Introduction

The university period represents a critical life stage marked by significant personal, academic, and emotional challenges. During this transition, students must develop adaptive coping mechanisms to manage stress, regulate emotions, and meet academic demands (Abdullah et al., 2020). Research suggests that fostering psychological protective factors during this period can buffer adverse effects and promote well-being (Hodges et al., 2014). Among these factors, meaning in life (MIL) has emerged as particularly relevant in the university context. MIL is associated with healthier self-esteem, greater hope, and increased positive affect, which in turn contribute to improved interpersonal relationships and enhanced resilience (Lund et al., 2022). However, the process of searching for meaning—especially in the absence of adequate social support—may also evoke distress and anxiety in young adults (Blattner et al., 2013; Malin et al., 2014).

Conceptually, MIL is commonly understood through two core dimensions: coherence (understanding one’s life) and purpose (having valued goals) (Steger et al., 2006). A third dimension, significance—the perception that one’s life has intrinsic value—has also been proposed (Martela & Steger, 2016). In the case of university students, the fulfillment of basic psychological needs (e.g., autonomy, competence, relatedness) has shown a bidirectional relationship with MIL (Zhang et al., 2022). Evidence from Spain further highlights its protective role, as MIL has been inversely associated with emotional dysregulation (Marco et al., 2017).

In recent years, growing empirical attention has focused on the psychological correlates of MIL among university populations. MIL has shown a positive association with adaptive emotion regulation strategies, particularly cognitive reappraisal, which enables individuals to reinterpret emotional situations in a constructive way (Chen et al., 2022; Hughes et al., 2011). In contrast, the use of emotional suppression is generally linked to lower levels of MIL.

MIL also correlates with affective states: it is positively associated with positive affect and negatively associated with negative affect and depressive symptoms (Ward et al., 2023; You et al., 2023). Moreover, studies have shown that positive affect mediates the relationship between MIL and adaptive coping responses (Hicks et al., 2012; Miao & Gan, 2020; Moss et al., 2023). Cultural factors have also been explored in relation to MIL; for example, Zhou et al. (2025) found that cultural identity positively influenced MIL through the mediating roles of perceived social support and resilience. In a sample of Spanish university students, Baquero-Tomás et al. (2023) reported an inverse relationship between MIL and depressive symptoms.

From a social–interpersonal perspective, MIL is inversely related to thwarted belongingness and perceived burdensomeness, two key constructs of the Interpersonal Theory of Suicide (Joiner, 2005). These associations have been observed in university samples, suggesting that MIL can buffer the negative effects of social disconnection and existential distress (Ploskonka & Servaty-Seib, 2015; Seo, 2020).

Taken together, these findings underscore the multifaceted nature of MIL and its close links with emotional, affective, and interpersonal functioning. Despite the growing interest in this construct, longitudinal studies examining how MIL evolves over time—particularly in university populations—remain limited. Most longitudinal research to date has focused on adolescents or individuals with medical conditions (Burrow et al., 2014; García et al., 2017; Marco et al., 2023). Only a few studies have tracked MIL trajectories in university students, revealing patterns of stability, increase, or decline depending on contextual variables such as mentoring experiences or emotional vulnerability (Luo et al., 2022; Morse et al., 2022).

Ecological Momentary Assessment (EMA) has recently gained prominence as a methodology for capturing real-time fluctuations in psychological constructs such as MIL. EMA reduces recall bias, enhances ecological validity, and allows for the detection of nuanced intraindividual changes (Shiffman et al., 2008). Previous EMA studies have shown that even brief, daily interventions can enhance MIL and overall well-being (Van Agteren et al., 2021). However, long-term EMA studies remain rare, particularly in university populations. Additionally, the high participant burden associated with EMA often results in attrition, which can complicate data analysis (Moskowitz & Young, 2006; Oleson et al., 2022). While most EMA protocols typically span between 1 and 4 weeks (McCarthy et al., 2015), longer-term designs are still scarce, with only a few exceptions documented in clinical populations (Cajita et al., 2023).

To date, no study has examined longitudinal trajectories of MIL using EMA over several months in Spanish university students. The present study addresses this gap by assessing MIL across 14 waves over a six-month period. This design enables a detailed analysis of developmental patterns in MIL and their psychological correlates within a community sample.

