Perceived stress (PS) is defined as the feelings or thoughts of individuals about the amount of stress to which they are exposed over a given time period (Cohen et al., Reference Cohen, Kamarck and Mermelstein1983). PS broadly involves frequencies of exposure to stressful events and appraisal of and ability to cope with such stressful events (Cohen et al., Reference Cohen, Kamarck and Mermelstein1983). Individuals greatly differ in the perception and adaptation to aversive and stressful stimuli (Ebner & Singewald, Reference Ebner and Singewald2017). Previous research demonstrates that PS is significantly associated with personality traits, such that extraversion, conscientiousness, and agreeableness were negatively related to PS, but neuroticism was positively related to PS (Ebstrup et al., Reference Ebstrup, Eplov, Pisinger and Jørgensen2011). PS is also a significant predictor of physical health (Baum & Posluszny, Reference Baum and Posluszny1999), mental health (Bjørndal et al., Reference Bjørndal, Kendler, Reichborn-Kjennerud and Ystrom2023; Kendler & Gardner, Reference Kendler and Gardner2010), work–life conflict (Bell et al., Reference Bell, Rajendran and Theiler2012), and wellbeing (Bolger et al., Reference Bolger, DeLongis, Kessler and Schilling1989).

Many researchers have pointed to gender differences in terms of level of stress and responses to stressful life events (Graves et al., Reference Graves, Hall, Dias-Karch, Haischer and Apter2021; Kendler et al., Reference Kendler, Thornton and Prescott2001). Young adult women tended to report higher levels of stress than their male counterparts (Graves et al., Reference Graves, Hall, Dias-Karch, Haischer and Apter2021). Furthermore, men and women varied in PS by category. Kendler et al. (Reference Kendler, Thornton and Prescott2001) illustrated that women reported more interpersonal, while men more legal and work-related stressful life events, and that women were more sensitive to stress due to social relationships, while men were more sensitive to stress due to loss of wealth and social status. Although the results were mixed, men were frequently engaged in problem-focused coping strategies, whereas women typically utilized emotion-focused coping strategies (Eaton & Bradley, Reference Eaton and Bradley2008; Tamres et al., Reference Tamres, Janicki and Helgeson2002).

Cultural psychologists have proposed that culture significantly influences one’s perception of stressful situations (Lee et al., Reference Lee, Masuda, Ishii, Yasuda and Ohtsubo2023). North Americans, who mainly belong to individualistic cultures, tend to perceive themselves as independent from others and prioritize independent goals and values, such as personal accomplishments, when evaluating situations. Conversely, East Asians, who belong to collectivistic cultures, tend to view themselves as embedded in relationships with others and prioritize interdependent values and goals, such as social harmony, when evaluating situations (Lee et al., Reference Lee, Masuda, Ishii, Yasuda and Ohtsubo2023). In support of this perspective, Hashimoto et al. (Reference Hashimoto, Mojaverian and Kim2012) found that Japanese undergraduates perceived stressful interpersonal situations as more frequent, which was associated with increased psychological distress, compared with American undergraduates. Lee et al. (Reference Lee, Masuda, Ishii, Yasuda and Ohtsubo2023) extended the study of Hashimoto et al. (Reference Hashimoto, Mojaverian and Kim2012) to include stressful noninterpersonal situations. The results revealed that European Canadian undergraduates perceived stressful noninterpersonal situations as more frequent compared with Japanese undergraduates. Alternatively, the Japanese undergraduates perceived stressful interpersonal situations as more frequent than did the European Canadian ones. Moreover, Lee et al. demonstrated that the European Canadian undergraduates perceived more psychological distress when envisioning themselves experiencing stressful noninterpersonal situations over interpersonal ones. Meanwhile, the Japanese undergraduates perceived more psychological distress when imagining themselves in interpersonal situations instead of noninterpersonal ones.

Prior twin studies have demonstrated genetic influences on PS. Federenko et al. (Reference Federenko, Scholotz, Kirschbaum, Bartels, Hellhammer and Wust2006) assessed PS, stress susceptibility, and six perceived chronic stress scales in German adult twins and found that heritability ranged from 5% to 45%, whereas shared environmental influences ranged from 0% to 22%. However, caution must be exercised in interpreting the results due to the small sample size (total N = 180 pairs). Another small experimental twin study on young adult American twins showed that heritability of PS was 44% with the remaining variance attributable to nonshared environmental influences (Bogdan & Pizzagalli, Reference Bogdan and Pizzagalli2009). Similarly, in a relatively large sample of Australian adolescent and young adult twins, Rietschel et al. (Reference Rietschel, Zhu, Kirschbaum, Strohmaier, Wust, Rietschel and Martin2014) found that genetic and nonshared environmental influences on PS were 52% and 48% respectively.

