Introduction

Withdrawal from higher education and switching majors are a widespread problem. According to the OECD, on average, across countries, around 32% of bachelor students leave without graduation (OECD, 2022). One important criterion for students’ persistence in higher education is students’ social environment on campus. Successful social integration of university students is characterized by positive and frequent interactions with fellow students and teaching staff and a sense of belonging to the college community (Tinto, 1975). In his widely applied model of student departure, Tinto (1975) argues that low social integration is one of the crucial factors for student retention.

Because of the importance of social integration for university students, researchers are interested in the determinants of students’ social integration. Previous studies investigated the effect of age (Rubin & Wright, 2015), ethnicity (Rienties et al., 2012), and peer mentoring (Collings et al., 2014) on students’ social integration in their first college years. One determinant of social integration that has not yet been considered in research is students’ Big Five personality traits.

Previous studies on social networking behavior illustrate that the Big Five personality traits shape social behavior (Selden & Goodie, 2018). University students’ social integration comprises different forms of social networking behavior that might be linked to their Big Five traits. In the present study, we investigate whether the Big Five personality traits are a determinant of students’ social integration in the university context. We aim to address this research gap by analyzing two research questions. First, we analyze the association of the Big Five dimensions with university students’ social integration with students and teaching staff. Second, we are interested to find out whether majors at a university are characterized by a specific personality pattern of their students. Based on person-environment-fit theory, we expect students with a greater “fit” to the average personality profile of their major to experience better social integration with fellow students and teaching staff.

We are not only interested in the effect of students’ personality traits on the extent of, but also on the development of their social integration. Previous studies have predominantly focused on the determinants of social integration in students’ freshmen and sophomore years. Little is known about the development of students’ social integration. To gain a more comprehensive understanding of the relevance of personality traits for students’ social integration, we trace social integration longitudinally. We aim to determine whether personality traits and a personality match with the major increase in students’ social integration over time.

We use longitudinal data from the student cohort of the German National Educational Panel Study (NEPS) to address these research questions. The data allow us to observe the social integration with fellow students and teaching staff of university students who enrolled in 2010 over the course of four years. We estimate growth curve models with random intercept and slope.

The impact of personality traits on university students’ social integration

To the best of our knowledge, the direct impact of the Big Five personality traits on students’ social integration has not yet been analyzed. However, a large body of literature shows that the Big Five are linked to different aspects of social behavior. Extraverted individuals prefer socializing over solitary activities. They have larger personal networks, more non-kin friends, meet their friends more frequently, have a more central network position, and are more prone to finding new contacts than introverted people (Selden & Goodie, 2018; Thiele et al., 2018).

Agreeableness facilitates maintaining relationships and positively affects close relationships (Vasalampi et al., 2014). Agreeable individuals report more relatives as friends than less agreeable individuals (Laakasuo et al., 2017). Some studies find a positive effect of agreeableness on the formation of new friendship ties, but others find no significant effects of agreeableness on network size (Selden & Goodie, 2018). Overall, previous studies suggest that agreeableness has some positive effect on network size.

The other three Big Five domains have weaker links to social behavior. Open individuals are prone to connect with new and diverse people (Tulin et al., 2018). They have more friends living further away and meet them less frequently than less open individuals (Laakasuo et al., 2017). Furthermore, studies suggest that open college students gain more new non-kin contacts over time (Wagner et al., 2014; Zhu et al., 2013). Neurotic individuals have less emotional resources to invest in others (Smółka & Szulawski, 2011). Laakasuo et al. (2017) show that neurotic individuals report more kin as friends and meet their friends less often. Selden and Goodie (2018) find that neuroticism is unrelated to network size or composition, even though it is negatively correlated to building new friendships. Conscientiousness is more important for task-related behavior than for social behavior (Wolff & Kim, 2012). Previous studies indicate that conscientiousness is associated with more kin ties within the personal network but is not related to network size as a whole (Laakasuo et al., 2017; Selden & Goodie, 2018).

