Abstract
Dynamic personality models highlight the importance of malleable personality states and social interactions for personality trait development. Integrating both ideas, this research examined how interactions with family members, friends, teachers, and others relate to, first, personality state expression, and, second, personality trait development in adolescence. In two separate German samples (Noverall = 445, Mage = 16.8; 80.9 % female), we combined data on personality states during daily social interactions (4268 reports) with self-reported personality traits measured over 6–12 months. Focusing on personality states, multilevel models indicated greater within-person variability compared to between-person differences. Conditional models revealed that adolescents experienced higher state extraversion, openness to new experiences, agreeableness, and conscientiousness during interactions with friends compared to interactions with family. In interactions with teachers and others, they reported higher state neuroticism, openness to new experiences, and conscientiousness. Considering adolescents’ personality traits over time, latent growth models showed that neuroticism declined on average. Apart from that, most traits showed high rank-order stabilities, no mean-level change, no substantial variance in change, and there was little evidence for bottom-up effects of social interaction frequency on personality changes. We conclude with a discussion of future directions in studying the role of social interactions in personality trait development.
Plain language summary
Adolescence is an important life phase to develop a better sense of oneself, that is, one’s personality. This research was interested in understanding how daily interactions with different types of people (family, friends, teachers, and others) are linked to adolescents’ momentary expressions of personality, called personality states. In a second step, we explored whether these social interactions with different people predict long-term changes in more stable aspects of personality, called personality traits. We collected data from 445 German adolescents (average age 16.8) in two ways: First, we tracked adolescents up to five times daily for one week to understand if and how their personality states varied across different social interactions. Second, we followed up with the same adolescents over 6–12 months to monitor changes in their personality traits. The results showed that adolescents acted more extraverted (i.e., sociable), agreeable (i.e., kind), open (i.e., interested in new experiences), and conscientious (i.e., reliable) during interactions with friends. In addition, they acted more neurotic (i.e., anxious), open, and conscientious during interactions with teachers and others. Over time, most adolescents became more emotionally stable, but did not change in terms of the remaining personality traits. In addition, there was no robust link between more frequent social interactions with specific interaction partners and personality trait change. We conclude that more research is required to clarify whether and how social interaction partners contribute to personality development in the long run.
Keywords
Introduction
During adolescence, individuals learn more about themselves and gradually form a clearer view of who they are (Erikson, 1959). Adolescence is thus a key phase for the development of a person’s personality, defined as relatively stable interindividual differences in behavior, thoughts, and feelings (Roberts & Del Vecchio, 2000). The age-specific patterns of adolescent personality development have been documented in a large number of studies, largely focusing on mean-level changes and rank-order stabilities of the Big Five traits (e.g., Borghuis et al., 2017; van den Akker et al., 2014). Despite this rich evidence, little is known about the mechanisms underlying personality development in adolescence.
To address this research gap, we integrate two central perspectives from dynamic personality models: First, the states perspective revolves around personality states, which fluctuate on a momentary level and form the basis for the manifestation and development of more stable personality traits in a bottom-up fashion (Fleeson & Jayawickreme, 2015; Wrzus & Roberts, 2017). That is, if certain, momentary thoughts, feelings, and behaviors are repeated for a sufficiently long period of time, they can lead to lasting personality changes. Second, the socialization perspective centers on the role of different types of social interaction partners that promote personality development of an individual by invoking certain (age-graded) social roles (Roberts & Wood, 2006). For example, adolescents increasingly spend time with their friends, which is assumed to shape their personalities (e.g., van Zalk et al., 2020). Combining these two ideas on momentary processes and formative social experiences and applying them to the developmental context of adolescence, the current paper examines two research questions: First, how are daily social interactions with different types of (age-relevant) interaction partners (i.e., family, friends, teachers, and others) related to differences in adolescents’ momentary Big Five personality states? Second, does the frequency of social interactions with certain types of interaction partners predict long-term changes in adolescents’ corresponding Big Five personality traits?
Personality development in adolescence
Potentially reflecting the multitude of biological and social changes that adolescence face (Denissen et al., 2013; Steinberg, 2005), extant findings on Big Five personality traits emphasize that adolescence is a key phase for personality development. First, meta-analytic research indicates that as an average trend from childhood to adulthood, people in Western industrialized countries become more emotionally stable, agreeable, and conscientious (Bleidorn et al., 2022; Roberts et al., 2006). These mean-level trends are often referred to as maturation. 1 However, they do not fully capture the more nuanced mean-level development during adolescence, which is also characterized by dips in emotional stability, agreeableness, and conscientiousness (disruption; Denissen et al., 2013; van den Akker et al., 2014).
Second, research shows that whereas rank-order stabilities of individual differences in personality tend to increase from childhood to adulthood, adolescence is also a period characterized by relatively low stability (Bleidorn et al., 2022; Roberts & Del Vecchio, 2000). Altogether, previous findings indicate that adolescents vary considerably in their personality trajectories, highlighting the need for research on the factors predicting individual differences in personality development.
Mechanisms underlying personality development: Integration of two theoretical perspectives
To date, the mechanisms underlying personality development are not well understood (for an overview of the current state of the field, see Bleidorn et al., 2021). From the states perspective, the basic idea is that malleable personality states form the building blocks of more stable personality traits, as personality traits are understood as individual differences in the repeated, relatively consistent expression of personality states across time and situations (Fleeson & Jayawickreme, 2015). It is assumed that if the expression of personality states systematically and frequently deviates from one’s traits across an extended period, this can translate to long-term changes in corresponding personality traits (Geukes et al., 2018; Wrzus & Roberts, 2017). The transition from state fluctuations to the internalization of enduring trait changes is thought to occur through associative or reflective processes that follow from self-observation or from feedback of others. For example, if people behave more extraverted than usual across several occasions, these short-term changes might be integrated in their self-view and translate into enduring increases in trait extraversion over time. Whereas existing theories make no predictions regarding the specific time frame in which bottom-up effects manifest (i.e., how state changes translate into trait changes; Hopwood et al., 2022), empirical research using adult samples provides support for naturally occurring bottom-up effects over a period of 2–6 years (Quintus et al., 2021; van Zalk et al., 2020; Wrzus et al., 2021). In addition, volitional personality change intervention studies targeting daily behaviors of adults observed bottom-up effects across 3 months (Hudson et al., 2019; Stieger et al., 2021). Thus, there is evidence for bottom-up effects predicting adult personality development both over longer periods of time and within narrower time frames. However, it remains largely unexplored whether the same processes apply to adolescent personality development and under which specific circumstances bottom-up effects can be expected.
