Abstract
Negative life events can lead to changes in self-esteem, with diverging effects across individuals. To better understand these individual differences, we examined self-esteem trajectories of 1,069 participants over 6 months following a negative life event. Using latent class analysis, we identified four classes with distinct self-esteem trajectories. While three classes showed comparable increases in self-esteem following negative events, indicating an adaptation to the event over time, one class exhibited no significant change, suggesting a lack of recovery. The Big Five personality traits and, to a lesser extent, perceived event characteristics predicted class membership and individual differences in self-esteem levels following negative life events. Neuroticism, followed by extraversion and social status change, emerged as the strongest predictors. Conversely, the type of event and demographic variables played a minor role in explaining individual differences in self-esteem development after negative events. Together, our findings highlight the importance of both event- and person-related factors in shaping self-esteem development following negative life events and may help identify and support individuals at risk for negative self-esteem trajectories.
Plain language summary
People differ in how their self-esteem responds to negative life events like job loss or the end of a romantic relationship. Some may experience a decrease in self-esteem, others may maintain it, and some may even see an increase. To better understand these differences, we studied the self-esteem of 1,069 people over six months after they experienced one of five negative events and found four distinct self-esteem development patterns. Three patterns showed an increase in self-esteem, suggesting that most people’s self-esteem recovers over time after negative events. However, one pattern showed no change in self-esteem, indicating some individuals’ do not recover. To understand why people show particular patterns of self-esteem development, we examined how event- and person-related factors predicted these patterns. We found that a person’s personality traits played a key role for self-esteem development after negative life events, with emotional stability—the tendency to react relaxed to stressful situations—being the most important factor, followed by extraversion—the tendency to be outgoing and energetic. People with lower emotional stability and extraversion were less likely to recover in their self-esteem after negative events. Additionally, peoples' perception of the event predicted patterns of self-esteem development, with those who perceived the event as more social status threatening being less likely to recover in their self-esteem. Together, our findings highlight the role of both peoples’ personality and their perception of the event in self-esteem changes after negative life events, with implications for both clinical practice and research. They may help identify people in need of support after negative events and inform theories of self-esteem development.
Keywords
Negative life events like job loss or a separation can threaten people’s self-esteem, which typically decreases in response to these events (e.g., Haehner et al., 2025; Luciano & Orth, 2017; Reitz, Luhmann et al., 2022). However, individuals exhibit notable differences in their reaction to negative life events: while some experience enduring drops, others manage to maintain or even enhance their self-esteem (e.g., Bleidorn et al., 2021; Luciano & Orth, 2017; Reitz, Weidmann et al., 2022).
To date, the variables that explain these individual differences have received surprisingly little attention (Reitz, 2022). Yet understanding which individuals are most affected by negative life events is crucial for developing strategies to enhance self-esteem and health in the general population (e.g., Lee et al., 2013; Orth & Robins, 2022; Reitz, 2022). Furthermore, identifying the variables underlying individual differences in self-esteem change after negative events could advance theories of self-esteem development by uncovering the mechanisms driving these changes (Reitz, 2022).
By analyzing data from a longitudinal study of participants who recently experienced a negative life event, the present study aimed to advance our understanding of individual differences in self-esteem development following negative life events. We addressed two research questions: (1) Are there classes of individuals with distinct self-esteem trajectories following negative life events? (2) Which variables explain individual differences in self-esteem changes in response to negative life events?
Self-esteem development across the lifespan and in response to negative life events
Self-esteem describes people’s subjective evaluation of their own worth (Rosenberg, 1965; Stinson & Fisher, 2020) and is of critical importance for various life outcomes, including the quality of social relationships (Harris & Orth, 2020), academic achievement (Valentine et al., 2004), work experiences (Krauss & Orth, 2022), and both mental (Orth & Robins, 2022; Sowislo & Orth, 2013) and physical health (Orth et al., 2012). People’s self-esteem changes normatively across life, typically peaking in middle and late adulthood (Orth et al., 2018). However, individual trajectories can diverge from this normative pattern (e.g., Hutteman et al., 2014; Orth et al., 2018; Wagner, Gerstorf et al., 2013). One factor that may contribute to individual differences in self-esteem development is the occurrence of negative life events.
Generally, life events can be defined as time-specific, personally significant disruptions of everyday life requiring psychological and behavioral adjustment (Luhmann et al., 2021). Contemporary theories propose that self-esteem development is, at least partly, driven by adaptations to (negative) life events (Reitz, 2022). For example, Sociometer Theory (Leary & Downs, 1995; Leary et al., 1995) proposes that self-esteem functions as a gauge of social inclusion, making it sensitive to social feedback. Negative life events like friendship dissolution often involve social rejection, which can reduce perceived inclusion and thus reduce self-esteem. Similarly, Hierometer Theory (Mahadevan et al., 2016, 2021) suggests that self-esteem serves as a measure of social status and is susceptible to fluctuations due to status-related events like job loss. Finally, the Integrative Process Framework of Self-Esteem Development (Reitz, 2022) proposes that negative life events can affect self-esteem by changing individuals’ everyday situations. For example, job loss may disrupt daily routines, strain social roles, and cause financial difficulties, all of which provide feedback about success or failure and may decrease individuals’ momentary self-esteem. Over time, these changes in momentary self-esteem can result in enduring shifts in trait self-esteem.
Building on these theories, several studies have examined self-esteem development following negative life events. Although findings are mixed across studies and types of negative events, consensus seems to emerge regarding the typical self-esteem trajectory in response to such events. Self-esteem has been found to decline in the months and years prior to the event occurrence, reach a low point shortly after the occurrence, and recover within the 4 years that follow (Bleidorn et al., 2021; Luciano & Orth, 2017; Mangelsdorf et al., 2019; Reitz, Luhmann et al., 2022). Thus, event-related changes in self-esteem seem to be somewhat more consistent than mean-level changes in the Big Five personality traits (Bühler et al., 2024), although effect sizes are rather small for both constructs (Haehner et al., 2025). Yet the average self-esteem trajectory provides little insight into how specific individuals change after negative life events. To fully understand the impact of such events, it is essential to investigate how and why individuals’ responses differ.