Overall, this review demonstrates that the Big Five impact social behavior but their influence differs from the observed aspects of social networks. University students’ social integration involves many different aspects of social networking behavior, i.e., building new contacts, maintaining existing ones, and positive interaction with teachers. We argue that the Big Five personality traits affect students’ social contacts in the university environment and, ultimately, their social integration with fellow students and teaching staff. We expect that extraversion, agreeableness, emotional stability, and openness have a positive effect on social integration with fellow students and teaching staff. Although conscientiousness is less important for non-kin networks, it is strongly connected to academic attainment (Vedel, 2014). High achievers represent valuable networking partners and are probably more popular with teachers. Therefore, we expect that conscientiousness has a positive effect on social integration. Furthermore, personality traits that foster continuously seeking new and maintaining existing social relationships should increase students’ social integration over time or at least prevent a decline in social integration. To summarize, we hypothesize that extraversion, agreeableness, emotional stability, openness, and conscientiousness are positively associated with students’ level and development of social integration with fellow students and teaching staff (H1).

The effect of student-environment-fit on social integration

The supplementary model of person-environment congruence argues that a good fit between a person’s personality and their environment fosters their achievement, satisfaction, and persistence in that environment. A good person-environment fit exists when a person possesses attributes similar to those of the other people in this environment (Muchinsky & Monahan, 1987).

Tinto (1975) applies the person-environment fit theory to his model of student departure. He argues that student-institution misfit, i.e., students’ incongruency with the social and academic environment at university, fosters drop-out. Tinto’s hypothesis is supported by many empirical studies. For example, Piepenburg and Beckmann (2021) show with a vignette study with German freshmen that low social integration with fellow students and teaching staff increases students’ intention to drop out of higher education. Müller and Klein (2022) show, based on a representative sample of German students, that low social and academic integration increases students’ drop-out risk. Consequently, students’ social integration depends not only on their individual social skills but also on their fit into the social environment of their university.

Within the educational institution, students interact primarily with fellow students and the teaching staff in their major. Hence, to feel socially integrated, students must “fit” with their fellow students and the teaching staff in their major. We argue that different fields of study might differ in terms of the typical personality patterns of the types of students they attract, such that a student’s personality may be a better or worse fit to the typical personality pattern of their field of study.

Previous German research emphasizes that each field of study has its own “culture.” Majors are characterized by different teaching cultures such as typical learning styles, forms of examinations, and the frequency and intensity of student–teacher interaction (Kim & Sax, 2014; Schaeper, 1997). Furthermore, students in a field of study share, for example, a clothing style, hobbies, political attitudes, communication behavior, and general patterns of perception, thoughts, and action (Weigand, 2014). An individual’s patterns of perception, thoughts, and action are shaped by their Big Five personality traits (McCrae & Costa, 1999). Hence, students in a major might represent a selective group regarding their personality, among other things. Evidence demonstrates that the Big Five are all linked to the field of study choice. For example, extraverted individuals are more likely to choose law and less likely to select natural science (Humburg, 2017). Based on these arguments, we assume that a specific personality pattern might be part of the major’s culture. We expect that students in the same major tend to share similar levels of extraversion, agreeableness, openness, neuroticism, and conscientiousness and test this assumption empirically with our data.

Additionally, social relationships tend to follow the principle of homophily. People seek relationships with others with similar characteristics regarding age, ethnicity, social class, and gender (McPherson et al., 2001). However, Laakasuo et al. (2017) find that people with high levels of openness are likelier to have friends of the opposite sex and another ethnicity. Nevertheless, evidence shows that university students tend to select friends with similar levels of extraversion, agreeableness, and openness, and the perceived similarity in personality traits predicts students’ higher friendship intensity (Selfhout et al., 2009). Furthermore, personality similarity regarding extraversion, conscientiousness, and openness eases students’ interaction with new acquaintances (Cuperman & Ickes, 2009; Kurtz & Sherker, 2003).

Based on these theoretical assumptions and research findings, we argue that the personality fit between students and the social environment in their field of study fosters social integration. Furthermore, a good student-major fit should facilitate the formation of stable and more intense relationships, increasing social integration over time. This leads to the following hypothesis: A higher deviation of students’ personality from the average personality traits of students in their major is negatively associated with the level and development of students' social integration with fellow students and teaching staff (H2).Footnote 1

Data and methods

Sample

We used the university students cohort of the German National Educational Panel Study (NEPS-SC5) for our analyses. NEPS-SC5 is a representative panel survey in Germany which follows 17,910 first-year undergraduate students who enrolled in a university or university of applied scienceFootnote 2 for the first time in the Fall 2010. Students are interviewed twice a year.