Thus, the aims of this study were: (1) to explore the longitudinal trajectories of MIL in a sample of Spanish university students over six months, and (2) to identify key psychological risk factors (e.g., negative affect, thwarted belongingness, perceived burdensomeness, emotional suppression, and depressive symptoms) and protective factors (e.g., positive affect, cognitive reappraisal, and MIL itself) associated with different trajectory groups.

Methods

Study design

This study employed a longitudinal observational design using EMA to assess MIL among university students. EMA is a real-time data collection method that captures individuals’ psychological states and experiences in their natural environments through repeated, momentary assessments. Compared to traditional retrospective designs, EMA offers several methodological advantages, including reduced recall bias, enhanced ecological validity, and the capacity to detect short-term intraindividual fluctuations over time (Moskowitz & Young, 2006; Shiffman et al., 2008).

Before the EMA phase, participants completed a comprehensive baseline assessment that included the full versions of all study instruments. These baseline data provided a detailed psychological profile for each participant and were later used to examine between-person differences associated with distinct MIL trajectories.

Following the initial assessment, data were collected over a six-month follow-up period using the MEmind application (Barrigón et al., 2017), a digital mental health platform designed to collect self-reported psychological data in real time. All assessments, including the baseline evaluation (T1), were administered via MEmind, which is accessible on smartphones, tablets, and computers. At baseline, participants completed the full set of study questionnaires in a single session through the application. During the follow-up phase (T2), only the assessment of MIL continued using EMA methodology. Participants received daily notifications prompting them to complete brief MIL assessments on their mobile devices.

The application presented items at various times throughout the day and week, enabling ecologically valid momentary data collection. Although this study focuses exclusively on MIL, other psychological variables were also assessed via EMA as part of the broader research protocol; however, these additional data fall outside the scope of the present analysis. To ensure a balanced distribution of all items, questions from the different constructs were presented across successive rounds before initiating a new cycle. During the EMA phase—comprising 14 measurement waves over six months—four selected items from the PIL-10 were repeatedly administered to assess MIL. These items were chosen to reflect key aspects such as purpose, goal-setting, and perceived life direction, while minimizing participant burden and promoting adherence to the repeated assessment schedule. To support compliance, participants received regular notifications prior to each entry, and the frequency of item presentation was progressively reduced over time.

Participants and procedures

A total of 737 Spanish university students were invited to participate, with 646 forming the final sample after providing informed consent and completing the baseline assessment. Participants were enrolled in programs such as Psychology, Speech Therapy, Nutrition, Medicine, Nursing, and Occupational Therapy. Although the age range spanned from 18 to 65 years (M = 23, SD = 5.7), the vast majority were between 18 and 25 years old. This distribution reflects the natural demographic diversity of Spanish university populations, including students enrolled in second degrees or continuing education programs (Alfageme et al., 2023).

Recruitment took place between September 2019 and January 2020. Inclusion criteria required participants to be enrolled in a university degree program, own a smartphone, understand the study procedures, and provide informed consent. Refusal to participate was the only exclusion criterion. Due to ethical and confidentiality constraints associated with encrypted data collection via the MEmind app, only participants’ academic major, gender, and age were recorded. As compensation, participants received a certificate and academic credit.

Participation gradually declined across the 14 EMA measurement waves, as shown in Fig. 1. This attrition was expected given the length and intensity of the protocol. To promote engagement, participants received scheduled notifications and academic incentives throughout the follow-up period.

Fig. 1
figure 1

Number of non - missing cases (bar plot) and average value of PIL scale (line plot) by moment of measurement

Full size image

Measures

Purpose in Life Test (PIL-10; Crumbaugh & Maholic, 1964; Spanish version

García-Alandete et al., 2013). This 10-item self-report questionnaire evaluates perceived meaning in life through two subscales: (1) life satisfaction and meaning, and (2) goals and purposes. Each item is rated on a 7-point Likert scale, with higher scores indicating a stronger perceived sense of meaning and purpose in life. An example item is: “In life I have… no goals or aspirations/many goals and defined aspirations”. The instrument has shown high internal consistency (α = 0.92), and in the present study, the baseline assessment yielded an omega coefficient of ω = 0.90. A confirmatory factor analysis supported the structure of the PIL-10 scale, showing good comparative and residual fit indices (CFI = 0.929; TLI = 0.904; SRMR = 0.040; GFI = 0.993). Although the RMSEA value exceeded conventional thresholds (RMSEA = 0.102), this may be partly due to the model’s low degrees of freedom (df = 33), a condition known to inflate RMSEA estimates (Kenny et al., 2015).

