We used data from a study on students’ experiences during the pandemic at the Faculty of Humanities of a large public university (more than 30,000 students) in Lower Saxony, Germany. Students were contacted twice per semester via a university mailing list and could join the study at any time point by providing their written consent. All administered surveys (T1–T5) are shown in Figure 1. Our overall sample consisted of 1,909 students who provided information on at least one of our variables of interest in the present study. Of these, 431 students completed at least two surveys and were included in our analyses (for additional study sample information by time point, see Supplementary Table 2). Our sample included all demographic groups of the faculty’s student population, with an overrepresentation of female students (78% female; 48% first-generation students).³

If not stated otherwise, we analyzed our data using Mplus version 8.5 (Muthén and Muthén, 1998–2017). Prior to our main analyses, we examined missing data patterns in our sample using Little’s missing completely at random (MCAR) test (Little, 1988) within SPSS’s Missing Value Analysis option (version 28.0; IBM Corp, 2021). Additionally, we tested longitudinal measurement invariance (Liu et al., 2017).

Based on research on within-person fluctuations in thriving over moderate periods (e.g., Brown et al., 2021), we were interested in changes in thriving between immediately consecutive time points (e.g., T1–T2, T2–T3). To address this, in our main analysis, we used a neighbor change model, a subtype of latent change score (LCS) models (Geiser, 2010, 2013). We first specified a neighbor change model without predictors to assess model fit and examine the overall pattern of intraindividual change in students’ thriving over time (e.g., Ferrer and McArdle, 2010). Next, we simultaneously included gender and college generation status as predictors of students’ thriving at T1 and intraindividual changes over time, given that these variables may be interrelated through intersecting identities (e.g., Wright et al., 2023). Alternative model structures (e.g., latent autoregressive models; Geiser, 2010) were deemed less appropriate due to their emphasis on the relative interindividual stability of constructs over time and the indirect modeling of change through residuals. In contrast, LCS models directly capture change via latent difference variables (e.g., Geiser, 2010; Kleinke et al., 2017; McArdle and Hamagami, 2001; Steyer et al., 1997), aligning more closely with our interest in modeling dynamic intraindividual processes across specific consecutive time points. Missing values were estimated using full information maximum likelihood (FIML; Enders, 2010). Unlike traditional imputation methods, FIML directly estimates model parameters using all available data without imputing missing values and generally yields less biased and more efficient parameter estimates than other methods under both MCAR and MAR conditions (Enders and Bandalos, 2001; in our case: MCAR).

According to Little’s MCAR test, the data (χ2 = 227.40, df = 220, p = 0.352) were missing completely at random. Based on the criteria proposed by Chen (2007), longitudinal measurement invariance was supported, with full configural and metric invariance and partial scalar invariance across all five time points (for detailed model fit indices, see Supplementary Table 1).

Students reported above-average levels of thriving at the beginning of the study period (T1, summer 2020; estimate = 3.01, SE = 0.06, p ≤ 0.001). From T1 toward T2 (fall 2020), during remote instruction, students had significant intraindividual decreases (estimate = −0.28, SE = 0.07, p ≤ 0.001). After these 6 months, from T2 toward T3 (early 2021), students’ thriving levels did not change when universities continued with remote instruction (estimate = −0.02, SE = 0.07, p = 0.833). As expected, we found intraindividual increases from T3 toward T4 (fall 2021), when universities had returned to in-person instruction (estimate = 0.24, SE = 0.09, p ≤ 0.01). There were no intraindividual changes from T4 toward T5 (early 2022), when universities had resumed remote instruction (estimate = −0.07, SE = 0.08, p = 0.381; for all results, see Table 1 and Figure 2; for the means of students’ thriving across all time points, see Supplementary Figure 1).

Neither students’ thriving at T1 nor their intraindividual changes in thriving toward later time points differed by demographic background—that is, by gender or college generation status (for all results, see Table 2).

While our data do not permit causal conclusions—given the limited scope of predictors included—we offer several theoretically grounded interpretations of the observed patterns.