Students’ self-report on social integration is measured yearly, up to four times per student. Hence, we restrict our sample to students’ first 4 years of study. Social integration is measured for the first time in Fall 2011, i.e., around 1 year after student enrollment. Hence, our sample is restricted to students who do not drop out or switch majors before the first measurement of social integration. We observe students only as long as they remain in the same major or majors at the same university, which they chose initially in 2010. A change of major or university implies a new social environment within which students must again integrate. Therefore, changes in majors or universities contort the development of students’ integration over time. Within the observation period of 4 years, 61% of the students in our sample graduated, 13% are still studying but have not yet graduated, 10% dropped out or switched their major, and 16% are panel dropouts.

Missing values are imputed for a subset of variables with multiple imputations by chained equations (m = 100). The sample restriction and listwise deletion for non-imputed variables leave a sample size of 8941 students (62% female and 38% male) from 163 universities. Table 1 illustrates the number of students in each year in our analysis sample.Footnote 3

Table 1 Sample
Full size table

Analyses

To analyze the development of students’ social integration over the 4 years, we conduct multilevel growth curve models with random intercepts and random slopes. Growth curve models allow the modelling of individual trajectories of change over time with longitudinal data. Furthermore, time-variant and time-invariant predictors can be included in growth curve models to explain inter-individual differences in trajectories. Due to the nested data structure, two-level models are estimated in which time represents level 1 and students represent level 2. The models account for the nesting of up to four academic years within the students. Furthermore, clustered standard errors account for the lowest sampling unit, i.e., the combination of subject area and university at the aggregate level. Time is included as the linear variable “academic year” in the models.

We estimate separate models for the two types of social integration, i.e., social integration with fellow students and teaching staff. For every kind of social integration, we conduct two models, one including the students’ absolute level of the Big Five traits and one with the students’ deviation of the Big Five from the subject mean.

Measures

Social integration

The measure of social integration follows Tinto’s theoretical model, which is adapted for the German context and captures the students’ perceived social integration (Dahm & Lauterbach, 2016). For social integration with fellow students, students rate how much the following statements apply to them: “I have been successful in building contacts with other students during my studies to date,” “I know a lot of classmates with whom I can exchange ideas about questions in my field of study,” and “I have many contacts with students in my cohort” (response options for all items range from 1, does not apply at all, to 4, completely applies). The items capture a mixture of the quality and quantity of students’ social contacts. The three-item scale shows a good internal consistency (Cronbach’s alpha 0.84).

Social integration with teaching staff focuses on student-faculty interaction on campus. Four items cover this aspect of social integration: “I feel accepted by the instructors,” “I get along well with the instructors in my degree program,” “most of the instructors treat me fairly,” “the instructors are interested in what I have to say” (response options range from 1, does not apply at all; to 4, completely applies). The items focus on the quality of contact with teaching staff.Footnote 4 The scale shows a sufficient internal consistency (Cronbach’s alpha = 0.79). For both types of social integration, we built an index with the unit-weighted mean over all items of each construct.

Big Five personality traits

The Big Five dimensions are measured by the 10-item Big Five Inventory (BFI-10), developed by Rammstedt and John (2007). The BFI-10 is an ultra-short instrument that consists of ten items, two for each Big Five dimension. In the NEPS, an additional third item for “agreeableness” was included to increase construct coverage and enhance the reliability of this dimension, leading to an 11-item instrument.Footnote 5 The BFI-10 shows sufficient reliability and validity (Rammstedt et al., 2013) and, despite its brevity, achieves similar predictive validity to (much) longer Big Five inventories (Rammstedt et al., 2021). The Big Five were measured once during our observation period, in Spring 2012, i.e., around 1.5 years after student enrollment and ca. 6 months after the first measurement of social integration. Ideally, the Big Five would be measured before or directly after enrollment. However, whereas personality change is most pronounced in childhood, much less change happens after age 20 (Soto et al., 2011). Roberts and DelVecchio’s (2000) meta-analysis illustrates that personality traits become more stable with age and are relatively robust. During age 18–21, the consistency is around 0.51; during age 22–29, it increases to 0.57.