According to the theory of gene–environment (GE) correlation (Plomin et al., Reference Plomin, DeFries and Loehlin1977), individuals play an active role in selecting and creating their environments. In support of GE correlation, twin studies have demonstrated that controllable events, in which the individual plays an active role, are substantially more heritable than uncontrollable events, in which the individual exerts little influence (Bemmels et al., Reference Bemmels, Burt, Legrand, Lacono and McGue2008; Bolinskey et al., Reference Bolinskey, Neale, Jacobson, Prescott and Kendler2004; Colodro-Conde et al., Reference Colodro-Conde, Couvy-Duchesne, Zhu, Coventry, Byrne, Gordon, Wright, Montgomery, Madden, Ripke, Eaves, Heath, Wray, Medland and Martin2018; Saudino et al., Reference Saudino, Pedersen, Lichtenstein, McClearn and Plomin1997). Using twins and adoptive siblings, Bemmels et al. (Reference Bemmels, Burt, Legrand, Lacono and McGue2008) found that additive genetic effects were the largest contributor to dependent, controllable life events (45%), whereas nonshared environmental effects were the largest contributor to uncontrollable, independent life events (57%). Shared environmental effects were the largest (71%) in the variance of familial life events (shared by members of a family). Bolinskey et al. (Reference Bolinskey, Neale, Jacobson, Prescott and Kendler2004) administered the stressful life events exposure scale (SLEs) to a large sample of American adult twins (N = 3938 pairs). The authors estimated genetic and environmental influences by category (personal vs. network events) and gender. Parameter estimates were similar across genders in the total SLEs. Specifically, genetic effects explained approximately 25% of variance in the total SLEs, while shared environmental effects accounted for only 10%–13% of variance. Nonshared environmental effects ranged from 61% to 65%. In personal events (i.e., events that directly occur to the individual), genetic effects were greater than shared environmental effects, and both effects were nearly the same across men and women (genetic estimates: 29% vs. 28%; shared environmental estimates: 9% for both). In network events (i.e., events that occur to a person within one’s social network, which, thus, indirectly affects the individual), shared environmental effects (26% for both genders) were greater than genetic effects (9% for men, 21% for women).

One of the limitations of twin studies of PS is that nearly all studies have recruited American or European twins raised in societies with individualistic cultures. To the best of our knowledge, no studies have examined genetic and environmental influences on PS in twins raised in collectivistic cultures. Given cultural differences in PS, exploring genetic and environmental influences on PS in East Asians is important to generalize the findings of twin studies to those in collectivistic societies. The present study aimed to examine genetic and environmental influences on PS in late adolescents and young adults in South Korea. Using sex-limitation variance component model-fitting analysis, we examined whether or not genetic and environmental effects on PS vary by gender and category.

Materials and Methods

Sample

The sample included 1372 twins who responded to an online survey conducted by the Korea Twin Research Institute (KTRI) in 2022–2023. The ages of the twins ranged from 16 to 27 years (mean ± SD = 22.4 ± 2.5 years). The survey link was uploaded to online communities in various universities and websites of the KTRI and twin clubs in South Korea. A gift coupon was sent to twins who completed the survey. The sample consisted of 286 (142 complete pairs and 2 cotwin missing individuals) monozygotic male (MZM), 91 (45 complete pairs and 1 cotwin missing individual) dizygotic male (DZM), 518 (258 complete pairs and 2 cotwin missing individuals) monozygotic female (MZF), 227 (113 complete pairs and 1 cotwin missing individual) dizygotic female (DZF), and 250 (125 complete pairs) opposite-sex dizygotic (OSDZ) twins. Women outnumbered men (63% vs. 37%) in the sample, partially because military service is mandatory for South Korean young adult men. Moreover, women tend to respond to online surveys more frequently than males (Wu et al., Reference Wu, Zhao and Fils-Aime2022).

The study determined the zygosity of twins using four questions on physical similarity: frequency of confusion about twins by family members, friends, or relatives; and self-perception of zygosity, adopted from the zygosity questionnaire developed by Ooki et al. (Reference Ooki, Yamada and Asaka1993). MZ twins exceeded DZ twins (59% vs. 41%), which likely reflected the twin birth rates in South Korea in the 1990s and early 2000s (Hur, Reference Hur2021).

Measure

We used the Life Stress Scale (LSS; Chon, Reference Chon1998) to assess PS across the past 6 months. Originally, the LSS included 20 items for measuring stress levels of late adolescents and young adults across categories. We performed factor analysis and extracted five factors, namely, Friendship (five items), Family Conflicts (four items), Family Financial Difficulty (FFD; two items), Future Career (two items), and Academic stress (six items). We excluded one item (‘I had conflicts with my lover’), because it did not substantially load on any of the five factors. Items were rated using a 4-point Likert-type scale ranging from 0 = not at all true to 3 = certainly true. The score for each scale was computed by summing the responses. We also summed the scores across the 19 items to produce the total score for PS. The internal consistency reliabilities (Cronbach’s alpha) of the five scales ranged from .74 for FFD to .89 for Future Career (Table 1). Cronbach’s alpha of the total scale reached .92.

Table 1.