At the onset of the pandemic (in our study: T1), university students reported slightly above-average thriving levels, aligning with cross-sectional study findings of moderate thriving levels among university students during the height of the pandemic (Sahin and Tuna, 2022; Warnock et al., 2024). Students showed intraindividual decreases in their experiences of thriving from the onset of the pandemic toward fall 2020 (in our study: T2). This pattern is consistent with declines observed in related constructs such as students’ well-being (e.g., Buizza et al., 2022) and study satisfaction (e.g., Gadosey et al., 2022) during the first academic year of the pandemic. Although not directly assessed in our study, the observed decline may reflect study-related stressors in online learning environments perceived as threatening (see Lazarus and Folkman, 1984), including reduced opportunities for interaction with instructors and peers (e.g., Elmer et al., 2020; Hollister et al., 2022) and stressors associated with transitioning academic work to virtual learning environments (e.g., El-Sakran et al., 2022). Despite evidence that students’ study satisfaction also decreased during winter 2020/2021 (Gadosey et al., 2022), we found no changes in students’ thriving from fall 2020 toward early 2021 (in our study: T3). One possible explanation is that students faced fewer disruptive changes in their academic life during this period. In contrast to the early stages of the pandemic, when digital tools and assessment formats were still less established, students in winter 2020/2021 may have begun to adapt, drawing on prior experience and developing learning- and performance-related (interpersonal) coping strategies (see Liebendörfer et al., 2023). This, in turn, may have resulted in greater continuity in students’ academic routines, and, consequently, stable levels of thriving (see Porath et al., 2012). Such a pattern aligns with frameworks describing distinct levels of functioning in response to adversity (O’Leary and Ickovics, 1995; see also Carver, 1998). When facing adversity, individuals can exhibit resilience by maintaining their typical, baseline level of functioning despite ongoing stressors or by showing a homeostatic return to prior levels of functioning after adversity. Supporting this, research indicates that stress levels of university students also remained stable during this period (e.g., Weber et al., 2022), despite escalating COVID-19 case numbers and renewed lockdown measures in Germany in winter 2020/2021 (Bundesministerium für Gesundheit Deutschland, 2023).

As expected, a key finding of our study was the pattern of intraindividual increases in students’ thriving levels toward fall 2021 (in our study: T4), when students had returned to their classrooms. This increase occurred despite ongoing study-related stressors such as social unease (Karnbach et al., 2024) and may be partly attributable to the renewed availability of in-person social interactions, which had previously been limited or absent during the first year and a half of remote instruction (e.g., Elmer et al., 2020; Hollister et al., 2022). In particular, supportive relationships with instructors and peers (see Imran et al., 2020; Kleine et al., 2019) may have offered key social coping resources that enabled students to successfully manage the study-related stressors upon returning to campus, thereby facilitating their thriving (see Feeney and Collins, 2015). This interpretation is consistent with a recent study on university students’ social belonging as a predictor of thriving during in-person instruction. Haase et al. (2025) found that the extent to which university students felt that they were part of their study program contributed to their experiences of vitality and learning. Social belonging is often positively related to the perceived availability of social support, for example, from instructors and peers at university (e.g., Walton and Cohen, 2007; see also Chiu et al., 2016). Moreover, research suggests that social belonging can be conceptualized as an indicator of the perceived quality of one’s interpersonal relationships in a given context (e.g., Shook and Clay, 2012; Walton and Cohen, 2007; see also Li et al., 2025). The return to campus may have increased the salience of the contextual environment (i.e., the respective study program), thereby strengthening students’ sense of social belonging and contributing to their thriving levels. However, these considerations remain speculative, as our study could not include social predictors of thriving across time due to the limited sample size and therefore cannot establish direct links between social experiences and thriving.

Consistent with pre-pandemic longitudinal research (e.g., Brown et al., 2021; Kaltenbrunner et al., 2019; Kleine et al., 2023), intraindividual changes in students’ thriving toward all later time points did not differ by gender. While not empirically tested in our study, it seems that both female and male students lacked sufficient coping resources to sustain thriving during the unprecedented study situation at the beginning of the pandemic (see Lazarus and Folkman, 1984), which likely contributed to similar decreases toward fall 2020 (in our study: T2). Interestingly, studies have found greater mental health impairments—such as higher stress levels and lower levels of well-being—among female compared to male students during this period (e.g., Barbayannis et al., 2022; Gestsdottir et al., 2021; Prowse et al., 2021). This may indicate that thriving in higher education reflects a relatively broad form of adaptive psychological functioning that is less sensitive to gender differences than other indicators, such as well-being or stress, which are more closely tied to immediate emotional states (see Graves et al., 2021; Ryff et al., 2021). In general, psychological functioning and development can include hedonic and eudaimonic dimensions (e.g., Ryan and Deci, 2001; Ryff et al., 2021), as already outlined. Since thriving has been proposed to encompass both hedonic (vitality) and eudaimonic (learning) dimensions (Spreitzer et al., 2005), it can be suggested that the interplay of these dimensions in thriving involves more stable aspects of personal development, which are suggested to be less sensitive to gender differences than hedonic well-being, which consists of less stable states of positive emotions (see Graves et al., 2021; Ryff et al., 2021). This may also explain the comparable patterns of stability in thriving levels observed among both female and male students beyond the first six months (in our study: T2–T3).