Further specifying the factors that drive personality development in adolescence, different theories point to the importance of social interaction partners: First, according to the dynamic-interactional paradigm, personality traits develop with and are molded by social relationships (Magnusson, 1990; Neyer & Asendorpf, 2001). Second, adding a developmental nuance to this general paradigm, the social investment principle (Roberts & Wood, 2006) emphasizes the accomplishment of age-graded social roles as key to normative personality development. We integrate these two theoretical ideas under the socialization perspective, suggesting that different types of social interaction partners not only represent distinct drivers of personality development, but also that these socialization effects systematically differ between age groups. Younger children usually spend much time with their parents or other family members (e.g., siblings or grandparents; Laursen & Bukowski, 1997). Accordingly, family members are one of the most important types of social interaction partners in this specific age group. However, as one central developmental task, adolescents become increasingly independent from parents and build meaningful relationships outside the family context (Grob & Jaschinski, 2003; Havighurst, 1948). That is, although family and parents remain important relationship partners, adolescents shift their social focus to relationships with close friends, including first romantic partners (Kahn & Antonucci, 1980; Laursen & Bukowski, 1997; Rubin et al., 2006; Wagner et al., 2014). Finally, school is a key developmental context of adolescence, where individuals spend large parts of their time and receive formal education (Israel et al., 2023; Rutter et al., 1979). As such, teachers represent a third central type of interaction partner in this age group. In sum, age-graded social roles in adolescence include separating oneself from family, behaving according to the norms displayed among friends, and fulfilling teacher expectations.
Integrating the states perspective and the socialization perspective, daily social interactions with different interaction partners are regarded as key cues that systematically trigger personality states (see also Geukes et al., 2018). Along these lines, we argue that momentary expressions of personality states during social interactions should be integrated as one building block in long-term changes in personality traits to better understand the sources of personality trait development. Given the comparable high change rate in personality trait development during adolescence as compared to later life phases (Bleidorn et al., 2022), bottom-up effects might be particularly pronounced during this earlier age period.
Linking different types of social interaction partners and personality in adolescence
There is a longstanding theoretical and empirical tradition connecting personality traits and social relationships in dynamic transactions (Asendorpf & Wilpers, 1998; Back et al., 2011; Deventer et al., 2019; Mund & Neyer, 2014; Neyer & Asendorpf, 2001; Wagner et al., 2014). However, while most research has focused on longitudinal transactions of personality and social relationships, few studies have looked into daily social interactions and the specific state expressions that people report when interacting with different types of relationship partners. The few existing studies are exclusively based on student and adult samples (e.g., Bleidorn, 2009; Lindner et al., 2024), limiting our understanding for younger individuals.
Social interaction partners and state expressions
We are aware of three previous experience sampling method (ESM) studies investigating how Big Five personality states vary across daily social interactions. In one recent study, Lindner et al. (2024) compared personality states of adults (mostly university students) reported during interactions with different types of social interaction partners to situations in which they were alone. Moreover, two studies have examined how acting within different social roles is related to differences in momentary personality states in the daily lives of university students (Bleidorn, 2009; Heller et al., 2007). Integrating these existing findings, we derive specific hypotheses regarding the associations between different types of social interaction partners and each of the Big Five personality states in adolescence (Research Question 1).
State neuroticism is expected to be highest in social situations that are in some form stressful for adolescents (Suls & Martin, 2005). Previous research suggests that people experience lower state values in neuroticism when interacting with familiar people, particularly with friends, compared to being alone, whereas interacting with strangers makes no difference to average state levels (Lindner et al., 2024). In addition, acting within a student role was related to higher state neuroticism than acting within a friend role (Bleidorn, 2009; Heller et al., 2007), suggesting that adolescents may feel more anxious or tense during interactions with teachers.
State extraversion is expected to be highest when adolescents are motivated to initiate or deepen social bonds with peers (Wagner et al., 2014). Along these lines, previous research suggests that although people already express higher state extraversion when interacting with others in general (Lindner et al., 2024; also see Breil et al., 2019), the positive link between social interactions and state extraversion is even stronger in social interactions with friends (e.g., Bleidorn, 2009; Lindner et al., 2024).
State openness to new experiences is expected to be highest when adolescents are able to pursue their own interests (MacDonald, 1995). Previous findings comparing state expressions in friend and student roles (Bleidorn, 2009; Heller et al., 2007), suggest that openness to new experiences is higher in interactions with friends than with teachers. Although interactions with teachers could also foster openness via intellectual stimulation, previous research does not give clear indications for such an association, rather pointing to similar openness levels in family and work-related interactions (Lindner et al., 2024).
Similar to state extraversion, state agreeableness is expected to be particularly high when adolescents aim to maintain social relationships, especially friendships (Wagner et al., 2014). Supporting this idea, research on social roles points to higher state agreeableness when acting in the friend role compare to the student role (Bleidorn, 2009; Heller et al., 2007). The role of interactions with family members for adolescents’ state agreeableness, in contrast, seems less clear: Whereas most people act more agreeable with familiar others, including family (Lindner et al., 2024), adolescents seek to become independent from their parents, which may also result in less agreeable behavior (Branje, 2018).
Finally, state conscientiousness is expected to be highest in social interactions where adolescents need to perform or to obey, such as social interactions in school or at home (Ashton & Lee, 2001). Previous studies found that acting within the student role was related to higher state conscientiousness as compared to acting in the friend role (Bleidorn, 2009; Heller et al., 2007). Furthermore, Lindner et al. (2024) found that participants reported higher state conscientiousness at their workplace, which may translate to adolescents’ interactions with teachers.
Social interaction frequency and trait development
Based on the expected links between different types of interaction partners and adolescents’ personality states investigated in our first research question, we also investigated whether more frequent interactions with certain types of interaction partners translate into personality trait development (Research Question 2). Whereas recent research demonstrates that fluctuations of personality states in daily life can shape personality trait development in a bottom-up fashion (e.g., van Zalk et al., 2020; Wrzus et al., 2021), little is known about the role of social interactions with different interaction partners herein.