In our data, the Big Five are measured a second time, in 2016 (outside our observation period). We compared the two measurements of the Big Five for students who also participated in the survey in 2016 and calculated the linear correlation coefficients of Pearson’s r. The correlations range from r = 0.53 to 0.68 per trait and point to a moderate but sufficient test–retest correlation. We, therefore, assume that personality is quite stable. For students who drop out or switch their major before the measurement of the Big Five, the missing values on the Big Five dimensions are imputed.

Deviation of the Big Five

The student-environment fit is defined as the deviation of the university students’ Big Five traits from the personality pattern of their major. To find out whether different majors can be characterized by a specific personality pattern of their student body, we first group the students’ majors into 12 subject areas. Figure 1 displays the distribution of the different subject areas in the first academic year. The three biggest subject areas are teaching (34%), engineering (14%), and economics and management science (12%).

Fig. 1
figure 1

Distribution of students in the 12 subject areas

Full size image

Students aiming for a teaching degree are grouped in a separate subject area, independent from the majors they are enrolled in. These students are oversampled in NEPS-SC5. They visit the same classes as other students enrolled in the same majors but who do not aim for a teaching degree. However, these students take additional pedagogical courses designed for students aiming for teaching degrees only, independent from their majors. Hence, they must also build social contacts with teaching staff and other teaching students outside the student body of their majors to feel fully integrated. Furthermore, studies show that German students aiming to become teachers differ from other students: Teaching students have a lower interest and competence in academic research (Besa et al., 2023; Neugebauer, 2013). They are more frequently motivated by job security, social interests, and work-life balance when choosing their career (Neugebauer, 2013). Additionally, teaching degree students are more extraverted than other students (Osada & Schaeper, 2021; Roloff Henoch et al., 2015). Moreover, many students aiming for a teaching degree (67%) have two or more majors which belong to a different subject group.Footnote 6 These students could also create a bias in the definition of the personality pattern of the subject groups if not treated as a separate group.

For each of the 12 subject areas, we estimate the mean for each of the Big Five dimensions over all students enrolled in the subject area in 2012, when the Big Five are measured. For this estimate, we also include students who, in 2012, are no longer enrolled in the first major(s) they chose in 2010. Forty-three of the students are enrolled in two or three majors belonging to different subject areas. The Big Five traits of these students are included in the estimation of the Big Five mean for the subject areas of all their majors. Figure 2 shows the average of the Big Five traits for each subject area. The differences between the means of different subject areas are small but, in most cases, statistically significant. We conducted regression analyses with each Big Five trait as the dependent variable and the subject areas as covariates (see Appendix). Results show, for example, that students of all subject areas except sports, mathematics, veterinary medicine, and agriculture/nutritional science report significantly higher levels of extraversion than engineering students. Mathematics/natural science students have significantly lower levels of extraversion than engineering students. Furthermore, engineering students show significantly lower levels of agreeableness, conscientiousness, neuroticism, and openness than students enrolled in linguistics/cultural studies, human medicine, and students aiming for a teaching degree.Footnote 7

Fig. 2
figure 2

Mean of the Big Five traits for all subject areas

Full size image

Overall, the results suggest that the student body of each subject area is characterized by a different pattern of the Big Five personality traits.

In the next step, we compute each student’s similarity with the typical personality pattern of students in the same subject area(s).

For students with major(s) that belong to the same subject area, we use the following equation:

$${\Delta }_{i}={\sum }_{k=1}^{5}{\left[{x}_{ki}- {\overline{y} }_{kl}\right]}^{2},$$
(1)

where \({\Delta }_{i}\) denotes the deviation of the student i from the average personality profile of students with the same subject area, k denotes each of the five personality traits, \({x}_{ki}\) denotes the individual score of the student i on the respective personality trait k, l denotes the student’s subject area, and \({\overline{y} }_{kl}\) denotes the mean over all students of the individuals’ subject area l on the respective Big Five dimension k. For example, for an economics student, we subtract the average extraversion score of all economic students (\({\overline{y} }_{kl})\) in our sample from the student’s individual extraversion score \({(x}_{ki})\) and take the square from the result. We repeat this calculation for all other four Big Five traits and sum up the results for all Big Five traits. The result is the student’s overall deviation from the economics personality profile.