During the EMA phase, only four items from the PIL-10 were repeatedly administered to assess within-person fluctuations in MIL. These items were chosen for their conceptual representativeness—capturing dimensions such as purpose, goal-setting, and perceived life direction—while minimizing participant burden. This EMA-based assessment was conducted in addition to the administration of the full PIL-10 at baseline (T1), which provided a comprehensive between-person measure of MIL. Given the intensive longitudinal design and the focus on intraindividual change, reliability estimates based on internal consistency (e.g., Cronbach’s alpha) and confirmatory factor analyses are not appropriate, as they assume between-person homogeneity and a unidimensional factor structure at the group level. Instead, single-item indicators in EMA are widely accepted when targeting specific constructs in real time (Fisher et al., 2016), and their validity is supported by the robustness of growth mixture modeling in handling measurement error at each occasion. A confirmatory factor analysis of the 4-item version of the PIL scale used for ecological momentary assessments showed excellent fit indices (CFI = 0.988; TLI = 0.965; SRMR = 0.019; GFI = 1.000). Although the RMSEA was slightly elevated (RMSEA = 0.088), the non-significant p-value and confidence interval suggest an overall good fit, consistent with expectations for models with very low degrees of freedom (Kenny et al., 2015).

Emotion Regulation Questionnaire (ERQ; Gross & John, 2003; Spanish version

Cabello et al., 2013). This 10-item scale assesses two emotion regulation strategies: cognitive reappraisal (6 items) and expressive suppression (4 items). Items are rated on a 7-point Likert scale, with higher scores indicating greater use of the respective strategy. An example item for cognitive reappraisal is: “When I want to feel less negative emotion, I change the way I’m thinking about the situation,” and for expressive suppression: “I control my emotions by not expressing them”. In the current sample, internal consistency at baseline was ω = 0.78 for both subscales. A confirmatory factor analysis supported the two-factor structure of the ERQ, showing an acceptable model fit (χ²[34] = 160.94, p <.001; CFI = 0.929; TLI = 0.906; RMSEA = 0.072; SRMR = 0.044).

The Interpersonal Needs Questionnaire (INQ-15; Van Orden et al., 2012; Spanish version

Silva et al., 2018). This 15-item self-report measure includes two subscales: thwarted belongingness (9 items) and perceived burdensomeness (6 items). Responses are given on a 7-point Likert scale (1 = not at all true for me; 7 = very true for me), with higher scores indicating greater levels of the respective construct. A sample item for thwarted belongingness is: “These days, I feel disconnected from other people,” and for perceived burdensomeness: “These days, I think I am a burden on society.” Internal consistency at baseline in the present study was ω = 0.82 for perceived burdensomeness and ω = 0.80 for thwarted belongingness. A confirmatory factor analysis supported the two-factor structure of the INQ-15, yielding an overall acceptable fit to the data (χ²[89] = 441.72, p <.001; CFI = 0.893; TLI = 0.874; RMSEA = 0.074; SRMR = 0.057). Although the CFI and TLI were slightly below conventional thresholds, the RMSEA and SRMR indicated an adequate fit, supporting the factorial validity of the two-factor solution (Hu & Bentler, 1999).

The Patient Health Questionnaire (PHQ-9; Kroenke et al., 2001; Spanish version

Diez-Quevedo et al., 2001). The PHQ-9 assesses the presence and severity of depressive symptoms over the previous two weeks. Items are rated from 0 (not at all) to 3 (nearly every day), with higher scores indicating more severe depressive symptomatology. An example item is: “Little interest or pleasure in doing things”. In this study, the baseline omega was ω = 0.72. A confirmatory factor analysis showed a reasonable fit for the PHQ-9 (χ²[27] = 118.38, p <.001; CFI = 0.905; TLI = 0.868; RMSEA = 0.082; SRMR = 0.044).