On a macro level, several studies on the interplay between personality and social relationships have examined how personality traits change across different social contexts in late adolescence and adulthood (Deventer et al., 2019; Jonkmann et al., 2014; Mund & Neyer, 2014; Wagner et al., 2014). For example, Jonkmann et al. (2014) found that young adults who moved out from their homes and started living with friends showed stronger increases in trait openness to new experiences and agreeableness compared to those who kept living with their families, but weaker increases in trait conscientiousness. In their study on personality and social network changes following high school graduation, Deventer et al. (2019) report that older adolescents who had more contact with their romantic partners decreased in negative affect, a facet of trait neuroticism, while those spending more time with their friends increased in goal striving, an aspect of trait conscientiousness. Whereas these studies generally support the key role of different social interaction partners for personality trait development, their macro perspective does not inform about the contribution of different social interaction partners in daily life. Moreover, given their hypothetical and cross-sectional nature, these findings do not allow for predictions regarding actual trait development. In sum, whether the frequency of social interactions with certain types of interaction partners translates into personality trait development in adolescence remains an open question.
The present study
Hypotheses Regarding Research Question 1.
As a second research question, we explored to what extent the frequency of interactions with different types of interaction partners predicts changes in adolescents’ personality traits across up to 1 year. We expected that the number of interactions with specific types of interaction partners that relate to the expression of certain personality states predict changes in adolescents’ corresponding personality traits in the same direction. For example, if, compared to social interactions with family, interacting with friends relates to higher state expressions of extraversion, we would expect that those adolescents who interact more frequently with friends increase more strongly in their trait extraversion. Due to the novelty of this research topic, however, we treat the research question on personality trait changes as exploratory.
Method
We analyzed data from two German longitudinal studies with adolescent samples and similar designs: SELFIE (Wagner et al., 2021) and SchoCo (Wagner et al., 2022), further called Sample 1 and 2. In Sample 1, data were collected from students who were tracked over the course of their final high school year and graduation across one year. Sample 2 originated from a research project tracking students in middle and late adolescence attending different school tracks during the COVID-19 pandemic across six months. In both samples, we used longitudinal trait data of three measurement points (t1-t3) and ESM data assessed in the week following t1. Hypotheses and data analyses were preregistered and the data and analysis code are available at our OSF page, together with an online Appendix (https://osf.io/vn7yp/).
Participants
In the original data sets, 220 adolescents in Sample 1 and 241 participants in Sample 2 took part in t1 and the ESM period. Since our research questions focused on different types of social interaction partners, we excluded participants who reported no social interactions during the ESM week (n = 2 in Sample 1 and n = 14 in Sample 2). 2 In our final Sample 1, N = 218 adolescents (76.15 % female) aged 16–22 (age: M = 17.70, SD = 0.98) rated an average of 13.51 (SD = 6.86, range: 1–29) social interactions during the ESM week. This resulted in a total of 2946 social interaction ratings. In our final Sample 2, N = 227 adolescents (85.46% female) aged 14–19 (M = 15.91, SD = 1.23) rated an average of 5.82 (SD = 4.71, range: 1–23) social interactions during the ESM week. This resulted in a total of 1322 social interaction ratings. Results from t-tests looking for differences between the participants of Sample 1 and 2 as well as between participants with and without longitudinal information are provided in Online Appendix A.
For our first research question regarding the link between different types of interaction partners and personality states, we ran parallel analyses in both samples as a test of robustness. 3 To estimate the power for detecting effects in the corresponding multilevel models, we conducted several simulation analyses with the R-packet “simr” (Green & MacLeod, 2016), following the tutorial by Arend and Schäfer (2019). The simulation results suggest that the given sample size provides sufficient power (>88%) to detect very small to small (b between .05 and .10) within-person effects in Sample 1 and small (b = .10) within-person effects in Sample 2. For our second research question regarding personality development, we collapsed data from Sample 1 and Sample 2 to achieve high power despite longitudinal sample attrition. Collapsing samples resulted in N = 445 participants providing self-reports on trait personality at t1. Out of these, N = 224 participants (50.34 %) remained in the sample at t2 and N = 176 participants (39.55 %) remained at t3. To determine the power for detecting effects in a three measurement points latent growths model given this sample size and the longitudinal missing values, we conducted a simulation in Mplus (Muthén & Muthén, 1998). Findings indicated that the sample size provided sufficient power (>90%) to detect small effect sizes (b = .11) of the covariate on the slope. The complete procedure and results of the power simulations are uploaded on our OSF page.
Procedure
Ethical approval for data collection of Samples 1 and 2 was granted by the German Psychological Society (DGPs) and by the ethics committee of the psychological institute of the University of Hamburg. The studies were promoted via social media platforms, personal outreach to schools, and leaflets in public spaces. Participants in Sample 1 received monetary compensation that was proportional to the number of completed study parts (up to €150), personalized feedback, and the chance to win prizes when completing the entire study. In Sample 2, participants could win gift vouchers in a lottery after each measurement point and received personalized feedback after full study completion.
All questionnaires were implemented with formr (Arslan et al., 2020) and the general procedure of both studies was very similar: At t1, participants completed a number of self-report questionnaires, including the assessment of their personality traits. While participants of Sample 1 completed the first measurement either in the laboratory in Berlin (Germany) or online, data collection in Sample 2 was completely online. After completion of t1, participants of both samples entered a weeklong ESM period in which they received five questionnaires per day (9 a.m., 12 p.m., 3 p.m., 6 p.m., 8 p.m.) on their smartphones. Among other variables, these ESM questionnaires included items regarding adolescent’s social interactions and their personality states. In both Samples 1 and 2, the ESM period was followed by two follow-up assessments measuring trait personality (t2 and t3). Whereas the time span in Sample 1 covered 1 year with half-year intervals between the three measurements, the time span in Sample 2 covered 6 months, with three-month intervals between measurement points.
Measures
All measures relevant to the current research were identical in Samples 1 and 2. If not specified otherwise, variables relating to the person were measured during the introductory session at t1, and during the follow-up sessions at t2 and t3, while momentary variables were measured during the ESM period following t1.
