Leaving the parental home during emerging adulthood

The transition to adulthood, which comprises a series of events that include leaving the parental home, starting a professional career, union formation, and entry into parenthood, has been characterized as “late, protracted, and complex” for the current generation [6]. The timeline of those events’ occurrence is lengthening, possibly reversible, and no longer in a certain order. This group of young people (aged 18–29) are referred to as “emerging adults”, describing those who feel somewhere in between adolescence and adulthood and who experience instability, role exploration, and change in life pathways [19].

The focus of this study is the transition of leaving the parental nest, which is considered an important step toward establishing a self-sufficient and independent adult life for adolescents and emerging adults. Due to the observed delays in home-leaving and the decreasing normativity of the timing of leaving [1, 20], researchers have focused their attention on investigating the consequences of continued co-residence with parents. First, on-time leavers reported higher rates of romantic activity in comparison to those who did not move out at the same age [21]. Second, those who continued to reside with their parents were less mature and had achieved fewer developmental tasks (e.g., starting a career) [22]. Third, while emerging adults who remain at home may feel as if they are treated like children and show lower levels of well-being [23], their parents are also more likely to feel financially and emotionally burdened [22].

Because of these consequences, we aim to study the reasons that contribute to emerging adults’ decision to leave. Studies have indicated that structural constraints (e.g., parental resources and de-standardization of traditional values) [2], demographic features (e.g., gender, education, and employment status) [3, 24, 25], and personality traits [8, 26] have an impact on individuals’ timing of leaving the parental home. Social network effects (e.g., siblings, colleagues, friends) may as well be crucial when making life course decisions [12, 27]. For instance, a sibling’s leaving was shown to be positively associated with one’s own leaving [7]. In this study, we aim to further elaborate on the sibling transmission effect, using siblings’ personality similarity.

Sibling similarity in life-course transitions: Observational learning processes and relationship closeness

Siblings typically live under the same roof and spend a considerable amount of time together during childhood and adolescence [11]. Having their relationship nurtured over years of exposure, they may have a stronger impact in each other’s transition to adulthood compared to other social network members. Previous studies suggest that social influence processes, namely, the processes of social contagion and social learning [12, 28, 29], explain similarities between siblings in terms of their life course trajectories [10]. The theory of social contagion suggests that individuals may influence those around them by their behavior, both intentionally and unintentionally. At the same time, people also unconsciously observe and imitate the behaviors of their networks and those they are close to (e.g, age peers, friends, and siblings) [12]. The notion of social learning points to the contagiousness of behaviors as well, but has a stronger focus on the observational learning aspect [28]. Similarly, the paving-the-way hypothesis posits that siblings who transit to a new chapter pave the way for those who follow [30]. Of relevance to the present study is that the three mechanisms all regard sibling similarity as a driver of mimicking behaviors.

Drawing on these theories, a number of studies have documented that siblings who are alike in terms of demographic characteristics are more likely to be a stronger source of contagion and role models and pave the way for one another. For example, siblings with a smaller age difference have a higher chance of influencing each other regarding dropping out of high school [31], leaving the parental home [7], and entry into parenthood [14]. Same-gender siblings are also more similar with respect to internal migration [30] and patterns of family formation [15]. One of the reasons sibling similarity is related to life-course resemblance is their relationship quality and emotional closeness [13, 31, 32], that siblings similar in age shared more experiences during childhood and could thus enhance relationship closeness in adulthood [33]. Other studies indicated that gender similarity is associated with increased sibling support and interaction [34, 35]. Nevertheless, while we know being similar in demographic traits is an important factor in reinforcing sibling transmission of life course transitions, including nest-leaving, it is unknown whether other forms of sibling similarity function the same.

Next to these social mechanisms, genetics are an important cause of sibling similarity. On average, full siblings inherit 50% of the same genetic variants from their parents, which makes them genetically more predisposed to being similar in their leaving home behavior, among other life course transitions [13, 36, 37]. Likewise, siblings’ personality similarity can also be explained by their shared genetics, given that personality traits are to some degree heritable [38, 39]. The focus of our study was to examine which factors moderate intragenerational transmission of leaving, specifically how personality similarity functions as a potential determinant.