Individual differences in self-esteem trajectories in response to negative life events
In recent years, research has increasingly focused on individual differences in reaction to negative life events (e.g., Bleidorn et al., 2020; Haehner, Bleidorn et al., 2024; Haehner, Wright et al., 2024). Prior evidence indicates that individuals differ in the direction, magnitude, and duration of self-esteem changes following such events. For instance, significant heterogeneity in self-esteem development has been observed among divorcees, with some showing self-esteem decreases and others increases (Bleidorn et al., 2021). Similar findings have been reported for breakups (Luciano & Orth, 2017), the death of a loved one (Reitz, Weidmann et al., 2022), and job loss (Reitz, Luhmann et al., 2022).
Despite this evidence of individual differences in self-esteem responses to negative life events, it remains unclear who is most affected by such events. The present study aimed to address this gap by identifying classes of individuals with distinct self-esteem trajectories following negative life events. Specifically, we sought to answer the question: Can classes of individuals with distinct self-esteem trajectories following negative life events be identified, and if so, what do these classes look like and what distinguishes them? Identifying such classes helps to understand individual differences in self-esteem development after negative events, for example, what percentage of the sample shows increases, decreases, or stability in self-esteem (Mancini et al., 2011). Furthermore, it allows examining whether these classes differ in important characteristics that may explain individual differences.
Why do people differ in their self-esteem trajectories after negative life events?
Knowledge on why individuals differ in their self-esteem trajectories following negative life events provides valuable insights beyond normative self-esteem development. Furthermore, it may help identify risk and resilience factors for self-esteem, supporting early identification of individuals vulnerable to lasting self-esteem decreases after negative life events and the development of tailored interventions.
Research on individual differences in well-being and personality traits has shown that changes following negative life events depend on event-related variables such as the type of event and an individual’s event perception (e.g., Haehner, Bleidorn et al., 2024; Haehner, Kritzler et al., 2024; Schwaba et al., 2023; Wilkinson et al., 2023). Furthermore, studies indicate that self-esteem development may also depend on person-related variables like personality traits and demographic variables (Orth et al., 2018; Wagner, Gerstorf et al., 2013; Weidmann et al., 2017, 2018). In the present study, we combined both perspectives and examined whether event-related (i.e., event type and perception) and person-related variables (i.e., personality traits and demographic variables) can predict individual differences in self-esteem development following negative life events. In the following, we describe our rationale for examining these variable clusters and present hypotheses about their relevance.
Event type
Negative life events differ in type (e.g., job loss vs. divorce), with each type affecting different life domains and social roles and recent research suggests that event-related changes in the Big Five personality traits depend on the type of event (Haehner, Rakhshani et al., 2023). Theory suggests similar expectations for self-esteem. For example, according to Sociometer Theory (Leary, 2005), events involving rejection-related emotions, such as friendship dissolution, should impact self-esteem more than other types of negative events. Conversely, other authors argue that events tied to success or failure, such as job loss, lead to greater changes in self-esteem (Crocker & Wolfe, 2001). To date, research has rarely compared how different types of negative events affect self-esteem. Therefore, we did not posit any specific hypothesis for event type. Instead, we examined the relation between five types of negative life events—the death of a loved one, the end of a romantic relationship, friendship dissolution, job loss and failing an important exam—and self-esteem development exploratorily.
Event perception
The perception of life events refers to an individual’s subjective evaluation and interpretation of an event (Haehner, Kritzler et al., 2024; Luhmann et al., 2021). For example, while one person may view a friendship dissolution as highly distressing, another may feel the opposite. Luhmann et al. (2021) proposed nine key characteristics of event perception: valence, impact, predictability, challenge, emotional significance, change in world views, social status change, external control, and extraordinariness. Drawing on existing evidence and contemporary self-esteem theories, we formulated hypotheses on how three of these perceived characteristics—challenge, impact, and social status change—relate to individual differences in self-esteem development following negative life events.
First, perceiving negative life events as more challenging is linked to higher levels of depressive symptoms and reduced well-being (Haehner, Kritzler et al., 2024; Haehner, Pfeifer et al., 2023; Haehner, Würtz et al., 2024). We expected these findings to extend to self-esteem and hypothesized that individuals who perceive negative life events as more challenging exhibit less positive self-esteem changes (Hypothesis 1).
Second, perceiving negative life events as more impactful is associated with stronger effects on people’s well-being (Luhmann et al., 2021) and personality traits (Löckenhoff et al., 2009; Roberts & Nickel, 2017). Based on these findings, we hypothesized that individuals who perceive negative life events as more impactful exhibit less positive self-esteem changes (Hypothesis 2).
Third, both Sociometer Theory (Leary, 2005) and Hierometer Theory (Mahadevan et al., 2016) conceptualize self-esteem as a social gauge, emphasizing social acceptance and status. Consequently, threats to one’s social status may lower self-esteem due to a declining sociometer or hierometer. Therefore, we hypothesized that individuals who perceive negative life events as more social status threatening exhibit less positive self-esteem changes following the event (Hypothesis 3).
For the other perceived event characteristics proposed by Luhmann et al. (2021), we did not formulate specific hypotheses but examined their relation to event-related self-esteem development exploratorily.
Big Five personality traits
Regarding person-related variables, the Big Five personality traits are expected to play an important role in explaining individual differences in self-esteem development. Higher self-esteem has consistently been associated with greater openness, conscientiousness, extraversion, and agreeableness, as well as lower neuroticism (e.g., Allemand et al., 2024; Amirazodi & Amirazodi, 2011; Bleidorn & Schwaba, 2018; Wagner, Gerstorf et al., 2013, Wagner, Lüdtke et al., 2013; Zeigler-Hill et al., 2015). Additionally, neuroticism has been linked to less positive self-esteem trajectories, with effects observed across time spans from several months to multiple years (e.g., Fetvadjiev & He, 2019; Poorthuis et al., 2014; Wagner, Lüdtke et al., 2013; Weidmann et al., 2017, 2018). Based on this evidence, we hypothesized that individuals with higher neuroticism exhibit less positive self-esteem changes (Hypothesis 4). Given the less clear associations of the other Big Five personality traits (i.e., openness, conscientiousness, extraversion, and agreeableness) with self-esteem development, these relationships were examined exploratorily.
Demographic variables
Beyond personality traits, demographic variables may explain why individuals differ in their self-esteem development following negative life events. Age, gender, education, income, and ethnicity are associated with self-esteem both cross-sectionally (e.g., Bleidorn et al., 2016; Robins et al., 2002; Twenge & Campbell, 2002) and longitudinally (e.g., Bleidorn et al., 2023; Orth et al., 2018; Wagner, Gerstorf et al., 2013). Additionally, recent research suggests that demographic variables shape self-esteem responses to negative life events. For example, Reitz, Luhmann et al. (2022) found that age and gender moderated self-esteem development following unemployment. However, longitudinal research replicating these effects and extending them to other relevant demographic variables like education and income is lacking. We thus explored how age, gender, education, income, and migration status relate to self-esteem development after negative life events.