Positive and Negative Affect Schedule (PANAS; Watson et al., 1988; Spanish version

Sandín et al., 1999). The PANAS includes 20 adjectives measuring positive and negative affect (10 items each). Participants indicate the extent to which they generally experience each emotion on a 5-point Likert scale, higher scores representing greater experience of the respective affect. Sample items include: “Excited” (positive affect) and “Upset” (negative affect). Internal consistency at baseline was ω = 0.86 for positive affect and ω = 0.85 for negative affect. The two-factor model of the PANAS showed acceptable fit, with RMSEA = 0.079 and excellent GFI = 0.975. Although CFI (0.854) and TLI (0.836) were slightly below ideal thresholds, most items loaded well on their respective factors, supporting the scale’s structural validity. The low correlation between factors (r = –.15) confirms the distinction between positive and negative affect.

All reliability and CFAs coefficients reported refer to the baseline assessment, as this was the only time point at which full versions of the instruments were administered, including the full version of the PIL-10.

Statistical analysis

In order to identify distinct longitudinal trajectories of MIL among participants, a Growth Mixture Model (GMM) was employed using scores derived from the four selected items of the PIL-10 scale, administered repeatedly during the EMA phase. The GMM allowed for the estimation of unobserved subpopulations within the sample, based on individual variability in MIL over time.

Model fit was evaluated using multiple indices, including the Akaike Information Criterion (AIC), the Bayesian Information Criterion (BIC), the Sample-Size Adjusted BIC (SSBIC), entropy values, and likelihood ratio tests, specifically the Vuong–Lo–Mendell–Rubin (VLMR) test, the Lo–Mendell–Rubin adjusted likelihood ratio test (LMR), and the Bootstrap Likelihood Ratio Test (BLRT). Higher entropy and lower AIC/BIC values were considered indicators of better model fit. Due to participant attrition at later time points, analyses were restricted to the first 11 waves of measurement. Accordingly, to address missing data related to participant attrition, we examined different patterns of missingness (MCAR – Missing Completely at Random, MAR – Missing at Random and MNAR – Missing Not at Random). As noted by Brown (2015), Maximum Likelihood (ML) assumes multivariate normality and MCAR or MAR, which can be tested using Little’s MCAR test (1988). For non-normal data, robust standard errors and test statistics can be obtained using the MLR estimator. In the context of missing data, ML estimation is frequently referred to as Full Information Maximum Likelihood (FIML). Nonetheless, the use of this term may lead to misconceptions, since ML inherently uses all available information regardless of data completeness. Little’s MCAR test yielded a non-significant result, χ² (df = 45) = 53.039, p =.192, showing that the null hypothesis—that the data are missing completely at random (MCAR) – could not be rejected. Since the pattern of missingness was consistent with the MCAR assumption, we assumed that missing values did not introduce systematic bias into the analysis. Consequently, the application of imputation techniques was deemed inappropriate, as it could lead to biased parameter estimates, including underestimated variances, inflated correlations, and underestimated standard errors (Brown, 2015).

Concerning the specification of the GMM model, linear and quadratic terms were included to enable the representation of nonlinear growth patterns, such as acceleration or deceleration over time, which are common in longitudinal data. The goodness-of-fit statistics for the linear model (AIC = 13016.9, BIC = 13126.8, SSBIC = 13050.6) and the quadratic model (AIC = 12991.2, BIC = 13133.2, SSBIC = 13034.7) revealed decreases between 15 and 25 points for AIC and SSBIC or negligible increases (below 10 points) for BIC (Burnham & Anderson, 2004). These differences indicate that the quadratic model provided a better fit to the data. Besides, the improvement in fitness was achieved without a substantial loss of parsimony, supporting the inclusion of the quadratic term to capture more nuanced developmental trajectories.