Combing psychological and sociological perspectives: Similarity-attraction effect and personality similarity

Building upon theories of cognitive dissonance and social comparison [40, 41], the similarity-attraction effect (SAE) suggests that individuals favor a logical and consistent view of the world [42]. The perspective suggests that we prefer to encounter people with similarity in personality traits because in doing so, our opinions, ideas, and attitudes are validated. Such reinforcement can be associated with positive feelings and leads to attraction. On the contrary, individuals who have different personality traits often disagree with us and create inconsistency in our worldview. We may thus feel anxious, confused, and repelled by those people. In other words, because similarity affords predictability, it enables individuals to interact in a more relaxed manner. This effect is also at the center of the notion behind homophily, which demonstrates that similarity can serve as a foundation for interpersonal attraction [43] and that relationship between dissimilar individuals dissolve more quickly [44]. Despite mixed empirical evidence [45], most literature suggests that personality similarity is linked to better interpersonal relationships than is personality dissimilarity [18, 46].

While the previously discussed sociological theories emphasize that similarity in demographics facilitates warm relationships, the SAE stresses the effect of similar personality traits. Research even showed that personality offers a stronger similarity effect than demographic characteristics do in both Canadian and Japanese contexts [18]. Moreover, the SAE may account for a great deal of social contagion and peer influence [47]. Taken together, it seems that the sociological and psychological theories share a commonality, that similar individuals tend to have increased transmission and contagion of behaviors, likely via enhanced interaction and close relationship between them. When it comes to cross-sibling effects on nest-leaving, it is important to examine whether and how siblings’ personality similarity, serving as a proxy for their relationship closeness, strengthens the intragenerational effect of leaving the parental home.

The role of sibling personality similarity in transmission of leaving the parental home

It is widely accepted that a five-factor model of personality (the Big Five) encompasses the most salient aspects of personality [48] that shape individuals’ behaviors and experiences [49]. Extraversion is characterized by being enthusiastic, assertive, and positive, and the tendency to enjoy and be dominant in social interaction. Agreeableness refers to the tendency to engage in prosocial behaviors and to value harmony, friendliness, and positive relationships. Neuroticism is defined by being vulnerable to stress and not being able to cope with negative emotions. Finally, openness to experience is expressed in the appreciation of art, unconventional ideas/creativity, curiosity, and intellect. Regarding the similarity effect of the Big Five model, there is consensus that the traits of extraversion, agreeableness and neuroticism are most relevant to people’s social behavior [46, 50, 51]. In the following, we discuss how siblings’ Big Five traits can be related to similarity in their timing of home-leaving.

As for extraversion, scholars found that dyads who have the same level of extraversion (both introverts or both extraverts) interact more easily than dyads composed of an introvert and an extravert do [46]. This suggests that the interaction style between people with the same personality makes their relationship more enjoyable. In a similar vein, people often get along with those showing the same level of extraversion because they share the same amount of sociability [16, 17]. Importantly, it was particularly dyads in which both are introverts that had higher odds of being friends and having joyful interactions [46, 52]. This suggestion goes hand in hand with the aforementioned social influence processes that similar demographic characteristics, associated with enhanced relationship quality, lead to life-course resemblance. Because better interaction and relationship closeness may increase the likelihood that a sibling’s leaving paves the road, Hypothesis 1 (H1) was formed: The association between siblings’ departures is stronger when they have similar levels of extraversion (a) and, particularly, when they are both introverts (b).

Agreeableness is often characterized by the ability to control undesirable emotions, develop healthy and harmonious relationships, and be less involved in aggressive, threatening, or conflict-related behaviors [53, 54]. In terms of homophily by agreeableness, studies have shown that adolescents and emerging adults are inclined to form friendships with and be closest to those who have a similar level of agreeableness [16, 17]. They demonstrated that two agreeable people form the most beneficial relationship, compared to a disagreeable pair or a pair in which one is agreeable and the other is disagreeable. Concerning the role of agreeableness in sibling relationships, agreeableness is one of the best predictors of high-quality sibling relationships [55]. In order to have a warmer and less conflictual sibling relationship, it is important for siblings to have high levels of agreeableness. Considering this, Hypothesis 2 (H2) is as follows: The association between siblings’ departures is stronger when they both have high levels of agreeableness.

While agreeableness is related to healthy and close sibling relationships [55], neuroticism is negatively associated with the quality of sibling relationships [55–57]. Concerning the similarity-attraction effect, similarity in neuroticism can encourage friendship and group formation [58]. Individuals who are similarly non-neurotic are more inclined to be friends, because conflicts in dyads may escalate when both members are emotionally unstable [52]. When homophily is manifested in neuroticism, it seems reasonable that when siblings are both not neurotic, their relationship and interactions are more pleasant and harmonious. As a result, we formed Hypothesis 3 (H3): The association between siblings’ departures is stronger when they both have low levels of neuroticism.