The present study
To better understand how negative life events shape self-esteem development, recent research has turned its focus to individual differences. The present study sought to contribute to this literature by examining individual differences in self-esteem trajectories after negative life events. Using latent class analysis, we identified classes with distinct self-esteem trajectories following negative life events and examined whether specific event- and person-related variables predicted class membership. As a complementary analysis approach, we computed latent growth-curve models to investigate whether these same variables explained individual differences in self-esteem levels and change after negative life events.
We tested four preregistered hypotheses: Individuals who perceive negative life events as more challenging (Hypothesis 1), more impactful (Hypothesis 2), and more social status threatening (Hypothesis 3), as well as individuals with higher levels of neuroticism (Hypothesis 4), exhibit less positive self-esteem changes following negative life events.
Method
Transparency and openness
The present study used data from the Post-Event Changes Study, which was approved by the local ethics committee at Ruhr University Bochum (Protocol Number 702) and preregistered at https://osf.io/hm3rv/. While data from this study have been previously analyzed in three publications (Haehner, Bleidorn et al., 2024; Haehner, Kritzler et al., 2024; Haehner, Sleep et al., 2023), none have examined event-related changes in self-esteem.
The analyses and hypotheses for the present study were preregistered here. Deviations from this preregistration are described in the supplemental materials. All relevant study materials can be retrieved from OSF.
Procedure
The Post-Event Changes Study was a five-wave, 6-month longitudinal online study conducted in Germany in 2021 and 2022. It focused on individuals who had experienced one of the following five negative life events: death of a loved one, separation, job loss, friendship dissolution, or failing an important exam. Participants were recruited via online platforms (including social media, topic-related forums, and mailing lists), flyers in public spaces (e.g., supermarkets and waiting rooms) and by approaching people who are regularly in contact with the target group (e.g., florists and morticians). To participate, individuals first had to register for the study, which involved providing written informed consent, providing an email address and verifying their eligibility. Inclusion criteria were fluency in German, a minimum age of 18 years, and having experienced one of the specified negative life events within the past 5 weeks.
After registration, participants received email invitations to complete surveys at each of the five measurement waves (T1 to T5), which occurred at 0-, 4-, 8-, 16-, and 24-weeks post-registration. Participants were invited to participate in each measurement wave, regardless of their participation in previous waves. To reduce dropouts, voucher raffles with increasing value were conducted after each wave.
At each measurement wave, participants completed assessments on various variables, including Big Five personality traits and self-esteem. During the initial wave (T1), participants further provided demographic information, specified the negative life event they had experienced, and rated its perceived characteristics using the Event Characteristics Questionnaire (Luhmann et al., 2021; see the study-design preregistration for more details).
Participants
A total of 1,673 individuals registered to take part in the Post-Event Changes Study. Participants were excluded from the analysis if their negative life event occurred more than 6 weeks ago. Additionally, data from measurement waves were excluded if participants either gave no or incorrect answers to instructed response items (e.g., “To ensure data quality, please select the response option rather not true”). The final sample size was N T1 = 1,069, N T2 = 759, N T3 = 688, N T4 = 613, N T5 = 577. 74% of the sample identified as female. The mean age of participants was 29.10 years (SD = 9.19). 87% reported having attained higher education (high school diplomas or above) and 12% of the sample reported having a migration background (see Section 1 of the supplementary materials for details).
Measures
Demographic variables (T1)
Following previous work on self-esteem development after negative life events (i.e., Bleidorn et al., 2021; Reitz, Weidmann et al., 2022; Tetzner et al., 2016), we examined five demographic variables as predictors of individual differences in self-esteem trajectories: age (in years), gender (female, male, non-binary), education (lower education, higher education), income (assessed on an 11-point scale) and migration background (yes, no).
Perceived event characteristics (T1)
The Event Characteristics Questionnaire (Luhmann et al., 2021) was used to assess participants’ subjective perception of the experienced negative life event. The questionnaire includes 38 items (e.g., “The event was stressful”) to measure nine event characteristics: challenge, emotional significance, external control, extraordinariness, impact, predictability, social status change, valence, change in world views. All items were rated on a 5-point scale ranging from 1 (not true at all) to 5 (absolutely true). Responses were reverse coded if necessary. Mean scores were calculated for each of the nine event characteristics.
Big Five personality traits (T1)
The Big Five personality traits (extraversion, agreeableness, conscientiousness, openness, and neuroticism) were assessed using the German BFI-2-XS (Rammstedt et al., 2020; Soto & John, 2017), a 15-item questionnaire with three items per trait (e.g., “I am somebody who is full of energy”). Each item was rated on a 5-point scale ranging from 1 (not true at all) to 5 (absolutely true). Responses were reverse coded as needed, and mean scores were calculated for each personality trait.
Self-esteem (T1 to T5)
Self-esteem was assessed using the German version of the Rosenberg Self-Esteem Scale (Collani & Herzberg, 2003; Ferring & Filipp, 1996; Roth et al., 2008) which consists of 10 items (e.g., “I feel that I am a person of worth”). Each item was rated on a 6-point scale ranging from 1 (strongly disagree) to 6 (strongly agree). Responses were reverse coded when necessary, and a mean score was calculated.
Data analysis
Data analysis was conducted in R (Version 4.4.1) and comprised three steps. First, we used latent class growth analysis (LCGA) and latent growth mixture modeling (LGMM) to identify latent classes with distinct self-esteem trajectories following negative life events. Second, we employed stepwise linear regression and multinomial logistic regression to assess how event- and person-related variables predicted individuals’ membership in the identified latent classes. Third, we computed latent growth-curve models (LGCM) to examine whether these same variables predicted self-esteem levels and change after negative life events. For all analyses, continuous variables were standardized using their T1 mean and standard deviation. To reduce risk of false positive findings, a significance level of α = .01 was used. This represents a deviation from the preregistration, where we preregistered to use a level of α = .05, which was incorporated during the revision of this manuscript.