Momentary variables
Big Five personality states
Adolescents’ Big Five personality states were measured with one item each, which were adapted for our age group from the CONNECT study (Geukes et al., 2019): Participants were asked “Thinking back to the time that has passed since the last survey, how would you describe your behavior overall?” to assess personality states. During the first survey of a day, they were given a different instruction asking them to think back to the time since they got up. Participants then rated their own behavior regarding each of the five states with one bipolar item on a scale ranging from 0 to 10 with the following anchors: Neuroticism (reversed; nervous, easy to upset – balanced, relaxed), extraversion (reserved, quiet – enthusiastic, sociable), openness to new experiences (conventional, uncreative – widely interested, profound), agreeableness (critical, combative – cooperative, warmhearted), and conscientiousness (lazy, careless – reliable, conscientious). Overall, the intraclass correlation coefficient (ICC) Type 2 (Lüdtke & Trautwein, 2007) indicated satisfactory reliability of the individuals’ momentary personality states across their ESM entries. In Sample 1 (Sample 2), the Type 2 ICCs were .84 (.72) for neuroticism, .80 (.64) for extraversion, .83 (.71) for openness to new experiences, .87 (.75) for agreeableness, and .88 (.79) for conscientiousness.
Type of interaction partner
During the ESM, participants were asked whether any other persons were present and who this person was as follows: Partner (1), family member (2), friend (3), classmate/colleague (4), neighbor/acquaintance (5), teacher/supervisor/person to be respected (6), service provider (7), other (8), and nobody (9). If more than one other person was present, they could select multiple options. In a follow-up question, participants then indicated which type of interaction partner was most important for them in the situation. This way, there was always one type of interaction partner per social interaction/prompt, which was analyzed in this study. Based on these answers, interaction partners were summarized in four categories: friend (friend and partner), family (family member), teacher (teacher/supervisor/person to be respected), and other (classmate/colleague, neighbor/acquaintance, service provider). Using family as a reference category, friend, teacher, and other was coded as dummy variables indicating no interaction (0) or interaction (1) with the given interaction partner.
Control variables
After each interaction, participants were asked to indicate how well they liked their interaction partner on a scale from 0 (not at all) to 10 (very much so). Furthermore, we computed two variables capturing temporal information of the ESM based on the dates that were automatically recorded: First, the day of the ESM relative to the individual ESM start date ranging from Day 1 (0) to Day 7 (6) and, second, whether the interaction took part on a weekday (0) or weekend (1).
Person variables
Big Five personality traits
At each measurement point (t1-t3), trait neuroticism, extraversion, openness to new experiences, agreeableness, and conscientiousness were measured with the German version of the Big Five Inventory 2 (BFI-2; Danner et al., 2019). All personality traits were measured with 12 items each on a scale from 1 (strongly disagree) to 7 (strongly agree). Total omegas (ω) indicated good internal consistencies at each measurement point: .91/.91/.91 for neuroticism, .89/.89/.90 for extraversion, .87/.88/.91 for openness to new experiences, .87/.87/.88 for agreeableness, and .89/.88/.89 for conscientiousness.
Interaction frequency
Based on the participants’ ESM reports on momentary interactions with different types of interaction partners, we counted the individual interaction frequency regarding interactions with family, friends, teachers, and others.
Control variables
In the questionnaire at t1, participants indicated their age in years and their gender by identifying themselves as female (0) or male (1). Furthermore, we categorized the sample of all participants as Sample 1 (0) or Sample 2 (1).
Analytic strategy
Analyses were conducted in R (R Core Team, 2023) using R Studio (RStudio Team, 2021) as well as in Mplus (Muthén & Muthén, 1998-2024). To answer our first research question, we used multilevel models. Based on the findings from these multilevel models, we used latent growth modeling to answer our second research question. As common in multilevel and latent growth modeling (Little, 2013), the full information maximum likelihood estimator (FIML) was used to handle different numbers of Level 1 units and longitudinal drop-out, respectively.
Multilevel models
Accounting for the nested data structure, models addressing the first research question were specified on two levels: Social interactions (Level 1) were nested within individuals (Level 2). We set up and illustrated these multilevel models with the R packages lme4 (Bates et al., 2015) and sjPlot (Lüdecke, 2023), respectively. Across models, all continuous within-person predictors were centered at the participants’ individual mean (within-person centered) and all continuous between-person predictors were centered at the grand mean (sample-centered) to separate between-person and within-person associations (Bolger & Laurenceau, 2013). To predict personality states from the type of interaction partner, social interactions with family were used as the reference category. Specifications of the multilevel models are shown in the Online Appendix B.
As a follow-up, we tested whether the effects in the multilevel regression models remain robust when entering control variables. On Level 1, we controlled for liking of the interaction partner, the day of the ESM, and weekends. On Level 2, we controlled for age and gender of the participant. For all estimates, we report exact p-values (McShane et al., 2019) and discuss all effects significant up to p < .05. To account for the substantial number of effects, we tested whether significant findings remained robust after adjusting p-values in all non-exploratory analyses using a false discovery rate procedure (Benjamini & Hochberg, 1995). As indicator of effect size, we calculated
Latent growth models
The latent growth models addressing the second research question were set up in Mplus Version 8 (Muthén & Muthén, 1998-2024) and plots for illustration were created with ggplot2 (Wickham, 2016). Based on the previously established associations between different types of interaction partners and personality states in Samples 1 and 2, the individual interaction frequency regarding interactions with family, friends, teachers, and others was used to predict slopes of adolescents’ corresponding personality traits in the combined data of both samples. Specifically, the slope predictors were entered based on two considerations: First, the frequency of interactions with family was entered in all models. Corresponding to the analyses regarding the first research question, where interactions with family members served as the reference category, we also accounted for adolescents’ baseline of interactions taking place in their home environment in the latent growth models. Second, we entered the frequency of interactions with only those types of interaction partners as predictors that demonstrated significant fixed effects in Samples 1 or 2 in the analyses regarding the first research question. For example, if only interactions with friends were associated with state extraversion, only the frequency of interactions with friends were entered as predictor besides the frequency of interactions with family (which was always included) to predict changes in trait extraversion. In contrast, if interactions with friends, teachers, and others were related to a personality state, all respective indices of interaction frequency were entered as predictors besides the frequency of interactions with family to predict changes in the corresponding personality trait. Finally, as a follow-up, the control variables age, gender, and sample were entered as additional predictors.
In all models, change in adolescents’ personality traits were modeled using a second-order latent growth model with three measurement points (Bollen & Curran, 2006). Specifications of the measurement models including tests on measurement invariance across time and across samples are reported in the Online Appendix C. Building on the longitudinally invariant measurement models which indicated good fit, we added two time-specific latent factors of personality trait development: The intercept factor, reflecting the level of adolescents’ personality trait at the first measurement point, and the slope factor, reflecting linear mean-level change in this trait over time. Variance of the intercept factor and the slope illustrate interindividual differences in personality trait levels at t1 and in change trajectories, respectively. All models accounted for the covariance between intercept and slope. Given the novelty and the exploratory character of the second research question, we discuss all effects of the latent growth models with p < .05, while reporting exact p-values to provide the reader with full information to evaluate the findings.