Following the identification of latent trajectory classes, participants were categorized accordingly, and subsequent analyses were conducted to examine associations between class membership and a set of psychological variables. These included cognitive reappraisal and expressive suppression (ERQ), positive and negative affect (PANAS), depressive symptoms (PHQ-9), and interpersonal variables (INQ). An exploratory factor analysis was performed using maximum likelihood estimation and varimax rotation to summarize latent dimensions among the psychological variables. A biplot graphical representation was generated to visually depict the distribution of participants by trajectory class across the extracted factors. In addition, multinomial logistic regression was applied to examine significant effects of psychological profiles across trajectory groups. To complement these analyses, a correlation matrix of all psychological variables was included as supplementary material (Table 1, Supplementary Material), providing a global overview of the relationships among constructs. All statistical analyses were conducted using SPSS and significance levels were set at p <.05.

Ethical implications

The procedure for this study was approved by the Ethics Committee of the University (Code: 072). All participants voluntarily provided informed consent prior to participation. Data collection was carried out in accordance with current Spanish legislation on the protection of personal data, and the evaluation procedures adhered to the ethical principles outlined in the Declaration of Helsinki. The data file was also registered with the Spanish Data Protection Agency.

Results

The full correlation matrix is presented in Table 1, Supplementary Material. To identify distinct student profiles based on longitudinal trajectories of MIL, an unconditional Growth Mixture Model (GMM) was conducted using the aggregated scores of the four selected items from the PIL scale. As shown in Table 1, the AIC, BIC, and SSBIC indices improved with increasing numbers of classes. All entropy values remained above 0.80, indicating good classification quality.

Table 1 Fit indices for two - to four class growth mixture models (unconditional)
Full size table

The VLMR and LMR tests indicated significant improvements in model fit when comparing the three-class to the four-class solution. No significant improvements were observed for models with fewer classes. Although the BLRT test showed significant differences across all models, it did not assist in identifying the optimal number of classes. A five-class model was also estimated, as recommended in previous research and to ensure model robustness. However, the additional class in the five-class model did not represent a qualitatively distinct trajectory pattern and was characterized by two groups with very small proportion of participants (around 2%), leading to interpretability and stability issues. Therefore, based on statistical criteria, parsimony, and theoretical interpretability, the four-class model was retained.

Table 2 presents the growth factor parameter estimates for the four-class conditional model. Classes 1 and 2 exhibited stable MIL trajectories, with non-significant slopes and curvatures. However, they differed markedly in their intercepts and class sizes. Class 1, representing a small minority of students (1.9%), reported very low MIL levels throughout the observation period, whereas Class 2 included most participants (74.1%) and showed consistently high MIL scores. Classes 3 (20.5%) and 4 (3.5%) started with high MIL levels but declined over time. Class 3 showed a gradual decrease, with significant slope and curvature estimates, while Class 4 demonstrated a steeper linear decline. Although Class 1 and 4 groups represented less than 5% of the sample, their distinct trajectories aligned with theoretical expectations and contributed meaningfully to the differentiation of student profiles.

Table 2 Growth factor parameter estimates of four-class conditional model
Full size table

The naming conventions used for each class are based on the combination of intercept levels and trajectory shapes: Class 1 was labeled as “low and stable,” Class 2 as “high and stable,” Class 3 as “declining gradual,” and Class 4 as “declining steep.” Quadratic parameters were included in the model to allow for nonlinear changes in MIL over time, especially relevant for detecting curvilinear patterns in trajectories such as the one observed in Class 3. Figure 2 visually represents these four trajectories with distinct color-coded lines, each corresponding to one of the identified classes.

Fig. 2
figure 2

Latent trajectory curves for the four groups identified based on the total purpose in Life score

Full size image

To further explore the psychological profiles associated with each MIL trajectory, an exploratory factor analysis (EFA) was conducted on a set of health and psychological variables measured at baseline. These included: cognitive reappraisal and emotional suppression (ERQ), thwarted belongingness and perceived burdensomeness (INQ), depressive symptoms (PHQ-9), and positive and negative affect (PANAS), along with the PIL total score. The rationale for applying EFA was to examine the underlying structure that captures shared variance across conceptually diverse constructs—some reflecting psychological vulnerability (e.g., depression, negative affect) and others representing adaptive functioning (e.g., positive affect, cognitive reappraisal). EFA was performed using maximum likelihood estimation, and based on Kaiser’s criterion, two latent factors were extracted, accounting for 55% of the total variance.

As shown in Fig. 3a, the first factor loaded highly (≥ 0.60) on depressive symptoms, perceived burdensomeness, and negative affect, capturing a general negative psychological functioning dimension. The second factor, represented on the Y-axis, was primarily defined by positive affect, interpreted as an indicator of positive emotional experience.