Openness to experience has traditionally been considered an intrapsychic characteristic which pertains to individuals’ intellectual lives and is less associated with their social behavior and relationships, unlike extraversion, agreeableness, and neuroticism [50, 59]. However, similarity in openness can also predict whether individuals become friends [16]. That is, although openness is not a particularly meaningful trait for interpersonal connection, adolescents’ self-rated openness tends to be similar to those of their ideal friends [60]. The aforementioned studies did not indicate whether the effect of personality similarity is especially strong at only one end of the trait dimension, whereas the other demonstrated that only those with greater openness have similar friends [52]. Based on these studies, Hypothesis 4 (H4) suggests the following: The association between siblings’ departures is stronger when they have similar levels of openness (a), and, particularly, when they both have high levels of openness (b).

Dataset selection and structure

The data for this longitudinal study are drawn from the survey Understanding Society: The U.K. Household Longitudinal Study (UKHLS) [61]. UKHLS is an ongoing annual and nationally representative panel survey that gathers a variety of information at the individual and family level. The dataset consists of 28 waves; waves 1 to 18 belong to the British Household Panel Survey (BHPS) and waves 19 to 28 are from Understanding Society. Together, the waves cover the entire period between 1991 and 2019. Understanding Society is a continued version of the BHPS, featuring data harmonized between both studies, allowing us to track households across the whole period. Over 60% of those who participated in BHPS decided to continue with Understanding Society [61].

To capture sibling effects of personality traits in function of parental home leaving, we considered all children from a family, unlike previous studies selecting families with only two children when studying sibling transmission [10, 62]. This approach results in a more representative sample of multiple-child families. To model the associations between all considered siblings, we created the following structure. As shown in Fig 1, our data structure consists of four levels: family level (level 4), child level (level 3), sibling dyad level (level 2), and wave/time level (level 1). The time level is nested in the sibling dyad level, which is nested in the child level, embedded in the family level. Because we considered all children from a family, the data structure is more complex compared to a design with only two children per family (as those are mirroring siblings). Therefore, the sibling dyad level is added to the model. Fig 1 provides an example with three children (X, Y and Z) at risk of leaving the household. Each of them can function both as a child at risk and as a sibling with potential influence on another sibling’s departure. For instance, when Child X is at risk, Child Y and Child Z are the siblings of influence (i.e., Sibling Y and Sibling Z), leading to sibling dyads XY and XZ. If Child Y is at risk, Child X and Child Z become the mirroring siblings (i.e., Sibling X and Sibling Z), resulting in sibling dyads YX and YZ. At the wave/time level the absence or occurrence of an event is measured for the child at risk.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Example of the four-level data structure.

https://doi.org/10.1371/journal.pone.0284808.g001

Siblings were identified based on the mother’s identification, in case of biological siblings. To take into account children from blended families, we included all children joining the household from either the mother’s or the father’s side in case of a parental dissolution. To determine whether children had left the parental home, we relied on the information of the household grid. In other words, when they were no longer registered in the household grid, they were considered to have left the nest. There were several conditions in which disappearing from the household was not counted as leaving, ensuring that children who were no longer reported in the household grid most likely left the parental household. First, when children left the household because of parental relationship dissolution, as long as they still lived with one of their biological parents, they were not regarded as having left but as new members of blended families. Second, children who passed away were right censored. Third, if a family stopped participating in the survey or migrated outside of the United Kingdom, attrition of the whole family was right censored as well.

Children in some families had already left the nest prior to participating in the survey. To avoid bias, we did not take those families into account. Finally, we left censored those children who had left the parental home before the age of 16. Given our focus on the departure of adolescents and emerging adults from the parental home, those who had not left the nest at the age of 29 were right censored. When children switched roles to the sibling of potential influence, they were required to be at least 13 years old. There was no maximum age restriction for the siblings of potential influence. Overall, we analyzed 33,612 observations in 4,976 sibling dyads (representing 3,717 children and 1,845 families).

Measures

Control variables.

