We conducted a randomized controlled trial with 93 student participants who had reported elevated levels of stress (score of ≥13 on the Perceived Stress Scale [PSS] [22,23]). Because of an expected dropout rate of about 50% [24], we recruited 183 participants who were randomly assigned to a digital autobiographical self-efficacy memory training and EMA (TG) or EMA only (control group [CG]). Independent researchers conducted random allocations. Which group participants were assigned to was blinded. During study participation, we excluded participants who did not fulfill inclusion criteria anymore at study start, who did not complete the baseline or postassessment, or who did not sufficiently participate in the app, which was defined as responding to at least 25% of the prompts. The final sample consisted of 93 participants (n=54 in the TG and n=39 in the CG; Figure 1). For a detailed description of the methods, please see our prior publication [21].

Participants were recruited from universities in Switzerland using mailing lists and postings on university websites and social media. Participants received detailed information about the study, its procedures, duration, potential risks, and benefits, but no specific information on the different groups. For example, they were informed about a 1-week smartphone app that assessed thoughts and emotions daily, which applied to both groups and secured blinding. All participants provided informed consent. They were offered monetary compensation, depending on compliance or course credits.

Interested students were asked to complete a web-based screening questionnaire assessing basic demographic information and their current stress level using the PSS. Inclusion criteria were (1) being a student at a Swiss university, (2) being 18-29 years old, (3) being a smartphone owner, (4) having fluent German language skills, and (5) having a score of ≥13 on the PSS. Exclusion criteria were a current psychiatric disorder.

In this study, we used 2 apps. One providing EMA and EMI (TG) and one providing EMA only (CG). Consecutively, both groups received EMA [25] 10 times per day, of which 3 were combined with the self-efficacy training in the TG. Besides, the TG received 8 additional EMA questions on specific self-efficacy every evening. EMA was adapted from Marciniak and colleagues [26]. As described in various studies using EMI and EMA, we also allowed participants to self-trigger interventions and assessments [4]. EMA questions assessed current mood, social and web-based contacts in the current situation, and specific self-efficacy (in TG only).

In addition to the scheduled prompts, participants could trigger additional EMA and EMI (TG), respectively, and EMA (CG) at any time. Both apps provided general information about study procedures and app use, and the TG app also offered a self-efficacy review section. EMI and the EMA were administered through the SEMA³ (Melbourne eResearch Group) platform, an open-source software application for delivering advanced smartphone surveys [27]. Participants were able to download the app from the app store or by using a link they received via email.

Before the app participation, participants in the TG watched video-based psychoeducation on self-efficacy [28] and were instructed to define 2 autobiographical self-efficacy memories. Specifically, they were asked about difficult life episodes they have overcome. The self-efficacy training consisted of instructions to slow down breathing and to perform an imagination task, specifically to recall their autobiographical memories step-by-step. For example, participants were instructed to initially build up contextual factors and, ultimately, to focus on the skills and character traits relevant to success [25]. After every training, we asked for the participant’s feedback on how well it went.

The trial obtained ethical approval from the Ethics Committee of the Faculty of Arts of the University of Zurich (20.4.24). Since this is not a clinical trial (ie, only healthy participants were involved), preregistration is not mandatory. The trial was registered retrospectively (ClinicalTrials.gov NCT05617248). The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and the Helsinki Declaration, as revised in 2013.

The total sample consisted of 93 (54 in the TG and 39 in the CG) participants. The mean age was 23.72 (SE 0.44) years in the TG and 22.64 (SE 0.61) years in the CG (group difference: P=.14). There were 42 female participants in the TG and 31 female participants in the CG (group difference: P=.84). Table 1 summarizes clinical variables. There were no statistically significant differences between groups in demographic and baseline psychological variables. Participants reported no harm or unintended effects.

Within participants in the TG, positive mood during the day was associated with higher specific self-efficacy in the evening (b=0.15, SE 0.05, P=.005; Figure 2). Neither negative mood during the day, dissatisfaction with social contacts, nor specific self-efficacy on the previous day had any significant effect on specific self-efficacy.

Between participants, more severe baseline depressive symptoms (BDI-II) were associated with lower specific self-efficacy (mean score over 1 week) during the self-efficacy training (b=–0.92, SE 0.35, P=.004, Figure 3). Other baseline variables showed no significant effect on specific self-efficacy. For detailed results, see Table 2.

Within participants, we found autoregressive effects for positive (b=0.34, SE 0.04, P<.001) and negative (b=0.48, SE 0.05, P<.001) mood in both groups. The effect of a positive mood on the following day’s mood was independent of group allocation. In contrast, the effect of negative mood was reduced in the TG (b=–0.13, SE 0.05, P=.01). Dissatisfaction with social contacts was associated with both a decreased positive mood (b=–0.56, SE 0.15, P<.001) and an increased negative mood (b=0.45, SE 0.12, P<.001) in both groups, with no significant group difference.

Between participants, trait anxiety at baseline (STAI, trait) was associated with lower positive mood (b=–0.64, SE 0.28, P=.02), while positive affect at baseline was associated with increased positive mood during study participation (b=0.34, SE 0.12, P=.004). More depressive symptoms at baseline (BDI-II) were associated with reduced negative mood (b=–1.05, SE 0.47, P=.02), while high state and trait anxiety (STAI, state and trait) were associated with increased negative mood (state anxiety: b=0.78, SE 0.38, P=.04; trait anxiety: b=0.73, SE 0.33, P=.03). Higher self-efficacy at baseline (GSE) was associated with reduced negative mood during study participation (b=–0.61, SE 0.30, P=.04); we observed no effect on positive mood. The other variables, including group membership, showed no significant effect on mood. For detailed results, see Tables 3 and 4.