Following the return to in-person instruction (in our study: T4), both female and male students’ thriving levels increased. This similar pattern of change is notable given that prior research suggests that thriving through social factors may be related to gender role beliefs (Feeney and Collins, 2015), with women often (self-)described as more community-oriented than men (e.g., Eagly, 1987; Kite et al., 2008; Kosakowska-Berezecka et al., 2023). Supporting this, women have been shown to thrive more during social interactions in face-to-face settings. For example, the association between positive leader–member exchange (e.g., mutual affect, professional respect) and thriving has been found to be stronger for women than for men (Di Milia and Jiang, 2024). However, first evidence suggests that men perceive themselves as similarly communal as women (e.g., Obioma et al., 2022), which could help explain why both genders experienced similar increases in their thriving levels when social coping resources became more accessible. Nonetheless, our study does not identify the specific social (or individual) coping resources that may have contributed to these gains in thriving among female and male students after the return to in-person instruction.

Likewise, first-generation (FG) and continuing-generation (CG) students showed similar developments in their thriving. This aligns with research showing that levels of flourishing did not differ between FG and CG students from the beginning of the pandemic until the end of the academic year in 2021 (Lipson et al., 2023; in our study: T1–T5)—despite the greater mental health impairments typically reported by FG students during the first academic year of the pandemic (e.g., Fruehwirth et al., 2021; Soria et al., 2020). Although not directly assessed, one possible explanation for the similar decreases toward fall 2020 (in our study: T2) and the similar increases in FG students’ thriving after having returned to campus (in our study: T4), compared to CG students, may be off-campus interpersonal relationships, social coping resources FG students frequently draw on. For instance, in a qualitative study, associations were found between FG students’ well-being and their off-campus interpersonal relationships in their home communities, in addition to on-campus social interactions (McCarron, 2022). These off-campus interpersonal relationships may have buffered against stronger decreases and supported similar increases in thriving levels compared to CG students.

First, although our study adds to the literature on positive outcomes as responses to the specific study situation during the pandemic, we could not include pre- and post-pandemic time points and therefore are not able to separate the specific effects of the pandemic from general effects during university students’ studies. For example, research has shown that positive psychological states generally tend to decrease over the course of students’ studies (for well-being, see Bewick et al., 2010; Cobo-Rendón et al., 2020).

Second, another limitation pertains to the time period examined in our study. Based on research that has investigated thriving over moderate periods in work contexts (Brown et al., 2021; Kleine et al., 2023), it seemed crucial to investigate university students’ thriving over a longer period. However, to gain a deeper understanding of the microdynamics of thriving in higher education, future research should examine university students’ thriving over shorter periods using methods such as experience sampling or diary studies (e.g., Hua et al., 2022; Niessen et al., 2012).

Third, our study did not include social predictors of thriving, limiting our ability to test whether changes in the availability of social coping resources actually contributed to the observed fluctuations. An interesting next step would be to examine how social coping resources—such as perceived support from instructors and peers—relate to students’ thriving over short and moderate time periods (for the role of host country support in international students’ thriving over a 9-day period, see Hua et al., 2022). In addition, future studies could investigate the underlying mechanisms through which social factors, such as close relationships, contribute to long-term thriving in higher education (Feeney and Collins, 2015). For example, in workplace contexts, a recent longitudinal study identified organization-based self-esteem as a mediator between social coping resources and thriving (Kim and Beehr, 2022). In developing such models, gender should be systematically considered, as it may shape the extent to which students benefit from social support. Although both female and male students showed increased levels of thriving after the return to in-person instruction, these patterns should be replicated in higher education, given that gender role beliefs—unmeasured in our study—may moderate how social coping resources are experienced and utilized (Feeney and Collins, 2015; Kosakowska-Berezecka et al., 2023).

Fourth, the lack of significant effects for gender and college generation status should be interpreted with caution due to the sample composition and missing data. The unequal gender distribution, with a smaller number of male participants, may have reduced the statistical power to detect gender-related effects (Aguinis, 1995; Alexander and DeShon, 1994). However, by accounting for measurement error through the use of latent variables, LCS models can enhance the overall precision of parameter estimates (e.g., Kline, 2016). While missing data on college generation status were handled via full information maximum likelihood (FIML; e.g., Lee and Shi, 2021; Newman, 2014), this approach does not fully mitigate the risk of reduced estimate precision and increased standard errors that can result from missing data.