Fig. 3
figure 3

a. Factor loadings of participants’ health and psychological strategies. Exploratory Factor Analysis b. Factor scores of participants’ health and psychological strategies. Exploratory Factor Analysis

Full size image

Figure3b displays participants’ factor scores, color-coded by their assigned MIL trajectory class. The majority group (Class 2, high and stable MIL) clustered along the Y-axis, near the positive affect factor. In contrast, participants located along the X-axis—high on the first factor—were primarily those in the low MIL group (Class 1) or in the classes showing declining trajectories (Classes 3 and 4). This distribution supports the relevance of MIL as a central variable differentiating emotional and cognitive profiles in students.

These patterns were further substantiated through a multinomial logistic regression analysis, which assessed the likelihood of individual belonging to one of the three clusters characterised by low or declining MIL (Clusters 1, 3, and 4), based on their score on psychological variable. The group of participants showing high and stable MIL values (Cluster 2) was used as the reference group in the estimation of this model. Table 3 presents the estimated regression coefficients for the specified model, along with the odds ratios associated with membership in each of the risk groups characterised by low or declining MIL values, relative to the reference group exhibiting high MIL levels.

According to Table 3, the cognitive reappraisal and positive affect scales were identified as protective factors, as evidenced by statistically significant negative regression coefficients and odd ratios below 1. Conversely, thwarted belongingness, emotional suppression, and perceived burdensomeness were classified as risk factors, given their statistically significant positive coefficients and odds ratios exceeding 1. A more in-depth analysis of the risk factors revealed that perceived burdensomeness was a strong predictor of steep decline in MIL (Cluster 4), whereas thwarted belongingness was a significant risk factor for both gradual and steep decline in MIL (Clusters 3 and 4).

Table 3. Means of participants’ health and psychological strategies. Significant differences were identified using ANOVA
Full size table

The pseudo-R² values showed an acceptable goodness of fit (Cox & Snell: 0.325; Nagelkerke: 0.455). Concerning the classification performance, the model correctly classified 82% of the participants. These values suggest an adequate explanatory power, which is acceptable given the complexity and variability inherent in the psychological constructs assessed.

Discussion

The present study had two main objectives: first, to examine the longitudinal trajectories of MIL over a six-month period in a sample of Spanish university students; and second, to identify key psychological risk factors (e.g., negative affect, thwarted belongingness, perceived burdensomeness, emotional suppression, and depressive symptoms) and protective factors (e.g., positive affect and cognitive reappraisal) associated with the different trajectory groups. In relation to the first objective, four distinct MIL trajectories were identified: a small subgroup with persistently low levels of MIL (1.9%), a large group with consistently high levels (74.1%), and two groups showing declining trends—one with a gradual decrease (20.5%) and the other with a sharper drop (3.5%). Although two of these classes comprised fewer than 5% of participants, their emergence is consistent with previous person-centered research, which frequently identifies small but theoretically meaningful subgroups in longitudinal psychological data.

These four trajectories were linked to distinct psychological profiles. The high-stable group, the most prevalent, displayed greater positive affect, fewer depressive symptoms, and more frequent use of adaptive emotion regulation strategies such as cognitive reappraisal. This profile aligns with expectations for non-clinical university samples, where the absence of significant psychopathology is generally associated with effective emotional regulation and psychological adjustment. As such, this group likely reflects normative functioning in students with preserved mental health (Nadeem et al., 2023).

In contrast, the low-stable group showed persistently low MIL levels, together with higher negative affect, interpersonal burden, depressive symptoms, and a greater reliance on suppression. This pattern may reflect individuals with chronic emotion regulation difficulties that, although not necessarily reaching diagnostic thresholds, may indicate a subclinical profile. These participants might represent a vulnerable subgroup with sustained psychological distress and limited access to adaptive coping strategies. In this regard, a similar pattern of high- and low-stable purpose trajectories was observed by Ko et al. (2016) in a longitudinal study with middle-aged adults. In their work, the high-stable group showed better overall psychological adjustment, while the low-stable group was associated with less well-being, and reduced engagement in personal growth. Although their population differed from ours, these parallels reinforce the theoretical plausibility and generalizability of stable MIL profiles across different life stages.