Step 1: Identifying latent self-esteem trajectory classes using LCGA and LGMM
We utilized LCGA and LGMM to examine individual differences in self-esteem trajectories following negative life events, as these approaches identify latent classes with distinct growth trajectories over time (Lennon et al., 2018; Van De Schoot et al., 2017; Wardenaar, 2020). While LCGA assumes that individuals within a latent class share the same trajectory, LGMM extends this by allowing for individual differences in both the intercept and slope within each class.
We used a stepwise approach, following the methodology outlined by Ballering et al. (2022) and adhering to van de Schoot et al.’s (2017) Guidelines for Reporting on Latent Trajectory Studies (GRoLTS) and Lennon et al.’s (2018) framework for latent class trajectory modeling. This involved fitting a series of progressively more complex models and ultimately selecting the one with the best fit. First, LCGA models with fixed class-specific intercepts and slopes were estimated. Next, LGMM models with random class-specific intercepts were fitted. Finally, LGMM models with both random class-specific intercepts and slopes were estimated. Self-esteem trajectories were modeled using time since event occurrence (in months) as the metric time variable. To account for potential non-linear change (Orth et al., 2018), all models were estimated separately with linear and quadratic time terms. Following Ballering et al. (2022) and Lennon et al. (2018), all models were fitted with 1–7 latent classes. To reduce the risk of local maxima and ensure convergence to the global solution, each model was initialized with at least 50 random start values from the one-class model and allowed up to 100 iterations (Ballering et al., 2022; Jung & Wickrama, 2008). Additionally, the intercept and slope variances in the LGMM models were allowed to vary across classes. All models were fitted using the hlme function from the lcmm package (Proust-Lima et al., 2017). Missing data was handled using full-information maximum likelihood (FIML).
Selection of the best-fitting model was based on four preregistered criteria: (1) low BIC (Ballering et al., 2022; Van De Schoot et al., 2017); (2) entropy value greater than .80 (Ballering et al., 2022); (3) fewer classes but larger memberships (Ballering et al., 2022; Infurna & Grimm, 2018; Jung & Wickrama, 2008); and (4) theoretically plausible classes consistent with previous research that accurately represent the data (Ballering et al., 2022; Jung & Wickrama, 2008; Ram & Grimm, 2009). The best-fitting model, as determined by these criteria, was then fully fitted with a minimum of 500 random start values and a maximum of 1000 iterations. Finally, a categorical class membership variable was created, assigning participants to a latent class based on their highest posterior class probabilities from the best-fitting model.
Step 2: Predicting latent class membership from event- and person-related variables
We used stepwise multiple linear regression and multinomial logistic regression to assess whether and how event type, event perception, Big Five personality traits, and demographic variables predicted membership in the latent classes identified in the best-fitting model.
Stepwise multiple linear regression models were estimated separately for each latent class, using participants posterior class probabilities as the dependent variable. First, a null model was compared to four regression models, each incorporating a single variable cluster as predictor (i.e., event type only, event perception only, Big Five personality traits only, or demographic variables only). Second, a full model incorporating all variable clusters as predictors was estimated and compared to four models, each of which excluded one variable cluster (i.e., event type excluded, event perception excluded, Big Five personality traits excluded, or demographic variables excluded). This stepwise approach allowed us to evaluate how much additional variance—measured in incremental R 2 —each variable cluster could explain. Stepwise multiple linear regressions were performed using the lm function available in the R base package stats (R Core Team, 2024).
Multinomial logistic regression models were estimated using the categorical variable representing latent class membership as the dependent variable. The latent class with the least amount of change in self-esteem served as the reference category. Similar to the stepwise multiple linear regression models, four multinomial logistic regression models, each incorporating a single variable cluster as predictor, were compared to a null model using McFadden’s Pseudo-R2. Then, a full model incorporating all variable clusters as predictors was estimated and compared to four models, each of which excluded one variable cluster. Following Ballering et al. (2022), participants with low posterior probabilities (<.50) for all classes were excluded from the analysis. 1 Odds ratios were calculated for each variable, reflecting the effect of the variable on the odds of being in a particular class compared to the reference class. All multinomial logistic regressions were performed using the multinom function of the R package nnet (Venables & Ripley, 2002).
Step 3: Predicting self-esteem levels and change from event- and person-related variables using LGCM (not preregistered)
Although latent class analysis fitted well to our overall research goal, it comes with certain disadvantages such as reduced power and categorizing continuous between-person differences. We thus included LGCMs as a complementary, non-preregistered analysis approach during the manuscript revision. LGCMs allowed us to examine whether the same event- and person-related variables from Step 2 predicted self-esteem levels and change following negative life events. Furthermore, they enabled the investigation of whether time since event occurrence predicted individual differences in self-esteem development after negative events.
We estimated five LGCMs, each including one variable cluster (Big Five, event perception, event type, demographics, and event timing) and compared them to a baseline model without covariates. Additionally, we estimated a full model with all covariates and compared it to five reduced LGCMs, each omitting one variable cluster or the event time variable. This approach enabled the evaluation of how much additional variance—measured in incremental R 2 —each variable cluster explained in both self-esteem levels and changes following negative life events.
Each LGCM comprised a latent intercept and a latent slope. Loadings for the latent intercepts were fixed to 1 for each measurement wave. Latent slopes were specified with a fixed loading of 0 for the first measurement wave (T1), 1 for the second (T2), 2 for the third (T3), 4 for the fourth (T4), and 6 for the fifth wave (T5), corresponding to the temporal distance in months between assessments. Covariates were regressed onto the latent intercept and slope factors. All LGCMs were estimated using the growth function from the lavaan package (Rosseel, 2012). Missing data was handled using FIML.
Results
This study is accompanied by an HTML document with supplementary materials, available here. Descriptive statistics of the examined variables are summarized in Section 1 of the supplementary materials.
Research question 1: Are there classes of individuals with distinct self-esteem trajectories following negative life events?
We fitted a series of progressively more complex linear and quadratic LCGA and LGMM models to identify the one with the best fit. Fit indices for all estimated LCGA and LGMM are summarized in Section 2 of the supplementary materials.
Fixed effects from the longitudinal linear four-class LCGA model.
Note. The table summarizes fixed effects (b), their standard errors (SE), and p-values (p) in the longitudinal four-class LCGA model. The fixed effects represent each class’s average intercept and slope, reflecting initial self-esteem and monthly change in T1-standardized units.