We controlled for a number of background characteristics that were known to impact the timing of children’s nest-leaving [3, 25, 68]. Those covariates were included at the family level, the child level, or the dyad level. First, at the family level, parents’ educational background reflected the highest educational qualification achieved. It was grouped into four categories: high (have a degree), middle (completed A-level or secondary high school), low (everything below the other two categories), and unknown. Employment status of the parents was categorized as either being an employee/self-employed, no employment, or unknown employment. The parents’ marital/relationship status could either be in a relationship (including being married or cohabiting with a partner) or single (i.e., divorced or separated). An unknown category was included to avoid losing a considerable number of observations due to missing values. For the parental covariates, when both biological parents were present, we selected the mother. Otherwise, we took the available biological parent as the targeted parent. The number of children each family had was included as a continuous variable.

Second, at the child level, age was modelled using both a linear and a quadratic term. Gender was a dummy variable (ref = male). Children’s educational attainment, employment status, and relationship status were measured using the same categories as for the parents. Third, because we also considered siblings of blended families, we controlled for whether the sibling was a child’s biological sibling at the dyad level by including a dummy indicator (ref = not a biological sibling). Moreover, we controlled for the age difference between siblings in its absolute value (e.g., an age difference of 2 or -2 were both counted as 2) to account for departing around the same time due to age similarity. Table 1 provides an overview of the descriptive statistics at each level. Whether these covariates were time variant or invariant is also indicated in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Descriptive statistics of the dependent and independent variables.

https://doi.org/10.1371/journal.pone.0284808.t001

Analytical strategy

The analysis drew on discrete-time event-history models predicting the timing of the at-risk children’s nest-leaving using a logit link function. The models were estimated by the PROC GLIMMIX procedure in SAS 9.4. We applied multiple strategies to adjust for potential sources of confounding. Besides a causal effect of someone’s leaving on their sibling via social learning or contagion, nest-leaving could also represent the effects of different endogenous factors. For instance, because siblings are raised in the same family context, they experience similar family norms and expectations of when to leave the parental home [69, 70]. As a result, the models included random effects (RE) at the family, child, and sibling dyad levels. Moreover, to control for household background, the models introduced various socioeconomic and demographic characteristics as covariates, aiming to avoid the impact of confounders on our results [9, 15]. The analysis indicated that allowing random intercepts at the different levels resulted in an absence of unexplained variance at the sibling dyad level. Therefore, as a first robustness check, we repeated the analysis by randomly selecting one sibling dyad per child (i.e., 3,717 sibling dyads for 3,717 children).

A second robustness check repeated the analysis by using fixed-effects (FE) models. Since at-risk children and their siblings stem from the same families, it is plausible that children from the same family simply leave the nest around the same age or time because they share a number of unobserved/endogenous characteristics (e.g., parental expectations, family norms) [69]. FE models reduce heterogeneity biases originating from time-invariant family influences [71] and are complementary to RE models, which are more generalizable [72]. Therefore, we draw on the FE models to address confounding by household background factors. The robustness checks can be found in the supplementary analyses (see the S1 File).

Main analysis

Table 2 presents the results of the event-history analysis (the full table is in the Table S1 in S1 File). All Big Five personality traits were modeled simultaneously when examining the main and interaction effects. To avoid a long table, interaction terms that were not significant or relevant to the hypotheses were not reported in Table 2 but can be found in Table S1 in S1 File.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Unstandardized coefficients of the multilevel discrete-time event-history analysis predicting at-risk children’s event of leaving (N time level = 33,612).

https://doi.org/10.1371/journal.pone.0284808.t002

The models were estimated stepwise. In Model 1, we included background characteristics at the family, child, and sibling dyad levels, together with the Big Five traits of the at-risk children and the siblings. The model showed that at-risk children whose parents had a higher educational background and were divorced/separated were more likely to leave the parental home. The number of children in the family was not related to the leaving of the parental home. At the child level, female, employed, and non-single children showed a higher tendency to leave. Children with a middle level of education had a higher chance than those with a lower educational background of remaining at home. At the sibling dyad level, a positive association between siblings’ departures was observed, indicating that an association between siblings’ timing of leaving. The at-risk children were less likely to leave when they had a biological bond with their sibling and when there was a larger age difference between them. In terms of the Big Five personality traits of the children at risk, our results suggest that among all the Big Five traits, only openness to experience had a statistically significant effect in that the more open one was, the less likely one was to leave. We did not find any significant associations between a sibling’s Big Five traits and the timing of a child’s leaving.