Because time intervals between assessment waves varied across individuals (ranging from three to six month), we used the TSCORE option of the Mplus software (Muthén & Muthén, 1998-2024) to implement individually varying instead of fixed slope factor loadings. With this option, the time structure of the latent growth model is not based on the measurement occasions (t1-t3), but relies on observed individual specific timing scores. 4 Compared to classical latent growth models, this time sensitive modeling approach allows for individual differences in assessment intervals. However, these models estimate only unstandardized effects and do not provide information on absolute fit indices such as CFI, TLI, or RMSEA. We thus report Aikaike’s Information Criterion (AIC) as a relative fit index to compare models.
Results
We first report findings regarding the link on different types of interaction partners and personality states in daily life (Research Question 1). Based on these findings, we then move on to the predictive role of the frequency of interactions with certain types of interaction partners for personality trait development (Research Question 2).
The link between different types of interaction partners and personality states
Research Question 1: Means, Standard Deviations, and Correlations of Momentary (Within-Person) Variables in Samples 1 and 2.
Note. N = neuroticism, E = extraversion, O = openness to new experiences, A = agreeableness, C = conscientiousness. Intercorrelations below and above the diagonal correspond to the data of Sample 1 and Sample 2, respectively. Intercorrelations in bold font are significant at p < .05.
Testing Hypotheses 1 to 5, Figure 1 illustrates the estimates from the multilevel models without covariates for Sample 1 and 2. Table 3 provides an integrative overview of the result patterns across the two samples and across the models without and with covariates (complete findings with all key estimates are shown in Tables D2 and D3 in the Online Appendix). Partly supporting Hypothesis 1, adolescents of both samples reported higher levels of state neuroticism during interactions with teachers and others compared to interactions with family, while there was no difference during interactions with friends. In Sample 1, however, the pattern of associations was not robust when including the control variables. Instead, models with covariates show that neuroticism was linked to liking of the interaction partner, such that adolescents reported lower state neuroticism when interacting with somebody they liked more. Estimates from the multilevel models predicting adolescents’ personality states from different types of interaction partners. Note. The gray line in each plot represents the average personality state level during interactions with family (reference category). Estimates left from the gray line represent lower state levels during interactions with friends/teachers/others compared to interactions with family, whereas estimates right from the gray line represent higher state levels. *p < .05, **p < .01, ***p < .001. Research Question 1: Overview Over Effect Patterns Across Samples and Models. Note. Effects in the multilevel models were coded as positive (+) or no (0) association. RQ2 = Research Question 2. The covariate model controlled for the effects of liking of the interaction partner, the day of the experience sampling, and weekends on Level 1, and age and gender on Level 2.
Partly supporting Hypothesis 2, adolescents in both samples reported higher state extraversion during social interactions with friends compared to interactions with family. Although all results remained stable when including covariates, in Sample 1, this model showed an additional effect of higher state extraversion levels when interacting with teachers and others. This effect was not significant in any of the other extraversion models.
Mainly in line with Hypothesis 3, adolescents of both samples reported higher state openness to new experiences during interactions with friends, teachers, and others compared to interactions with family. Of note, the difference between social interactions with teachers and family members was not hypothesized and only occurred in the models including covariates.
Supporting Hypothesis 4, adolescents in both samples reported higher state agreeableness during interactions with friends compared to interactions with family. In models including covariates, two exceptions from this pattern emerged: In Sample 1, state agreeableness was linked positively to interactions with teachers and others. In Sample 2, the positive link between interactions with friends and state agreeableness was not significant after including covariates.
Partly consistent with Hypothesis 5, the overall result pattern suggested that adolescents reported higher levels of state conscientiousness during interactions with friends, teachers, and others compared to interactions with family. The difference between social interactions with teachers and family members, however, was only observed in Sample 1.
Overall, the results illustrate that compared to social interactions with family, adolescents expressed higher levels of state extraversion, openness, agreeableness, and conscientiousness when interacting with friends. Furthermore, they reported higher levels of state neuroticism, and, less consistently, of state openness to new experiences and state conscientiousness when interacting with teachers. Finally, adolescents reported higher state levels of neuroticism, openness to new experiences and conscientiousness when interacting with others in their daily lives. Thus, compared to daily interactions with family, interacting with different types of social interaction partners related to systematic differences in the expression of personality states.
The role of different types of social interaction partners for personality trait development
Research Question 2: Means, Standard Deviations, and Correlations of Person Variables in the Combined Sample.
Note. Results are based on NT1 = 445, Nt2 = 224, and Nt3 = 176 observations. Internal consistencies are provided as total omega (ω). Intercorrelations in bold font are significant at p < .05. Underlined intercorrelations represent retest reliabilities (e.g., rt1,t2).
Unconditional models
We first examined adolescents’ personality trait change between t1 and t3. The results of the unconditional models (Table D5 and Figure D1) are shown in the Online Appendix. Only one significant mean-level change emerged for the Big Five traits: Adolescents decreased in trait neuroticism over time (Mslope = −0.22, p = .001). Moreover, across traits, few reliable interindividual differences in personality change trajectories across time occurred. Only in the case of trait openness to new experiences the slope variance was significant (p = .028). Despite the lack of significance in the remaining Big Five slopes, all latent growth curve models could be fitted and the slope variances could be estimated. As an additional criterion besides testing for significance, Raudenbush and Bryk (2002) suggested calculating a 95% plausible value range (PVR) to indicate the amount of between-person variance in the individual slopes (formula on p. 78). Results indicated that personality trait change parameters ranged between −1.24 and 0.80 in the case of neuroticism, −0.82 and 0.84 in the case of extraversion, −1.07 and 0.85 in the case of openness to new experiences, −0.36 and 0.60 in the case of agreeableness, and −0.20 and 0.20 in the case of conscientiousness. Thus, although the variance parameters were not substantially different from zero, adolescents in our sample showed both increases and decreases in all five traits across the time of the study.