Class-specific mean-predicted self-esteem trajectories. Note. The figure illustrates class-specific mean-predicted self-esteem trajectories in the best-fitting four-class linear LCGA model. Slopes (b) represent changes in self-esteem in T1-standardized units per month, and p-values (p) indicate statistical significance.
Class-specific self-esteem trajectories of 15 random individuals. Note. The figure shows self-esteem trajectories for 15 randomly selected individuals from each of the four classes identified in the best−fitting linear LCGA model. The colored prominent lines (solid, long-dashed, two-dashed or dot-dashed) represent the mean-predicted self-esteem trajectory for each latent class.
Research question 2: Which variables explain individual differences in self-esteem changes in response to negative life events?
Building on the previously identified linear four-class LCGA model, we used multiple linear and multinomial regression analysis to investigate whether event- and person-related variables predicted membership in the four self-esteem trajectory classes. To assess the contribution of each variable cluster—event type, perceived event characteristics, Big Five personality traits and demographic variables—we applied a stepwise approach.
Stepwise multiple linear regression analysis
R2-values from the stepwise multiple linear regression models.
Note. R2-values from the stepwise multiple regression models predicting the probability for membership in each latent self-esteem trajectory class. Rows correspond to variable clusters (event type, event perception, Big Five personality traits, and demographic variables). Columns represent different model configurations: This predictor only contains regression models with only the specified variable cluster; All except this predictor includes regression models with all variable clusters except the specified one; All predictors contains the regression model with all variable clusters simultaneously.
Event-related and person-related variables predicting class membership. Note. The figure illustrates regression coefficients of the multiple linear regression models predicting the posterior probability for membership in each class, using all variable clusters as predictors. Coefficients for each variable are represented by squares, with significant coefficients (α = .01) shown in black. Error bars indicate 99% confidence intervals (CI). Event type: reference category “End of a romantic relationship”; gender: reference category “female”; education: reference category “lower education”; migration: reference category “no”.
Together, all variable clusters explained 22% of the variance in the probability of belonging to the Very Low-Stable Class, 14% to the Low-Increasing Class, 4% to the Moderate-Increasing Class, and 36% of belonging to the High-Increasing Class. Among all variable clusters, the Big Five personality traits accounted for the highest amount of variance in the probability of class membership (1% ≤ R 2 ≤ 31%), followed by perceived event characteristics (1% ≤ R 2 ≤ 8%), whereas demographic variables (2% ≤ R 2 ≤ 5%) and event type (R 2 = 1%) accounted for considerably less variance.
A closer look at specific associations between person- and event-related variables and class membership probabilities in the models including all variable clusters simultaneously revealed no significant effects for event type. Among the demographic variables only one significant effect emerged: individuals were more likely to be assigned to the High-Increasing Class if they were older (β = 0.05, p < .001). Similarly, with the exception that individuals were more likely to belong to the High-Increasing Class, if they perceived negative events as less social status threatening (β = −0.06, p < .001), no significant effects were found for perceived event characteristics. Thus, neither Hypothesis 1—that individuals who perceive negative life events as more challenging exhibit less positive self-esteem changes—nor Hypothesis 2—that those who perceive negative events as more impactful exhibit fewer positive changes—were supported by the multiple regression models including all variable clusters. Only Hypothesis 3—that individuals who perceive negative life events as more socially status threatening exhibit less positive self-esteem changes—was partially supported as individuals who perceived negative life events as less social status threatening were more likely to be assigned to the High-Increasing Class.
For the Big Five personality traits, we found that individuals were more likely to belong to the Very Low-Stable Class if they had higher levels of neuroticism (β = 0.09, p < .001) and lower levels of extraversion (β = −0.04, p < .001). Likewise, individuals were more likely to belong to the Low-Increasing Class, if they had higher levels of neuroticism (β = 0.09, p < .001) and lower levels of extraversion (β = −0.04, p = .001). Finally, individuals were more likely to be assigned to the High-Increasing Class, if they had lower levels of neuroticism (β = −0.14, p < .001) and higher levels of extraversion (β = 0.07, p < .001), agreeableness (β = 0.03, p = .009), and conscientiousness (β = 0.04, p = .001). Together, these results largely support Hypothesis 4—that individuals high in neuroticism exhibit less positive self-esteem changes.
In summary, the Big Five personality traits, particularly neuroticism and extraversion, emerged as the most important predictors of membership in a specific self-esteem trajectory class. They explained a substantial amount of the variance in posterior class probabilities, with neuroticism significantly predicting membership in three classes. While perceived event characteristics also seem to be relevant predictors given the variance explained in class probabilities, their effects were smaller than those of the Big Five traits, with only social status change emerging as significant in the High-Increasing Class. The associations between posterior class probabilities and event type and demographic variables were less clear.
Stepwise multinomial regression analysis
In addition to the multiple linear regressions, we applied stepwise multinomial regressions to assess whether event- and person-related variables predicted the odds of belonging to a specific self-esteem trajectory class. For this analysis, we used the Very Low-Stable Class as the reference category and evaluated the contribution of each variable cluster using McFadden’s Pseudo-R 2 . The Pseudo-R 2 -values and odds ratios for all stepwise multinomial regressions are provided in Section 4 of the supplementary materials.
Overall, the stepwise multinomial regressions yielded results consistent with those of the multiple linear regressions. Again, the Big Five personality traits contributed the most to explaining class membership (McFadden’s-R 2 = .21), followed by perceived event characteristics (McFadden’s-R 2 = .05), demographic variables (McFadden’s-R 2 = .03) and event type (McFadden’s-R 2 = .01). Together, all variable clusters achieved a McFadden’s-R 2 of .25.
The odds ratios from the multinomial regression models including all variable clusters simultaneously as predictors of class membership largely mirrored the findings observed in the corresponding multiple linear regression models, with only three differences, leading to the same overall conclusions. First, although individuals with lower levels of extraversion were more likely to belong to the Low-Increasing Class in the multiple regression analysis (β = −0.04, p = .001), extraversion was not significantly associated with higher odds of membership in this class compared to the Very Low-Stable Class in the multinomial regression. Second, while perceiving negative life events as more world view changing was associated with higher odds of membership in the High-Increasing Class (OR = 1.81, 99% CI = [1.11, 2.95]) compared to the Very Low-Stable Class in the multinomial regression, it was not significantly associated with class membership in the multiple regression analysis. Third, while individuals with higher levels of agreeableness were more likely to belong to the High-Increasing Class in the multiple regression analysis (β = −0.03, p = .009), agreeableness was not significantly associated with the odds of membership in any class relative to the Very Low Stable Class in the multinomial regression.