For students with majors that belong to two or three different subject areas, Eq. (1) can be modified as the following:

$${\Delta }_{i}={\sum }_{k=1}^{5}{\left[{x}_{ki}-\left(\frac{\sum_{j=1}^{m}{\overline{y} }_{klj}}{m}\right)\right]}^{2}$$
(2)

Here, m denotes the number of subject areas to which the students’ majors belong (m = 2 or m = 3). For example, for a student enrolled in art and economics, we calculate the mean of all art students for the particular Big Five trait and the mean of all economics students for the Big Five traits. Then we sum up the two means and divide the result by two. We subtract the result from the student’s individual score on the respective Big Five trait and take its square. Finally, we repeat this calculation for the other four Big Five traits and sum up the results for all Big Five traits.

Control variables

We control for students’ background characteristics that may be related to both personality and social integration in all models: migration background (yes/no), both parents have no tertiary degree (yes/no), students’ high-school leaving certificate GPAs, and age at enrollment as a centered variable. We include characteristics of the student’s program: university of applied science (1) or a full university (0) and the subject areas. For the students with more than one major that belongs to different subject areas, the subject area of the major that students report as the first major is used. Additionally, the employment status (working yes/no) and children in the household (yes/no) are included as time-variant variables and measured yearly. These factors restrict students’ time on campus and energy to build social contacts.

Results

Social integration with fellow students

We estimate growth curve models with social integration with fellow students as the dependent variable.Footnote 8 In all models, all control variables are included. Table 2 shows the results for the two models.

Table 2 Growth curve models with random intercept for social integration with fellow students
Full size table

Both models illustrate that students’ perceived social integration with fellow students significantly declines over time. Social integration slightly decreases by about 4.4 percentage points with one additional academic year. The items for social integration capture not only the quality but also the quantity of social integration. Hence, the decline might partly be explained by the fact that students probably have contact with many other students at the beginning of their studies, but after a while, they focus on a closer circle of friends.

In Model 1, the individual levels of all Big Five traits are included in the model as predictors of the random intercept. The results are partly in line with our Hypothesis 1. Extraversion, agreeableness, emotional stability, and conscientiousness are significantly positively connected to social integration with fellow students at the end of the first academic year. However, openness is positively but not significantly connected to social integration. Extraversion and agreeableness have the strongest association with social integration with fellow students. In Model 2, only the deviation of students’ Big Five traits from the subject group mean is included in the model, but not the individual Big Five dimensions. In line with Hypothesis 2, the deviation from the subject group mean is negatively associated with social integration with fellow students at the end of the first academic year. The association is relatively weak but statistically highly significant. Hence, students with a worse person-environment fit regarding their personality report lower social integration at the end of the first academic year than students with a better fit.

To find out whether the Big Five traits themselves and the deviation of the Big Five traits from the subject group mean also affect the development of students’ social integration with fellow students, i.e., the random slope, we estimate interaction effects between each covariate and the time identifier “academic year.” Table 3 depicts the results. The analyses illustrate that all interaction effects are close to zero and non-significant. Consequently, neither the absolute Big Five traits nor the deviation from the subject group has an impact on the decline of students’ social integration with fellow students over the course of their study. Consequently, Hypotheses 1 and 2 can only be supported in terms of the association of most personality traits and the deviation with the absolute level of social integration but not with the development of social integration.

Table 3 Growth curve models with random slope for social integration with fellow students
Full size table

Social integration with teaching staff

Table 4 shows the results of the growth curve models with social integration with teaching staff as the dependent variable. Students’ perceived social integration with teaching staff increases marginally but not significantly over time. Model 9 contains the Big Five traits as predictors of the random intercept. In line with our expectations, agreeableness, conscientiousness, and openness are positively linked with social integration with teaching staff, while neuroticism is negatively associated with social integration at the end of the first academic year. In contrast, extraversion is not significantly connected to social integration with teaching staff. Therefore, Hypothesis 1 is partly supported. The impact of students’ absolute Big Five traits on social integration with teaching staff is weaker than their impact on social integration with fellow students. Furthermore, a good relationship with university teachers does not depend on students’ level of extraversion. At the same time, extraversion is the personality trait that has the strongest association with students’ relationships with other students. However, openness seems more important for a good relationship with teachers than fellow students. In line with our Hypothesis 2, the deviation of the Big Five from the subject mean is negatively linked to social integration with teaching staff (Model 10). The effect is weak but statistically significant.