Model 2 tested whether siblings’ similarity in the Big Five traits was related to their similarity in the timing of nest-leaving by including the sibling personality similarity variables and the terms of interaction with sibling left. To better understand the interaction effects, we discussed the results by means of predicted probability plots [73]. As shown in Model 2 and Fig 2, the more similar the siblings were regarding their level of extraversion, the more likely a child was to leave following their sibling’s departure (b = 0.144, SE = 0.064). In other words, sibling similarity in extraversion was positively associated with similarity in their timing of leaving the parental home, confirming H1a. When a sibling did not leave the nest, similarity in extraversion was linked to a longer stay in the parental home. Although the effect of openness was not statistically significant (b = -0.103, SE = 0.060), we observed from the plotted interaction of predicted probabilities (Fig 3) that, if siblings were dissimilar in their level of openness, the effect of sibling left was stronger, which contradicts H4a.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The association between siblings’ departures as moderated by siblings’ similarity in extraversion.

The x-axis represents how siblings are similar in terms of their level of extraversion, and a value of 6 indicates that they are highly similar.

https://doi.org/10.1371/journal.pone.0284808.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. The association between siblings’ departures as moderated by siblings’ similarity in openness.

The x-axis represents how siblings are similar in terms of their level of openness, and a value of 6 indicates that they are highly similar.

https://doi.org/10.1371/journal.pone.0284808.g003

In Model 3, we further examined whether the effect of personality similarity among siblings may differ depending on the level of extraversion, agreeableness, openness, and neuroticism. It was possible that the association between siblings’ departures is only strengthened or weakened on one end of the spectrum (e.g., both of them are highly agreeable). To test this, three-way interaction effects were modeled using at-risk children’s Big Five traits, those of their siblings, and the indicator of whether the sibling had left or not. As shown in Model 3 and Fig 4, a significant relationship between sibling left, the at-risk child’s level of extraversion, and the sibling’s level of extraversion was found (b = 0.102, SE = 0.046). When the sibling of an at-risk child had left the parental home, the child was most likely to leave when he/she was an introvert and when the sibling was also an introvert. To understand what introversion entails in this case, we explored different introversion cut-off scores. We found that introverted siblings were more likely to leave home together when they both were above average introverts. If the sibling was introverted and the child was extraverted, the likelihood of leaving decreased. Similarly, introverted children were not motivated by their extraverted siblings’ departure. When both of them were extraverted, the probability of leaving was higher than in dissimilar pairs, but still lower than for two introverts. This supports H1b that the relationship between siblings’ departures is particularly strong when they are both introverts. In addition, if both the child at risk and the sibling lived at home, the child’s risk of departure was the highest if they were extraverted and the sibling was introverted. Given that we did not find a significant three-way interaction effect for agreeableness, neuroticism, and openness, H2, H3, and H4b were not confirmed.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. The association between siblings’ departures as moderated by child’s extraversion and sibling’s extraversion.

(Predicted probabilities for event = 1, with 95% confidence limits).

https://doi.org/10.1371/journal.pone.0284808.g004

Robustness check and supplementary analysis

For the first robustness check, the models estimating a three-level data structure (excluding sibling dyad level) yielded almost identical results compared to the main analysis (see Table S2 in the S1 File). For the second robustness check using FE models, although the intercept of sibling left was shifted to a lower level and its effect became negative, the extent to which similarity in siblings’ personality moderated the effect was similar to that in Model 2. The three-way interaction effects in the FE model were also in line with those in Model 3 (see Table S3 in the S1 File).

As discussed, agreeableness predicts better interpersonal relationships. To further ensure that the effects found in the main analysis were not confounded by sibling relationship quality, we repeated the analysis based on sub-samples of agreeable and unagreeable dyads (agreeableness scores = 5 being the cutoff point; see supplementary analysis 4 in the S1 File). For both agreeable and unagreeable sibling dyads, the interaction effect of sibling similarity in extraversion was close to the significant level of 0.05, which is likely due to the smaller sample size. However, the direction and magnitude of the effect remained the same, that sibling similarity in extraversion positively moderated sibling transmission of parental home leaving. As for the three-way interaction effect of extraversion, the results of agreeable dyads showed high resemblance to those of Model 3 in Table 2, whereas no effect was found for unagreeable dyads.

Furthermore, we controlled for parental agreeableness, a proxy for parent-child relationship quality, and regions in which the family resided on a reduced sample as a sensitivity check (see supplementary analysis 5 in the S1 File). They largely mimicked the main analysis, confirming that our findings are robust. Finally, we explored whether there was any non-linear effects of the Big Five traits. The results can be found in Fig S1 in the S1 File.