LGCMs
Effects of the LGCM including all variable clusters and event timing as covariates.
Note. The table summarizes standardized regression coefficients (β), standard errors (SE), p-values (p) and 99% confidence intervals (CI (99%)) from the LGCM including all variable clusters and the event timing variable as covariates. Significant coefficients (α = .01) are bold. Event type: reference category “End of a romantic relationship”; gender: reference category “female,” education: reference category “lower education” and migration: reference category “no”.
Together, all variable clusters explained 57% of the variance in the intercept and 20% in the slope. Regarding the intercept, the Big Five personality traits accounted for the highest amount of variance (R 2 = 51%), followed by perceived event characteristics (R 2 = 14%), demographic variables (R 2 = 7%) and event type (R 2 = 1%). In contrast, the slope variance was best explained by perceived event characteristics (R 2 = 9%), followed by the Big Five traits (R 2 = 6%), event type (R 2 = 7%) and demographic variables (R 2 = 1%). Event timing explained less than 1% of the variance in both the intercept and slope, indicating that the temporal distance between study enrollment and event occurrence had negligible influence on self-esteem development. 2
A closer inspection of the LGCM that included all variable clusters simultaneously revealed that the intercept was not statistically significant due to our standardization on the T1 mean and SD. However, its variance was significant (Var Intercept = 0.36, p < .001), indicating that individuals significantly differed in their initial self-esteem levels following negative life events. Furthermore, social status change (β = −0.15, p < .001) and neuroticism (β = −0.44, p < .001) were associated with lower initial self-esteem levels, whereas extraversion (β = 0.19, p < .001), conscientiousness (β = 0.10, p < .001), agreeableness (β = 0.07, p = .005) and age (β = 0.09, p < .001) positively predicted the intercept.
The average slope was not statistically significant, indicating no meaningful changes in self-esteem on average. However, the significant variance of the slope (VarSlope = 0.004, p < .001) indicates individual differences in self-esteem change over time. Among all covariates, only friendship dissolution (β = −0.04, p = .003) and job loss (β = −0.07, p < .001) significantly predicted these individual differences.
Thus, the findings from the LGCMs aligned broadly with those of the four-class LCGA model. However, while the LCGA model identified three classes exhibiting significant self-esteem increases after negative life events, LGCMs showed no significant change. Moreover, the LCGMs extended the regression analyses by showing that the variable clusters primarily predicted the self-esteem intercept, suggesting that associations between these variables and class membership may reflect differences in initial self-esteem levels, not change in the subsequent months.
Discussion
Research indicates that individuals differ in their self-esteem trajectories after negative life events. In the present study, we examined these individual differences by identifying classes with distinct self-esteem trajectories and examining event- and person-related variables associated with class membership. Using LCGA, we identified four distinct self-esteem trajectory classes: three classes exhibiting slight increases in self-esteem following negative life events, while one class showed no significant change. Furthermore, we found that the Big Five personality traits, especially neuroticism and extraversion, were the most important variables in explaining membership in the four self-esteem trajectory classes followed by perceived event characteristics, particularly social status change. Additional LGCMs revealed that the event- and person-related variables explained more variance in self-esteem levels than in changes following negative life events.
Are there classes of individuals with distinct self-esteem trajectories following negative life events?
Using LCGA, we identified four classes with distinct self-esteem trajectories after negative life events. The Moderate-Increasing Class, with 36% of participants assigned, was characterized by slightly above-average self-esteem shortly after negative life events and a modest upward trend in the subsequent months. The Low-Increasing Class, comprising 27% of participants, was marked by below-average self-esteem after experiencing negative life events, followed by a slight increase thereafter. The High-Increasing Class, with 26% of participants assigned, showed noticeably higher-than-average self-esteem after negative life events, with the most pronounced increases in self-esteem in the following months. Lastly, the Very Low-Stable Class, comprising 11% of participants, exhibited markedly low self-esteem after negative life events, with no significant changes over time. Thus, on average, most individuals experienced a self-esteem increase of approximately 0.18 SD over the 6-month study period, which corresponds to a small effect. Compared to event-related changes in other variables, these changes in self-esteem were larger than corresponding changes in the Big Five personality traits (Haehner, Rakhshani et al., 2023; Haehner, Sleep et al., 2023), but smaller than event-related changes in well-being (Haehner, Kritzler et al., 2024; Reitz, Luhmann et al., 2022).
With the exception of the Very Low-Stable Class, our results align with the typical pattern of self-esteem development expected following negative life events. Specifically, existing research found that self-esteem reaches a low point shortly after negative life events and then increases in the subsequent months, reflecting a recovery process over time (Luciano & Orth, 2017; Mangelsdorf et al., 2019; Reitz, Luhmann et al., 2022). Interestingly, despite differences in self-esteem levels across classes shortly after experiencing negative life events, the increases in self-esteem over the following months were highly similar across the Low-Increasing, Moderate-Increasing and High-Increasing Classes. One explanation for this pattern could be a universal recovery process, with habituation and adaptation processes occurring similarly across individuals, regardless of stable self-esteem levels or the initial reactions to the event.
However, the self-esteem trajectory following negative life events of the Very Low-Stable Class diverged from the other classes. The Very Low-Stable Class not only showed the lowest self-esteem levels shortly after negative life events but also exhibited no significant increases in the months thereafter, suggesting a disruption in the typical recovery process expected following negative life events (e.g., Haehner, Kritzler et al., 2024; Luciano & Orth, 2017; Mangelsdorf et al., 2019; Reitz, Luhmann et al., 2022). One explanation for the persistently low self-esteem levels of individuals assigned to the Very Low-Stable Class could be the existence of preexisting vulnerabilities. According to common diathesis-stress models (e.g., Cohen et al., 2019; Ingram & Luxton, 2005; Monroe & Simons, 1991; Rosenthal, 1963), vulnerabilities such as negative self-schemas or dysfunctional attributional styles interact with stressful negative life events, increasing individuals’ sensitivity to their impact. Specifically, negative life events may activate and reinforce negative self-schemas, leading to low self-esteem, hindering its recovery, and potentially contributing to the development of psychopathology (Beck, 1967; Beck & Bredemeier, 2016). Similarly, a dysfunctional attributional style, which attributes negative events to internal, stable and global causes, may contribute to a self-reinforcing cycle of low self-esteem (Abramson et al., 1989). For instance, after experiencing a job loss, individuals with preexisting vulnerabilities like negative self-schemas may interpret the event as proof of their unworthiness, reducing self-esteem and impeding its recovery.