Table 4 Growth curve models with random intercept for social integration with teaching staff
Full size table

In the next step, we investigate whether the Big Five traits and the deviation of the Big Five traits from the subject mean affect the development of students’ social integration with teaching staff, i.e., the random slope. Table 5 depicts the interaction effects between the time identifier and all personality covariates. Similar to the results for social integration with fellow students, there are no significant interaction effects, with one exception. High levels of conscientiousness negatively impact the development of social integration with teaching staff (p = 0.09). Figure 3 depicts the small interaction effect in detail. Students with very high conscientiousness experience a significantly smaller increase in social integration with teaching staff over time. However, very conscientious students report better social integration in the first place. The differences in the social integration increase are only significant in the first 2 years. Over time, students with lower conscientiousness seem to catch up regarding their social integration with teaching staff, and in the fourth academic year, there are no longer any significant differences. Overall, we conclude that the Big Five traits and the deviation of the Big Five mostly do not positively affect the development of social integration with teaching staff. Hence, Hypothesis 2 cannot be supported.

Table 5 Growth curve models with a random slope for social integration with teaching staff
Full size table
Fig. 3
figure 3

Marginsplot for social integration with teaching staff: interaction effect between conscientiousness and academic year

Full size image

Robustness checks

We performed additional analyses with a more restricted sample to test the robustness of our results. We limited the sample to students who either earned a degree in their first major within the observed four academic years or are still enrolled in their first major in the fourth academic year. The results confirm our previous findings and point to robust results (see Appendix).

Furthermore, previous research suggests that the importance of specific personality traits for group formation varies by sex. Laakasuo et al. (2020) investigated friendship groups within a European fraternity and found that a higher individual distance from one’s group members in conscientiousness was significantly negatively associated with a sense of belonging for men but not for women. We conducted all our models with interaction terms between sex and the single Big Five traits, the deviation of students’ Big Five traits from the subject group mean (random intercepts), and the interaction terms between those variables and the academic year (random slopes) to reveal potential sex differences. We only found one significant, but very small interaction effect: The positive effect of extraversion on social integration with fellow students (random intercept) is 0.03 points higher for men than for women (p = 0.03) (see Appendix). Hence, extraversion seems to play a very slightly larger role for men than for women when socially integrating with other students.

Conclusion and outlook

This study aimed to ascertain whether university students’ Big Five personality traits and person-environment fit affect the level and development of students’ social integration. The study focuses on students’ perception of their social integration with fellow students and teaching staff. Person-environment fit was defined as the similarity of students’ individual Big Five traits and the average Big Five traits of the other students in their subject group. Regression analyses showed that most but not all majors have a significantly different personality pattern in their student body. We conducted separate growth curve models for social integration with fellow students and teaching staff with random intercept and random slope.

Previous studies demonstrated that agreeableness and extraversion have the strongest theoretical and empirical link to social behavior. Our results for the random intercept confirm that agreeableness has one of the strongest positive associations with both forms of social integration at the end of the first academic year. Extraversion is the most important personality trait for social integration with fellow students. Furthermore, extraversion is slightly more relevant for mens’ than for womens’ social integration with other students. However, extraversion does not play a significant role in social integration with teaching staff for both sexes. A possible explanation for this phenomenon could be that the measurement of social integration with teaching staff in the NEPS-SC5 focuses more on the quality than the quantity of contact with teaching staff. Extraversion is mainly important for building new contacts and less relevant for weaker ties (Roberts et al., 2008). Although previous empirical studies cannot find a strong link between conscientiousness and networking behavior (Wolff & Kim, 2012), we expected it to be significant for social behavior in an academic setting. Our results illustrate that conscientiousness is significantly positively linked to social integration with fellow students and teaching staff at the end of the first academic year. According to previous empirical studies, neuroticism and openness play a smaller role in social networking behavior (Selden & Goodie, 2018). Our analyses show that both emotional stability and openness are positively linked to social integration at the end of the first academic year. However, openness is the least important personality trait for social integration with teaching staff, and the connection between openness and social integration with fellow students is not significant. Open individuals are curious and interested in new information. In an academic setting, it seems reasonable that openness is especially valuable for the relationship with teaching staff within the classroom. Interested students are likely to be more popular with teachers.