Conditional models
In the next step, we continued with our preregistered analyses and extended the unconditional latent growth models by entering social interaction frequencies with different types of interaction partners as predictor variables. 5 Notably, significant slope variances are currently not considered as a precondition to test for the effects of predictor variables on individual differences in change (Snijders & Bosker, 2011). Besides the frequency of social interactions with family, which was included in all latent growth models, remaining model specifications were based on our findings in the multilevel models (see Table 3). Specifically, we used the frequency of social interactions with teachers and others to predict changes in trait neuroticism, the frequency of social interactions with friends to predict changes in trait extraversion and agreeableness, and the frequency of social interactions with friends, teachers, and others to predict changes in trait openness to new experiences and conscientiousness.
Research Question 2: Conditional Latent Growth Models Predicting Big Five Trait Development From Social Interaction Frequencies.
Note. (Residual) variance: Variance estimates represent the variance in case of the mean, and the residual variance accounting for the variance explained by the predictors in case of the slope. CovIntercept, Slope = Covariance between intercept and slope. Δ AIC refers to the difference in fit relative to the unconditional model.
In a final step, we entered control variables to predict the intercepts and slopes of the latent growth models (see Table D6 in the Online Appendix). As a consequence, the effects of interaction frequency with family on declines in openness to new experiences and increases agreeableness were not significant anymore. In the case of openness to new experiences, however, adding the control variables led to a substantial decrease in model fit (Δ AIC: 10.509), which suggests that the previous model fit the data better. Moreover, none of the covariates had a significant effect on the intercept or slope (all ps > .560). Together, these results suggest that the control variables did not contribute substantially to explaining differences in the level and change of adolescents’ openness to new experiences. In contrast, adding the covariates improved the fit of the model predicting changes in agreeableness (Δ AIC: −7.865). Older adolescents reported higher (b = 0.09, p = .020) and male adolescents reported lower (b = −0.36, p = .033) mean levels of trait agreeableness at t1, while there were no covariate effects on the slope. In sum, the respective models with the best fit indicated that the predictive effect of the interaction frequency with family on increases in adolescents’ agreeableness may not be robust.
Discussion
Our study provides first insights into the role of social interaction partners for adolescent personality on the state and trait level, by integrating and testing current theoretical approaches on personality development (Fleeson & Gallagher, 2009; Geukes et al., 2018; Roberts & Wood, 2006; Wrzus & Roberts, 2017). Combining ESM data with longitudinal trait data from two adolescent samples, three key findings emerged: First, interactions with different types of interaction partners (family, friends, teachers, other) related to systematic differences in adolescents’ personality states. Second, most Big Five traits showed surprisingly little mean-level change and variance in their developmental trajectories. Third, results provided little evidence for bottom-up effects of social interaction frequency on personality trait changes across the time of study. In the following, we discuss our findings and point out future directions.
Personality states in daily life: Systematic links to social interactions with different types of interaction partners
During interactions with friends, adolescents expressed higher state extraversion, openness to new experiences, agreeableness, and conscientiousness compared to interactions with family members. While the findings regarding state extraversion and openness to new experiences align with prior research using adult samples (Bleidorn, 2009; Heller et al., 2007; Lindner et al., 2024), the findings regarding state agreeableness and conscientiousness add to the current picture and could be age-specific: During adolescence, individuals are highly motivated to connect with same-aged peers and to deepen friendships (Rubin et al., 2006; Wagner et al., 2014). Strategies to achieve this developmental task include acting more sociable (high state extraversion), embracing new people and ideas (high state openness to new experiences), but it appears it might also be beneficial to be kind and cooperative (high state agreeableness), as well as reliable (high state conscientiousness). The increased levels of state agreeableness during interactions with friends contrast with adolescents’ interactions with family that are typically marked by more conflicts (Branje, 2018). While the increased levels of state conscientiousness were contrary to our hypothesis, it may reflect adolescents’ efforts to fit in. In addition, it may mirror the beneficial effect of conscientiousness in school settings, where peers might particularly appreciate conscientious behavior, such as helping with homework (Lösch & Rentzsch, 2018). Thus, interactions with friends seem to co-occur with personality states that are useful in maintaining and building friendships. Together, these findings illustrate how personality and adolescent developmental tasks can promote each other on a momentary state level, and point to the relevance of friends as important socializing agents outside the family (Hurrelmann & Quenzel, 2018).
Apart from interactions with friends, interactions with teachers and with others were also linked to higher levels of state openness to new experiences and conscientiousness. Thus, adolescents reported being more open and conscientious during social interactions with all types of interaction partners compared to interactions with family. This could suggest that leaving the “safe heaven” of their families encourages adolescents to be more open-minded and reliable. These findings largely match previous research on personality states focusing on adults (e.g., Lindner et al., 2024), but we emphasize that adolescents’ interactions with teachers are unique to the lives of school students and therefore difficult to compare. Another line of research illustrates the longitudinal interplay between the quality of student-teacher relationships and personality traits in adolescence (Israel et al., 2023). Our results add to this picture by showing that daily interactions with teachers may promote expressions of state openness and conscientiousness, two features that have been consistently related to academic success at the trait level (Brandt et al., 2020; Israel et al., 2023). Looking closer at interactions with teachers, we note that differences in state openness to new experiences were only found when accounting for the control variables (liking, ESM day, weekends, age, and gender) and the differences in state conscientiousness were only significant in Sample 1. This inconsistent pattern may be explained by the fact that interactions with teachers made up only a small portion of all reported social interactions (7% in Sample 1 and 8% in Sample 2). Furthermore, the data in Sample 2 was collected during the COVID-19 pandemic. Although schools remained open during most of the study time, adolescents’ interactions with their teachers may have been different in this specific time period (Soemantri et al., 2023). Future studies should further clarify the link between social interactions with teachers and personality states.
With regard to neuroticism, adolescents’ state levels were higher during interactions with both teachers and others in contrast to interactions with family, while there was no difference in state neuroticism between social interactions with friends and family. It appears that adolescents experience more anxiety and nervousness (i.e., higher neuroticism levels) during interactions with less close (teachers) or unfamiliar (others) people, whereas social interactions with close partners do not elicit similar experiences. This result was also supported by a negative link between state neuroticism and liking interaction partners. Aligning with previous research using adult samples (e.g., Bleidorn, 2009; Lindner et al., 2024), the role of close others in reducing state neuroticism may thus generalize to younger age groups. In this context, our findings also illustrate how the unique function of family relationships (Neyer et al., 2011) may be reflected in the daily lives of adolescents. In interactions with family members, adolescents acted less neurotic, but also in ways that are often seen as less socially desirable (i.e., less extraverted, open, agreeable, and conscientious). Given the stability of family relationships (Laursen & Bukowski, 1997), interacting with family members might allow adolescents to make less effort and act more reserved, uncreative, combative, and lazy around their relatives without worrying about the consequences.