A markedly different pattern emerged in the gradual-decline group, which showed a progressive reduction in MIL from initially high levels, along with intermediate emotional resources and greater use of avoidance-based strategies. Similar downward trends in meaning during emerging adulthood have been observed in prior research, including longitudinal declines in the search for meaning among university students (Luo et al., 2022), likely reflecting difficulties in sustaining meaning amid academic and developmental demands (Morse et al., 2023). One possible explanation for this decline is the accumulation of academic and personal stressors—such as time pressure, exam anxiety, relational difficulties, financial strain, and uncertainty about the future—which, if not effectively managed, can lead to maladaptive symptoms that undermine well-being and MIL (Park & Kang, 2022).

Finally, a more abrupt and maladaptive trajectory emerged in the sharp-decline group, which exhibited a marked decrease in MIL, accompanied from the outset by high levels of emotional distress and limited use of adaptive regulation strategies. Although this subgroup represented a small proportion of participants, person-centered approaches often identify minority trajectories that carry theoretical and practical significance. The study by Hill and Weston (2019) documented substantial inter-individual variability in purpose in life trajectories among older adults, including general declines associated with factors such as self-rated health, educational level, and marital status. These findings suggest that abrupt reductions in purpose are conceptually plausible—particularly in response to life events, the breakdown of existential goals, or emotional disconnection. This trajectory may reflect a critical point of vulnerability, underscoring the importance of early detection and targeted interventions aimed at restoring meaning and fostering relational support.

Together, these trajectories may illustrate the heterogeneous nature of MIL developmental pathways during emerging adulthood and highlight the value of differential approaches for identifying psychologically vulnerable profiles. Moreover, our findings partially align with those of Morse et al. (2023), who identified similar stable and declining MIL trajectories in American college students. However, unlike our sample, they also reported a group whose MIL increased over time, possibly due to the influence of a structured mentoring program. In our case, data collection occurred during the academic year and was not linked to structured interventions, which may explain the absence of increasing trajectories. This interpretation is consistent with previous evidence suggesting that participation in prosocial or altruistic activities fosters both MIL and well-being (Eakman, 2013).

Furthermore, our results are in line with those of Luo et al. (2022), who reported stable MIL trajectories over a one-year period among Chinese university students, indicating that in the absence of targeted interventions, MIL tends to remain stable during emerging adulthood. Similarly, Van Agteren et al. (2021) found that daily reflective practices, as part of a one-week EMA-based intervention, significantly increased MIL and well-being. Although our study did not implement an intervention, the self-monitoring nature of EMA may have encouraged some participants to engage in personal reflection, potentially contributing to the stability of their MIL levels.

In addition, our findings are partially consistent with those of Zhou et al. (2025), who found that cultural identity influences MIL among university students through perceived social support and resilience. However, unlike their cross-sectional study focused on static mediating variables, our research employed a longitudinal design using EMA, which allowed us to capture intraindividual changes in MIL over time.

However, while our results suggest relative stability, studies such as Chen and Cheng (2020) have reported gradual increases in MIL during adolescence, which may be attributable to the distinct developmental tasks and support structures characteristic of that life stage. In our sample, some students were already immersed in mid-semester academic demands, which may have influenced their sense of purpose and overall psychological adjustment.

The second objective of the study was to identify psychological risk and protective factors associated with different MIL trajectories. Two broad latent profiles emerged: participants with low or declining MIL trajectories (Groups 1, 3, and 4) reported higher levels of depressive symptoms, perceived burdensomeness, and negative affect. Conversely, participants with stable and high MIL (Group 2) displayed elevated levels of positive affect and more frequent use of cognitive reappraisal strategies.

These findings are consistent with prior literature. Perceived burdensomeness has been shown to be more strongly associated with psychological distress than thwarted belongingness (Chu et al.,2017), including links to anxiety and suicidal ideation (Hill et al.,2018), as well as mental health risks in marginalized youth (Baams et al., 2018). Likewise, depressive symptoms have been negatively related to MIL and life satisfaction, whereas positive affect and the presence of purpose in life (PIL) have been found to serve as psychological buffers (Seo, 2020). In our sample, positive affect emerged as a particularly robust correlate of high MIL, which is consistent with prior evidence that emotional well-being and spiritual orientation predict a stronger sense of purpose (Heydarinasab & Ghomian, 2019).