This study applied an EMI providing a 1-week digital self-efficacy training to university students reporting elevated stress levels. We analyzed the daily collected EMA data and investigated the prediction of specific self-efficacy and how mood and other momentary parameters interacted with each other and with baseline variables. Positive training effects have been reported elsewhere [21].

We found several associations between mood and self-efficacy in university students with elevated stress levels: (1) a positive association between momentary positive mood and specific self-efficacy in the evening of the same day, (2) an association between higher baseline self-efficacy and lower mean negative mood during study participation, and (3) an association between higher baseline depressiveness and lower mean specific self-efficacy during study participation. Additionally, we found associations between anxiety and mood: (1) higher baseline trait and state anxiety were associated with a higher mean negative mood during study participation, and (2) higher baseline trait anxiety was associated with a lower mean positive mood during study participation. Moreover, we found autoregressive mood effects in both groups, but the autoregressive effect of negative mood was reduced in the TG.

Our results confirm previous studies that show an association between positive mood and self-efficacy [6,19]. The relation between high baseline depressiveness and low specific self-efficacy during study participation may be due to the negative association between depression and motivation, interests, and cognitive performance [44-46]. This finding also matches previous results suggesting that negative mood in depressed people is associated with a lower capacity to learn from new experiences [47], which may impede engagement in and understanding of the self-efficacy training. Interestingly, it has been shown that specific self-efficacy moderates the relationship between mood and cognitive performance, as it positively affects cognitive performance, but only during a positive mood [48].

The relationship between anxiety and mood is in line with literature widely investigating and discussing that depression and anxiety share some aspects and are correlated [49]. Within participants, positive mood and negative mood were self-reinforcing, a common effect of inertia, and in line with the assumption that such effects are implicated specifically in individuals with psychological maladjustment, low self-esteem, depression, stress, and rumination [50-53]. Interestingly, the self-reinforcing effect of positive mood was independent of group membership, while the self-reinforcing effect of negative mood was alleviated by self-efficacy training.

The training may enhance mood and enable participants to manage difficulties better when faced with challenging situations. The training may also have contributed to higher emotional flexibility [50], which is particularly important for adaptation and coping in constantly changing environments [54,55] and is associated with well-being, resilience [56], emotional functioning, and quality of life [57]. Interestingly and complementary to our results, a recent pilot study showed improvements in general self-efficacy in a group that had received emotional flexibility training [58].

Specific self-efficacy correlates with self-efficacy [59,60]. Previous research has shown that the specificity of self-efficacy plays an important role within a domain and even within a specific task [61,62]. Future investigations could go deeper and, for instance, assess specific self-efficacy concerning the content of the self-efficacy memories and, if applied in different settings, such as before or in addition to specific therapies, participants’ therapy motivation, therapeutic relation, and other parameters related to the specific situation, since self-efficacy has been associated with improved therapy motivation [63], treatment adherence, and outcomes [64,65].

This study is not without limitations. First, the part of our analyses this study focuses on was exploratory. Aiming to gain more knowledge on specific self-efficacy and a more nuanced picture of the potential effects of the self-efficacy training, we examined the prediction of specific self-efficacy in the TG and associations of mood with other EMA and baseline parameters in addition to our main research questions focusing on the effects of the training, which we described in another publication [21]. Such an approach amplifies the probability of false-positive findings.

The study was conducted during the second wave of the COVID-19 pandemic, and we included participants with elevated stress levels detected by using the PSS, which measures the degree to which life has been perceived as unpredictable, uncontrollable, and overwhelming [22,23]. Thus, we intended to include a subpopulation that is at higher risk since psychological stress is, for example, associated with negative consequences for academic performance and with mental disorders in students [66-68]. However, we did not investigate if or to what extent the elevated stress levels were related to the pandemic. Due to the pandemic, college students may have had to change their lifestyles, which may have had negative consequences on their mental health [69]. However, other stressors are arguable, as college students experience a wide range of distressing factors related to the university environment and its challenges and responsibilities [70]. Additionally, we cannot preclude that other active ingredients besides the self-efficacy training might have contributed to the observed effects. Additionally, the option to self-trigger EMI and EMA leads to less random EMA data collection, which might bias the results. Other limitations are the homogeneity of our sample, for example, regarding education, age, and origin; the unequal group sizes; and the reduced generalizability since we investigated a sample with elevated stress. Furthermore, the study is slightly underpowered because our final sample consisted of 93 participants instead of the intended 94. Therefore, replications with larger, more heterogeneous samples and in different settings are warranted.

Our study has several strengths. First, we showed multiple associations between mood and self-efficacy, respectively, specific self-efficacy leading to the conclusion that participants with lower baseline depression scores benefit more from the training than participants with higher baseline depression scores, giving the practical implication that enhancing self-efficacy might be more effective if individuals additionally receive training on how to elicit positive emotions or learn antidepressant strategies. The self-efficacy training showed a positive effect on the emotional flexibility needed to counterbalance the carryover effects of low mood.

EMA data is extremely powerful in describing and explaining momentary symptoms and momentary symptom fluctuations. In this study, we focused on the within- and between-person relationships between various symptom measures, mood, and self-efficacy. It would be an interesting avenue for future research to also explore the time-varying effects of how the constructs impact each other.

This study showed several significant associations between mood and self-efficacy as well as between mood and anxiety in students with elevated stress levels. Improving mood in participants with low mood could enhance the effects of self-efficacy training. The self-efficacy training could be especially beneficial for individuals with low emotional flexibility and contribute to mental health improvements. Additionally, analyses could focus on time-varying effects in which the constructs impact each other. Learning more about when and in which time frames changes in symptoms, mood, and self-efficacy occur might support practitioners in developing new psychological interventions.