Taken together, our findings indicate that the latent self-esteem trajectory classes primarily differed in initial self-esteem following negative life events, rather than in how-self-esteem changed over time. This is further supported by the small slope variance relative to the larger intercept variance observed in the LGCMs. Thus, while individuals showed notable differences in self-esteem levels shortly after negative life events, their self-esteem development in the following months was relatively similar, pointing to a universal recovery process. However, this recovery process may be disrupted for individuals with preexisting vulnerabilities and very low levels of self-esteem, as observed in the Very Low-Stable Class. Targeted support may be needed to address the specific challenges these individuals face. An important first step in this regard is identifying variables that help predict who is at risk for such a negative self-esteem trajectory.
Which variables explain individual differences in self-esteem changes in response to negative life events?
To advance our understanding of why individuals differ in their self-esteem trajectories, we investigated whether event- and person-related variables predicted membership in the four identified self-esteem trajectory classes. Additionally, we computed LGCMs to examine whether these same variables were associated with self-esteem levels and changes after negative life events. We discuss these findings for each variable cluster, sorted by explained variance across all analyses.
Big Five personality traits
Among all variable clusters, class membership was most strongly predicted by the Big Five personality traits, which accounted for up to 31% of the variance. The Big Five traits also explained 52% of the variance in the intercept and 6% in the slope in the corresponding LGCM. These findings align with previous research linking the Big Five to self-esteem and its development (e.g., Bien et al., 2024; Fetvadjiev & He, 2019; Weidmann et al., 2018).
Neuroticism emerged as particularly relevant for membership in self-esteem trajectory classes. Consistent with Hypothesis 4, higher neuroticism was associated with membership in classes characterized by less positive self-esteem changes, such as the Very Low-Stable Class. Neuroticism is defined as a predisposition to experience negative emotions (Costa & McCrae, 1992) and is linked to heightened emotional reactivity (Hisler et al., 2020; Mader et al., 2023; Zautra et al., 2005) and maladaptive emotion regulation (Barańczuk, 2019). These tendencies could negatively influence psychological states such as affect and cognitions in response to negative life events, which are thought to play a key role in self-esteem development (Reitz, 2022). In this context, and in line with the idea that individuals in the Very Low-Stable Class may exhibit preexisting vulnerabilities, neuroticism might represent one such vulnerability. However, the LGCMs showed that neuroticism was significantly associated with self-esteem levels shortly after negative life events, but not with changes over the subsequent 6 months. This indicates that neuroticism is not related to longer-term self-esteem development in the present research. Nevertheless, its association with self-esteem levels following negative events suggests that neuroticism is relevant for individuals’ immediate reactions to negative life events. However, this interpretation cannot be directly tested, as our study lacks data on self-esteem and the Big Five prior to the negative event. An alternative explanation is that neuroticism is generally related to lower self-esteem, independent of the occurrence of negative events (e.g., Amirazodi & Amirazodi, 2011).
Beyond neuroticism, extraversion also exhibited notable associations with class membership, being negatively related to membership in the Very Low-Stable and the Low-Increasing Classes and positively related to membership in the High-Increasing Class. This may be understood in light of research linking extraversion to more positive and less negative affect (e.g., Anglim et al., 2020; Costa & McCrae, 1980) and more adaptive emotion regulation (Barańczuk, 2019), characteristics that could contribute to more favorable self-esteem responses following negative life events. Importantly, similar to neuroticism, extraversion was associated only with self-esteem levels shortly after negative events, but not with change over time. As discussed above, the observed associations involving extraversion could reflect either its link to individuals’ initial self-esteem response to a negative event or more general associations with self-esteem.
Notably, research has shown that the Big Five personality traits can itself exhibit short-term fluctuations, particularly following negative life events (Haehner, Bleidorn et al., 2024; Haehner, Rakhshani et al., 2023). These fluctuations could therefore also affect how the Big Five traits relate to self-esteem levels and change following negative life events.
Perceived event characteristics
In addition to the Big Five personality traits, membership in the self-esteem trajectory classes was also explained by perceived event characteristics, which accounted for up to 8% of the variance in class membership. Furthermore, in the corresponding LGCM, perceived event characteristics explained 14% of the variance in the intercept and 9% in the slope.
However, our analyses did not support Hypothesis 1 (individuals who perceive negative events as more challenging exhibit less positive self-esteem changes) or Hypothesis 2 (individuals who perceive negative life events as more impactful exhibit less positive self-esteem changes), suggesting that associations found between event perception and well-being development do not seem to generalize to self-esteem (Haehner, Kritzler et al., 2024; Haehner, Pfeifer et al., 2023; Luhmann et al., 2021). While perceiving an event as impactful and challenging may temporarily disrupt well-being, such as decreasing life satisfaction, this may not directly threaten self-esteem. For example, an individual may feel challenged by a negative event like the death of a loved one but not necessarily experience a change in self-esteem if they feel confident in their ability to cope with it.
Conversely, we found support for Hypothesis 3 (individuals who perceive negative life events as more social status threatening exhibit less positive self-esteem changes). Perceiving a negative event as more social status threatening was associated with a lower probability of being assigned to the High-Increasing Class. Additionally, social status change was negatively associated with self-esteem levels immediately after negative life events in the corresponding LGCMs. These results are consistent with Hierometer Theory (Mahadevan et al., 2016), which conceptualizes self-esteem as a measure of social status. Specifically, perceiving a negative event like a job loss as more social status threatening is expected to lead to stronger reductions in self-esteem, making it less likely for individuals to be assigned to the High-Increasing Class, characterized by above-average self-esteem levels shortly after the event.
Demographic variables, event type and event timing
Compared to the Big Five personality traits and perceived event characteristics, demographic variables (up to 5%) and event type (1%) explained only little variance in self-esteem trajectory-class membership. Similarly, demographic variables accounted for 7% of the variance in the intercept, and event type accounted for 1%, in the corresponding LGCMs. However, event type proved relevant for self-esteem changes over time as it explained 10% in slope variance.