Fifth, we used a shortened three-item version of the Thriving at Work Scale by Porath et al. (2012) to ensure longitudinal measurement invariance. While short scales can reduce conceptual breadth, reliability, and content validity (Credé et al., 2012; Marsh et al., 1998; Xiao et al., 2024), research shows that well-selected short scales can retain psychometric strength, often comparable to longer item versions (e.g., Gogol et al., 2014; Kjell and Diener, 2021; Kleine et al., 2023). Our items reflected both theoretical core dimensions of thriving, i.e., vitality and learning, suggesting that intraindividual changes in this three-item score still meaningfully captured changes in students’ overall thriving (see Porath et al., 2012). Still, future studies should replicate these findings using more comprehensive measures.

Lastly, because our study was limited to one specific university in one German state, our findings should not be generalized to the experiences of vitality and learning among university students in Germany during the pandemic. For instance, prevalence in Lower Saxony during our survey period was below the national average at most time points (see Supplementary Table 3), potentially leading to an overestimation of thriving increases at T4. However, particularly during earlier phases (T1–T3), infection rates were comparable to national levels.

Although our study focused on the unique circumstances of the COVID-19 pandemic, our findings may reflect more general patterns that are relevant to higher education institutions beyond this specific context. Accordingly, we outline several implications that may inform how higher education institutions can support students’ thriving across different stages of their studies.

Our findings show that students’ thriving can fluctuate substantially over two academic years (see Brown et al., 2021; Kleine et al., 2023). Given the dynamic nature of thriving, higher education institutions may benefit from systematically monitoring medium-term developments in students’ vitality and learning across academic terms. To implement such monitoring, short, validated thriving assessments (e.g., Porath et al., 2012; see also Haase et al., 2025) could be embedded into existing structures, such as institutional surveys on students’ mental health and well-being, administered in paper-and-pencil or digital formats (e.g., Lahtinen et al., 2023).

Given that fluctuations in thriving can occur in response to substantial changes (e.g., Porath et al., 2012; see also Brown et al., 2021), monitoring may help to examine whether changes in students’ thriving are associated with transitions between different study phases, for example, transitions from regular coursework to thesis work or high-stakes group projects. Given the important role of social relationships in fostering thriving (Feeney and Collins, 2015; Imran et al., 2020; Jiang et al., 2020), it seems important to examine how the relevance of various dimensions of social relationships with instructors and peers might shift across these transitions. Thriving could therefore be examined alongside instructor-related and peer-related factors that have been shown to support thriving in work contexts. Instructor-related factors encompass mentoring (see Jiang et al., 2020)—including challenging assignments and emotional support—and feedback on important academic decisions (see Kim and Beehr, 2020), such as students’ grades and academic progress (for a measure of instructors’ self-perceived ability to support students, see also Haase and Zander, 2022). After the transition from coursework to thesis work, for instance, students may particularly rely on individualized mentoring and expert feedback (see Jiang et al., 2020; Kim and Beehr, 2020), which may positively contribute to their thriving. At the peer level, relevant factors include support from fellow students (see Zhai et al., 2020), feelings of connection (see Carmeli and Spreitzer, 2009), and mutual recognition of strengths (see Moore et al., 2022). After the transition to demanding group-based tasks, for example, thriving may be particularly shaped by interpersonal dynamics such as feelings of connectedness (Carmeli and Spreitzer, 2009). Taken together, fluctuations in students’ thriving may partly reflect variation in the availability and quality of social relationships, resulting in greater intra- and interindividual variability (Brown et al., 2021; Kleine et al., 2023).

In light of our findings suggesting that remote instruction may pose specific challenges for students’ thriving, higher education institutions may pay particular attention to creating positive and socially supportive online learning environments, to, in turn, promote students’ thriving. Effective instructional strategies may include interactive communication tools that support direct exchange between instructors and students. These tools have been shown to increase students’ satisfaction—a construct conceptually close to thriving—through instructors’ social presence (i.e., students’ perception that instructors are attentive and involved in the online interaction; see Park and Kim, 2020). Equally important, collaboration tools that facilitate peer-to-peer interaction (see Maxrath et al., 2025), such as shared social annotation platforms, can enhance students’ perceived social belonging in online courses (Kelly and Clinton-Lisell, 2025). Social belonging, in turn, has been found to predict students’ thriving (see Haase et al., 2025).

As our findings suggest that thriving fluctuated not only across time but also appeared sensitive to shifts in the broader learning environment, targeted, phase-specific interventions may help buffer against declines and foster recovery (for a systematic review on peer support interventions in higher education, see Pointon-Haas et al., 2024).