Overall, our findings on personality states in adolescents’ daily lives demonstrate that the presence of different types of interaction partners elicits systematic differences in personality state expressions. These results, which were largely robust across the two samples, were in line with our hypotheses regarding state agreeableness (Hypothesis 4) and partly supported the hypotheses regarding the remaining personality states (Hypotheses 1, 2, 3, and 5). Furthermore, the results resonate with theoretical (e.g., Geukes et al., 2018) and recent empirical work (Stadel et al., 2024), by highlighting the pivotal role of interaction partners in explaining within-person fluctuations in personality states. Future studies should explore more nuanced characteristics of daily social interactions with different types of relationship partners to better understand why they evoke systematic differences in personality states. To this end, research on different types of interaction partners could be supplemented with research on situation characteristics (Rauthmann et al., 2015).
Interaction frequency: A driver of interindividual differences in personality trait development?
No change and little variance in most Big Five trajectories
Before turning to the results addressing the second research question regarding the role of social interaction frequency in personality development, we examined the average developmental trajectories in our sample. Consistent with previous findings on late adolescence (Bleidorn, 2012; Luan et al., 2017; cf Borghuis et al., 2017), we found that, on average, neuroticism decreased over the time of the study (6–12 months). In line with the maturity principle, this trend is often considered “normative” in Western industrialized countries (Roberts et al., 2006; for a critique, see Klimstra & McLean, 2024), possibly reflecting improved abilities to handle developmental tasks, such as filling out new social roles and coping with stress at school (Denissen et al., 2013). Thus, adolescents seem to increasingly meet social expectations and become more relaxed as they grow up. With respect to the remaining Big Five traits, we found no average mean-level changes across time. These findings do not resonate with the maturity principle, which predicts increases in agreeableness and conscientiousness (Roberts et al., 2006), and are partly inconsistent with prior research (Bleidorn, 2012; Borghuis et al., 2017; Luan et al., 2017; van den Akker et al., 2014), although patterns are also mixed across different studies. Specifically, existing studies often differ in their specific sample composition (e.g., gender and education) and tracked adolescents’ personality trajectories over quite diverse periods of time, ranging from one to several years. Consequently, the patterns of adolescent personality development remain a topic for further investigation (also see Gillespie et al., 2024).
Furthermore, we found no substantial variance in the trajectories of adolescents’ personality traits across 6–12 months with respect to neuroticism, extraversion, agreeableness, and conscientiousness. This lack of variance is surprising given that previous research on the Big Five across different stages of adolescence and young adulthood consistently indicates significant variance in change trajectories (Bleidorn, 2012; Borghuis et al., 2017; Luan et al., 2017; van den Akker et al., 2014). Our diverging findings might stem from methodological features: First, tracking adolescents over the course of 6–12 months might have been too short to discover profound individual differences in personality development, as most studies on this topic adopted research designs spanning several years (e.g., Borghuis et al., 2017; Luan et al., 2017, but see also Bleidorn, 2012). Moreover, although the participants in Sample 1 graduated from high school during the study period, the adolescents in our study may have experienced rather few systematic changes in their social roles. As such, the social investment principle as an explanation of normative personality development and variance herein may not have fully applied (Roberts & Wood, 2006). To date, however, a theoretical framework on the specific interplay between social interactions and personality development in adolescence, specifying the time frames in which meaningful personality change can be expected, is still missing. Second, our sample was rather homogenous (i.e., mostly female and from higher school tracks). This homogeneity may have reduced interindividual differences in personality development. Accordingly, future studies examining how social interactions with different types of interaction partners predict differences in adolescent personality trait development should track adolescents over longer periods of time and collect data from more diverse samples. This kind of research may also inform theories on adolescent personality development.
The lack of variance in change trajectories of personality traits is contrasted with the high degree of intra- and interindividual variability at the level of personality states. That is, although adolescents’ momentary expressions of state neuroticism, extraversion, agreeableness, and conscientiousness in daily life varied strongly, this variability did not seem to translate into long-term changes of the corresponding personality traits. Two explanations may account for this discrepancy: First, the TESSERA model (Wrzus & Roberts, 2017) suggests that personality states must be expressed repeatedly, deviate from a person’s typical state expressions, and become part of a person’s self-concept via reflective processes to manifest in changes of relatively stable traits. Thus, although adolescents in our study varied substantially in their daily personality states, this does not necessarily mean that they deviated from their personal average, nor that they integrated their behavior into their self-concept (i.e., self-perceived personality traits). Second, it is an ongoing debate how personality states should ideally be measured and what their psychometric properties are (Horstmann & Ziegler, 2020). Relatedly, it remains an open question how personality states and traits correspond to each other (e.g., Borkenau & Ostendorf, 1998; Fleeson & Gallagher, 2009). Indeed, while corresponding personality states and traits in our study showed higher correlations than non-corresponding states and traits, these correlations were only moderate in size (rs ranging from .22 to .47). Thus, a person’s average expression of state agreeableness over several situations cannot be equated with their level of trait agreeableness. Related to this, adolescents in our study were instructed to report on their personality states “since the last survey” (or “since getting up”), which potentially results in a wide range of hours as reference period, depending on the response behavior. Accordingly, the instruction introduces some uncertainty in the measurement of personality states in the current study. Instructions to think of a specific time period (e.g., one hour) may be a possibility to minimize this measurement inaccuracy in future ESM studies. Accordingly, unsolved questions relating to the conceptualization and measurement of personality states may have limited the likelihood to discover corresponding variation of personality states and trait development.
Weak evidence for bottom-up effects of interaction frequency
The consideration of interaction frequency with specific types of interaction partners as predictors of individual differences in personality trait development revealed few effects. Specifically, adolescents who interacted more often with family members showed weaker declines in trait openness and stronger increases in trait agreeableness. However, both of these findings were somewhat inconsistent with regard to previous research and/or with regard to the overall result patterns in our study. In the case of openness to new experiences, it appeared that, while all adolescents became more conservative and less curious, this trend was less pronounced among those with more frequent family interactions. This result contrasts with research on young adults suggesting a stronger increase in openness among individuals who started living with friends compared to those who stayed with their families (Jonkmann et al., 2014). Moreover, the results on the trait level do not match the findings on the state level, which indicated lower openness levels during social interactions with family compared to interactions with friends. In the case of agreeableness, results suggested no change across time on average, but an increase among those with more frequent family interactions. Although family relationships have been linked to agreeableness before (Wagner et al., 2014), the interpretation of this finding is limited by two facts: First, as in the case of openness to new experiences, the results on the trait level do not match the findings on the state level. As such, we emphasize that the link between interaction frequency with family and changes in adolescents’ trait agreeableness needs further research to be fully understood. Second, the effect was not significant in the covariate model, which had a substantially better model fit.