Two findings stood out when considering demographic variables and event type. First, age was positively associated with both membership in the High-Increasing Class and the intercept in the LCGMs, indicating that older individuals tend to exhibit higher self-esteem following negative life events. This may be explained by greater emotional stability and regulation with age (e.g., Bleidorn & Hopwood, 2024; Charles & Carstensen, 2010; Heckhausen & Schulz, 1995), which could help protect self-esteem against negative events. However, the higher self-esteem levels with increasing age could also reflect normative increases in adulthood, rather than effects of negative life events (Orth et al., 2018). Second, the end of a friendship and job loss negatively predicted the slope in the LGCM including all variable clusters, suggesting that experiencing these events contributes to less positive self-esteem changes compared to the end of a romantic relationship, which served as the reference category. One possible interpretation is that job loss may undermine mastery experiences and likely also perceived social status in the months that follow, making it difficult to recover self-esteem after such an event (e.g., Crocker & Wolfe, 2001; Mahadevan et al., 2016; Wojciszke et al., 2011). However, the effects of the end of a friendship and job loss on the slope were small, and both events did not emerge as significant predictors of class membership in the regression analyses.
Summary
Taken together, our study highlights the importance of both event- and person-related variables in explaining individual differences in self-esteem development following negative life events. Specifically, neuroticism, extraversion, and perceived social status changes exhibited notable associations with both trajectory-class membership and self-esteem levels following negative life events. Thus, these variables may be interpreted as potential vulnerabilities or resources for self-esteem in the context of negative life events. From a health services perspective, this information is valuable not only for identifying those most affected by negative life events, but also for providing them with early support to help mitigate the adverse effects of such events.
Limitations and future directions
The present study aimed to investigate individual differences in self-esteem development following negative life events and to identify variables explaining these differences. However, several limitations highlight the need for further research on this topic.
First, our sample was predominantly female, highly educated, and from a Western democratic country. Therefore, the findings may not be directly applicable to other sociodemographic groups or cultural settings (Henrich et al., 2010; Muthukrishna et al., 2020). Future research should address this limitation by including more diverse samples from multicultural contexts.
Second, participants for the Post-Event Changes study were recruited after they had experienced a negative life event. Therefore, no data on self-esteem, as well as event- and person-related variables prior to the occurrence of the event, were available. This has implications for the conclusions that can be drawn based on this dataset. Specifically, individual differences in self-esteem observed at the first measurement wave confound preexisting self-esteem differences and the initial reaction to the event. Furthermore, without an appropriate control group, it is impossible to examine whether the examined event-related and person-related variables are only relevant in the context of negative events or if they are generally related to self-esteem development over time. Finally, without pre-event assessments, we were not able to examine potential selection effects for life events. For example, low self-esteem has been shown to predict the end of a romantic relationship (Luciano & Orth, 2017) and job loss (Reitz, Luhmann et al., 2022). Testing these possibilities requires future research that includes data both prior to and following negative life events, as well as the inclusion of control groups. This is particularly important for investigating causal relationships (Grosz et al., 2020; Rohrer, 2018) and regarding potential preexisting vulnerabilities in individuals with negative self-esteem development following negative life events. In light of these considerations, our findings should thus be interpreted as being in the context of negative events or occurring after negative events but not necessarily as being driven by negative events.
Third, the present study used data based on self-report measures, which can be biased due to measurement artifacts such as social desirability or response styles (e.g., Paulhus & Vazire, 2007) To gain a more precise understanding of individual differences in self-esteem changes following negative life events, other assessment methods, such as peer-reports, should be incorporated in future research, as findings may differ between these methods.
Fourth, we examined changes in self-esteem across five measurement occasions within 6 months after a negative life event, as research indicates that self-esteem changes might be more pronounced in the early post-event period (Luciano & Orth, 2017; Tetzner et al., 2016). However, we were unable to investigate longer-term changes in self-esteem beyond this period. Future research should adopt a perspective that considers both short- and long-term development to better understand self-esteem changes following negative life events.
Fifth, as is common in longitudinal research, some participants dropped out during the Post-Event Changes study. Attrition analyses (see Section 1 in the supplementary materials) indicated that dropout was partially related to characteristics like neuroticism or conscientiousness. For example, individuals who struggle more with negative life events or those with lower conscientiousness may be more likely to drop out, potentially leading to biased self-esteem trajectories. Additionally, other unobserved factors may influence dropout, limiting the generalizability of our findings.
Conclusion
Individuals differ in their self-esteem development following negative life events. In the present study, we sought to better understand these individual differences by identifying latent classes with distinct self-esteem trajectories after negative events. Additionally, we examined whether event- and person-related variables predicted membership within these classes, as well as self-esteem levels and changes following negative life events. Our results revealed four distinct self-esteem trajectory classes. Three classes showed comparable increases in self-esteem in the months after negative life events, indicating a recovery process. Conversely, the class with the lowest initial self-esteem levels showed no significant change, suggesting that the recovery process might be disrupted for individuals in this class. These individuals may be most affected by negative life events and may require additional support to initiate recovery. Membership in the four self-esteem trajectory classes was most strongly predicted by the Big Five personality traits, followed by perceived characteristics of the event. Neuroticism followed by extraversion and social status change emerged as the strongest predictors of both class membership and self-esteem levels following negative life events. These variables may be interpreted as potential vulnerabilities or resources in the context of self-esteem responses to negative events. Our findings have important implications for health services, self-esteem development theories and future research on adaptation to negative life events.
Footnotes
Acknowledgments
The authors thank Sophia Salzburg, Melina Sostmann, Debora Brickau, Janine Kaltbeitzer, Maren Koß, Lennart Martens, and Mahmoud Aldalati for their help with data collection for this article.
Author contributions
Lukas Schellenberg: Conceptualization; Methodology; Formal analysis; Writing – original draft; Visualization; Project Administration. Marco Joe Altorfer: Conceptualization; Methodology; Formal analysis; Writing – original draft. Peter Haehner: Conceptualization; Methodology; Investigation; Supervision; Writing – review & editing.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Open science statement
Ethical considerations
This study based on data from the Post-Event Changes Study. Data collection for the Post-Event Changes Study was approved by the local ethics committee at Ruhr University Bochum (Protocol Number 702).
Consent to participate
All respondents of the Post-Event Changes Study provided informed consent before participating in the study.
Supplemental Material
The present study was preregistered at https://osf.io/8x5b9?view_only=3092920b4e0d4214a17547453e3ba1e0. All relevant materials, including the dataset, R scripts, codebook, and supplementary materials, are available at 
.