Results for the random intercept on the person-environment fit support our expectations. Low student-major fit regarding personality is significantly negatively associated with both forms of social integration at the end of the first academic year. Students with a personality that deviates from the other students’ average personality in their major tend to report less and lower quality contact with fellow students and teaching staff. However, the effects of person-environment fit are quite small. Overall, the direct effect of students’ individual-level Big Five traits seems more important for their social integration at the end of the first academic year.

In the next step, we were interested in the development of social integration over time. Results demonstrate that students’ perception of social integration with fellow students decreases slightly but significantly over time. In contrast, perceived social integration with teaching staff remains constant over time. The results of the random slope illustrate that the development of both forms of perceived social integration is not affected by students’ individual personality traits or their person-environment fit regarding their personality. Hence, students’ personalities cannot protect them from a loss of perceived social integration over time or even foster an increase in social integration over time. Overall, personality seems to shape mainly students’ initial level of perceived social integration in the first academic year. It has to be kept in mind that the measurement of social integration in our study reflects students’ perceptions of their social integration. Our study cannot reflect the development of students’ objective social integration.

This study has some limitations. Firstly, there is a selection effect. Students’ Big Five personality traits affect their choice of major (Humburg, 2017). Therefore, most students that enrol in a specific major already possess a relatively good person-environment fit regarding their personality from the start. Students with a very bad person-environment fit probably would not enrol in that particular major in the first place or will leave their first chosen major and will no longer be part of our sample. Thus, with increasing academic years, the sample will contain fewer students with a bad environment fit. This might explain the relatively small effects of person-environment fit on social integration found in this study. For a robustness check, we replicated the analyses with a sample restricted to students who remained in their initially chosen major for all observed four academic years. The results were the same as in our original sample.

Secondly, students’ social integration and Big Five personality traits are not measured directly after enrollment but 1 year and 1.5 years, respectively, after enrollment. Although personality is quite stable in college years, it might still change slightly over time. Furthermore, students who leave the major before the first measurement of social integration are not included in our sample. Therefore, our analyses might underestimate the effects of person-environment on social integration because students with a bad fit and, hence, inadequate social integration might drop out before completing the first academic year.

However, the results illustrate that students’ Big Five traits are important factors for students’ formation of social integration and, consequently, their well-being and dropout risk. Person-environment fit regarding their personality is a weaker determinant, and for social integration with teaching staff, it is negligibly small. Our empirical tests demonstrated that most, but not all majors show a significantly different Big Five pattern of their student body. Although previous studies also show that students in various subjects differ regarding their personality traits, majors differ regarding many other aspects, e.g., examination style, teacher-student interaction, communication behaviour, etc. (e.g., Weigand, 2014). These other aspects might play a larger role in defining a person’s fit to a major than their personality. We do not observe these aspects in this study, but future research might consider a broader definition of student-major fit.

Another interesting aspect for future research would be a closer-range investigation of students’ peer networks to define student-major fit. Our paper focused on a general personality pattern typical for a field of study, independent from students’ university. Nevertheless, it would be fruitful to analyze the direct fit of a student’s personality only to the personality of students who are enrolled in the same major at the same university to capture the immediate social environment of students. For example, Laakasuo et al. (2020) analyzed 25 friendship groups in a European fraternity and investigated the impact of students’ similarity to the other members of their respective friendship groups regarding their Big Five personality traits on group formation. Such a smaller-scale perspective on student-major fit could reveal whether there are regional differences between universities regarding the personality profiles of the same majors. It could take into account that people with similar personality traits are geographically clustered, as evidence for the US and UK shows (Jokela et al., 2015; Rentfrow et al., 2013). Furthermore, it could show whether students’ fit into their immediate peer group regarding personality plays a larger role in social integration than their fit to a general major personality. However, our data does not allow such an additional analysis because our sample does not contain all students in the same major at a university.Footnote 9 Therefore, it would be an interesting research question for further research with different data.