Students with the lowest MIL levels also reported greater negative affect, depressive symptoms, and interpersonal distress (e.g., perceived burdensomeness and thwarted belongingness). These findings reinforce previous research suggesting that MIL can serve as a buffer against the negative impact of psychological vulnerabilities (Marco & Alonso, 2019; Marco et al., 2021 ). He et al. (2023) also identified depression as the strongest emotional predictor of diminished MIL, although negative affect may be particularly salient among university students at risk of suicidal ideation (Yang et al., 2020).

Finally, participants with high and stable MIL reported greater use of cognitive reappraisal strategies, consistent with evidence indicating a positive association between this form of emotional regulation and MIL (Park et al., 2008). Troy et al. (2018) proposed that individuals with higher levels of MIL are better equipped to reinterpret difficult experiences, thereby reducing emotional distress. In our study, this capacity may have contributed to the psychological stability observed in this group, despite ongoing academic demands.

To identify the psychological variables that statistically predicted class membership, we employed a multinomial logistic regression model using the high-stable MIL group (Cluster 2) as the reference. This analysis revealed that cognitive reappraisal and positive affect were significant protective factors, reducing the odds of belonging to risk groups with low or declining MIL. In contrast, perceived burdensomeness and thwarted belongingness emerged as key risk factors, particularly for the steep-decline group. Emotional suppression also predicted gradual decline in MIL. These findings suggest that difficulties in emotion regulation and interpersonal connectedness may play a central role in the erosion of meaning over time, whereas adaptive strategies like cognitive reappraisal may help buffer against such decline.

These findings also underscore the importance of early identification of students with declining or chronically low levels of MIL, as they may be at heightened psychological risk. University support services could incorporate brief screening tools to systematically detect such profiles and implement targeted interventions to enhance meaning and psychological resilience. These may include structured mentoring programs, meaning-centered psychotherapies, or cognitive reappraisal training. Interventions that promote positive affect, adaptive emotion regulation, and interpersonal connection may serve as effective protective strategies to foster well-being and support students’ academic adjustment.

Limitations and future directions

Despite the contributions of this study, several limitations should be noted. First, the sample consisted exclusively of Spanish university students, which may limit the generalizability of the findings to other cultural or educational contexts. Future studies should replicate these analyses in more diverse populations and across different countries to examine potential cultural variations in MIL trajectories. Second, although EMA allowed for capturing intraindividual changes, the time frame was limited to six months; longer follow-up periods are needed to understand the long-term evolution of meaning in life during emerging adulthood. Third, although we identified key psychological correlates of MIL trajectories, future studies should explore causal mechanisms and mediating processes, for example, by examining how interventions targeting emotion regulation or social connectedness may influence changes in MIL over time. Finally, the study used a brief 4-item version of the PIL scale adapted for ecological momentary assessment. Despite its brevity, the scale showed adequate fit, supporting its unidimensionality. Overall, findings support the scale’s validity for capturing meaningful within-person fluctuations in naturalistic contexts. Future studies could test its invariance across populations and further refine short-form ecological tools.

Conclusion

This study provides novel insights into the longitudinal trajectories of MIL in Spanish university students using EMA. The results suggest that maintaining stable levels of MIL over time is possible, particularly among students who report higher levels of positive affect and cognitive reappraisal. These variables appear to buffer the impact of depressive symptoms, negative affect, and perceived burdensomeness.

To our knowledge, this is the first study to explore MIL trajectories over a six-month period using EMA in a non-clinical Spanish university sample. While most EMA studies span 1 to 4 weeks (McCarthy et al., 2015; Rowan et al., 2007), our extended design allowed for the observation of mid-term trends and stability patterns.

These findings underscore the need for university-based interventions that promote emotional well-being and strengthen MIL. Programs encouraging self-reflection, emotional regulation, and positive affect may help students navigate academic and personal challenges. Future studies should explore predictive models of MIL trajectories and assess the effectiveness of meaning-centered interventions in fostering resilience and academic engagement.