2016). Similarly, the impact of these interventions on other outcome variables that have already been examined in cross-sectional studies in the context of the SD-R framework such as life satisfaction or performance could be examined (Schaufeli et al., 2002; Lesener et al., 2020).

Second, currently there are no validated scales for assessing study crafting and self-undermining among students. Therefore, scales developed and validated for the work context were adapted to the higher education context. Although most scales exhibited satisfactory reliability, the results concerning study crafting and self-undermining should be interpreted with caution. Therefore, future research should prioritize the development and validation of such measurement instruments.

Third, the changing study conditions during the course of our study (online vs. face-to-face lectures, exam phase at follow-up) may also have influenced our results, as discussed earlier. Future research should consider varying time points and examine the longer-term impact of SD-R-based interventions (i.e., one semester later) as some effects may only emerge over time, as evidenced in a job crafting intervention study in the work context (van Wingerden et al., 2017b).

Fourth, our study’s generalizability is limited by the specific demographic composition of the sample. The sample consisted predominantly of women and bachelor’s students at a single university of applied sciences. At the same time, Herbst et al. (2016) emphasize that women show higher stress levels compared to men, students at universities of applied sciences compared to students at universities, and bachelor’s students compared to master’s students. Thus, our intervention targeted a particularly vulnerable population, which has a high practical value for this target group, but further limits its transferability of the results. Future studies could address these points and also investigate whether sociodemographic variables or personality traits (e.g., regulatory focus) influence how students respond to the resources-intervention compared to the demands-intervention.

Last, self-report data, while appropriate for capturing subjective well-being and personal perceptions regarding the own study environment, is susceptible to common method bias (Podsakoff et al., 2003). Future studies might incorporate peer or lecturer ratings for a more comprehensive assessment of certain variables.

Our study demonstrated that SD-R-based interventions are effective in enhancing mindfulness and study crafting behavior, and mitigating self-undermining and exhaustion among students. Notably, our study underscores that brief interventions, requiring only 3–4 h over 2 weeks, can yield positive outcomes, which also meets students’ preference for compact stress management solutions and online interventions (Seidl et al., 2018; Harrer et al., 2019). Therefore, a key practical implication of our study is the great potential of our interventions for higher education institutions as a cost-effective, and time- and resource-efficient way to support students. A notable observation is that the intervention group focusing on resources (IG2) exhibited more positive intervention effects. This suggests that interventions emphasizing resource enhancement are particularly impactful.

However, beyond implementing targeted interventions, it is crucial for higher education institutions to create a study environment that facilitates the adjustment of resources and demands. This involves creating opportunities for personal growth and autonomy, such as diverse extracurricular activities or flexible course choices. In addition, lecturers should pay attention to fostering social resources such as communication, support, feedback, and interaction. To counteract accumulation of demands and associated stress, higher education institutions should also ensure a distribution of academic demands across the semester, for example, by staggering exam schedules and assignment deadlines.

In conclusion, it should be noted that demands and resources can fluctuate greatly over the course of a semester, and study conditions can change rapidly due to external circumstances such as the COVID-19 pandemic (Körner et al., 2021). The SD-R framework provides an adaptable and responsive basis for interventions that can help customize demands and resources to respond flexibly to changing circumstances. While our intervention warrants further refinement and testing, we conclude that SD-R-based interventions offer a promising measure for improving student well-being by creating an optimal balance of demands and resources.