In sum, our findings provide no robust evidence for the theoretically expected bottom-up processes of adolescent personality development over 6–12 months (Fleeson & Jayawickreme, 2015; Wrzus & Roberts, 2017) or for a key role of social interactions herein. Given that this part of our research was exploratory and that our study is the first to investigate the predictive role of interaction frequency in personality development, we view our findings as a first step that should be replicated in future work.
Limitations and future directions
By linking data from daily social interactions with both momentary personality states and long-term personality trait development, this study provides important first insights into the contribution of different types of interaction partners to adolescent personality. However, there are several limitations and open questions that should be considered.
First, we assumed that interactions with different types of interaction partners would evoke different personality states, such as higher state extraversion during interactions with friends. The direction of causality, however, might have been reversed. For example, adolescents in a state of higher extraversion might be more likely to elicit social interactions with friends. In the future, laboratory research or experience sampling designs that are enriched with experimental elements (Klasnja et al., 2015; Qian et al., 2022) are needed to answer questions on causality.
Second, the links between different types of social interaction partners and differences in personality states were measured over one week. However, little is known on the specific time or number of experiences on the level of personality states need to translate into changes at the trait level (Hopwood et al., 2022). Correspondingly, the periods over which personality change should be measured and how time should be statistically modeled are among the most pressing yet unsolved questions in the field of personality development (e.g., Bien et al., 2024; Roemer et al., 2024). Therefore, future studies should test different time scales on which social interactions might feed into adolescent personality development via bottom-up processes. Related to this, recent research has shown that participants in ESM studies may report more social interactions and more unique social interaction partners when using an event-contingent design (i.e., assessment initiated by the participant after the occurrence of a social interaction) instead of a signal-contingent design (i.e., assessment invited by prompts distributed over the day; Stadel et al., 2024). Since we applied a signal-contingent design in our study, future studies may explore whether the same result patterns emerge in an event-contingent research design.
Third, the latent growth curve models in our study only demonstrated significant slope variance in the case of trait openness to new experiences. The lack of slope variance in the remaining Big Five traits may have limited our ability to detect additional bottom-up effects from social interactions on personality trait development. Besides adopting sampling strategies that increase the heterogeneity of observed trait trajectories by including more diverse populations (Klimstra & McLean, 2024), next steps could include the analysis of changes at the level of specific personality facets instead of broader personality traits. Research on adolescents from the US with Mexican origin suggests that Big Five facets diverge from corresponding Big Five traits with respect to both mean-level changes and rank-order stabilities (Ringwald et al., 2023). Moreover, in the context of personality-social relationship transactions, it has been argued that effects of specific social relationships on broad personality traits are rather unlikely, while effects on specific personality facets are more common (Deventer et al., 2019; Mund & Neyer, 2014). Therefore, more variance in adolescents’ personality change trajectories and, relatedly, more predictive effects of social interactions with different types of interaction partners may be observed when personality is examined at the facet level.
Fourth, we considered the interaction frequency with different types of interaction partners to explore possible bottom-up effects of differences in personality states during social interactions on long-term personality trait development. Interaction frequency, however, might be only one potential and a somewhat indirect indicator of this assumed mechanism. Since different types of social interaction partners represent discrete categories, we were unable to extract within-person contingencies between interaction partners and personality states, as has been done in previous studies (Mueller et al., 2021; Quintus et al., 2021; Wagner et al., 2024). A task for future research is therefore to explore alternative approaches to model bottom-up effects on personality trait development with regard to the role of different types of social interaction partners or by accounting for average levels of personality states across social interactions.
Finally, the findings’ generalizability may be limited since the participants were above-average female adolescents (over 75% in both samples) and exclusively represented individuals from a so called WEIRD (Western, Educated, Industrialized, Rich, and Democratic) society (Henrich et al., 2010). Social roles and age-related societal expectations, however, differ between genders and cultures (Bornstein & Cheah, 2005; Seiffge-Krenke & Gelhaar, 2008). For example, adolescents’ increased social orientation towards peers may be more normative in individualistic than in kin-oriented societies (Edwards, 1992). Accordingly, future research needs to examine whether the links between different social interaction partners and personality (states and trait development) replicate in more diverse samples. Furthermore, in Sample 2, the COVID-19 pandemic likely impacted adolescent’s daily lives, leading to increased contact with parents at home and reduced face-to-face contact with peers due to temporal school closures and regulations to stay at home (Branje & Morris, 2021). Thus, although the results were fairly consistent across samples when analyzed separately and the sample variables were included as a control variable in the analyses with the collapsed data, we cannot exclude that the historical context of the pandemic might have led to some differences.
Conclusion
Do adolescents vary in their personality expression depending on their interaction partners in daily life? It seems that they do: Social interactions with different types of interaction partners were systematically linked to differences in adolescents’ personality states. Our findings suggest that social interactions with age-relevant interaction partners are an important factor in explaining variability in adolescent personality on a momentary level. While adolescents’ trait neuroticism levels on average decreased over the study period, there was little evidence for substantial variance in adolescents’ long-term personality development. As an exception, adolescents significantly differed in their development of the personality trait openness to new experiences, but there was no robust evidence for a predictive role of social interaction frequency herein. Future research should explore different aspects of social interactions and varying timeframes to enhance our understanding of bottom-up-effects in personality trait develop during adolescence and beyond.
Footnotes
Acknowledgments
We thank Natalie Hofbauer, whose Bachelor thesis was the starting point of this research.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the German Research Foundation with a grant awarded to Jenny Wagner (WA 3509/3-1).
Open Science Statement
Hypotheses and data analyses were preregistered and the data and analysis code are available at our OSF page (https://osf.io/vn7yp/). The complete wordings and response formats of the items used in the current study can be obtained from the codebooks provided at the OSF pages of the original studies of Sample 1 (https://osf.io/4gnz9/) and Sample 2 (
).