Items of the Perceived Stress Scale

Statistical Analysis

Using the raw data option in Mx (Neale et al., Reference Neale, Boker and Xie2003), we computed the maximum likelihood correlations for the five zygosity groups of twins and performed the general sex-limitation model-fitting analysis (Neale & Cardon, Reference Neale and Cardon1992) to determine additive genetic, shared environmental, and nonshared environmental influences on PS. Specifically, we modeled the phenotypic variance for each PS scale and the total PS scale as a function of an additive genetic variance component (A), a shared environmental variance component (C), and nonshared environmental variance component, including measurement error (E). Twin correlations for A were 1.0 and 0.5 for MZ and same-sex DZ twins respectively. For OSDZ twins, however, we allowed the correlation for A to vary between 0 and .5 with the assumption that genes for PS may be qualitatively different between men and women. To test this qualitative sex difference in genes for PS, we set the additive genetic correlation for OSDZ twins to .5 and examined the resultant model-fit statistics. Twin correlations for C were 1.0 for MZ and same-sex DZ twins. For OSDZ twins, we allowed the correlation for C to vary between 0 and 1.0 with the assumption that shared environmental effects for PS may be qualitatively different between men and women. To test for this qualitative sex difference in shared environmental influences on PS, we set the shared environmental correlation for OSDZ twins to 1.0, and examined the resultant model-fit statistics. Because we cannot test qualitative sex difference in genetic and shared environmental effects simultaneously, we tested them separately. Twins were not correlated for E. For correlational and model-fitting analyses, the study considered age as a covariate to control for its main effect. Additionally, we allowed the A, C, and E components to differ between men and women (Am ≠ Af, Cm ≠ Cf, and Em ≠ Ef) with the assumption that the magnitudes of additive genetic and shared and nonshared environmental influences on PS may quantitatively differ between the two genders. To test for this quantitative sex difference in the A, C and E components, we constrained the three variance components to be equal across men and women and examined the resultant model-fit statistics.

The raw data option in Mx calculates twice the negative log likelihood (−2LL) of the data. As the difference in −2LL is chi-square distributed with degrees of freedom equal to the difference in the degrees of freedom, we used the log likelihood ratio test to evaluate the significance of the constraint when two models were nested. A nonsignificant change in chi-square between the full and constrained models indicates that the reduction in parameter is acceptable, while a significant change indicates that the parameter should be retained in the model. When two models were not nested, we utilized Akaike information criterion (AIC; Akaike, Reference Akaike1987) to evaluate the models. AIC quantifies the information content of a model in terms of the joint criterion of fit and parsimony (Akaike, Reference Akaike1987). Thus, the smaller the AIC, the better the fit of the model to the data.

Results

Descriptive Statistics

Supplementary Table 1 presents the means and standard deviations of all PS scales. Except for Academic Stress and Future Career, the means were significantly higher for women than for men. However, the differences were small (Cohen’s d < .30). Variances were also significantly higher for women for Friendship, Future Career, and FFD. Means were significantly higher in MZ than in DZ twins in a few scales. Again, however, the differences were small (Cohen’s d < .30). Variance differences between MZ and DZ twins were significant For Friendship and Family Conflicts. In general, the correlations of the PS scales with age were nonsignificant (r ≤ .05) except for Academic Stress (r = −.12).

Twin Correlations

Figure 1 depicts the correlations for the PS scales for the five zygosity groups of twins, and Table 2 shows the correlations and their 95% confidence intervals. For all scales, correlations for MZ twins were greater than those for DZ twins, which points to genetic influences on PS. However, correlations in FFD for DZ twins were very similar to those for MZ twins, which points to the importance of shared environmental influences. In general, correlations for OSDZ twins were not substantially lower than those for DZM or DZF twins, which indicates little evidence of the presence of sex-specific genes for PS. There were a number of hints of nonadditive genetic influences on Future Career for men, Friendship for women, and the total PS for men, because the correlations for DZ twins were much lower than half of those for MZ twins in these scales. But, because the estimates of dominance effects were not significant, only the additive genetic effects were reported.

Figure 1.

Maximum likelihood twin correlations for the Perceive Stress scales.

Note: MZM, monozygotic male twins; DZM, dizygotic male twins; MZF, monozygotic female twins; DZF, dizygotic female twins; OSDZ, opposite-sex dizygotic twins

Table 2.

Maximum likelihood twin correlations (95% CI) for Perceived Stress scales

Note: MZM, monozygotic male; DZM, dizygotic male; MZF, monozygotic female; DZF, dizygotic female; OSDZ, opposite-sex dizygotic twins; CI, confidence interval.

Model-Fitting Analysis

Supplementary Table 2 shows the results of model-fitting analysis. For all PS scales, AIC values were lower for the ACE than those for the ADE models. Thus, we selected ACE as the full model for all scales. Sex-specific genetic and environmental effects were not significant in any of the PS scales, because dropping sex-specific genes or shared environmental effects did not worsen the fit of the full models. Additionally, the difference in the magnitudes of A, C and E between men and women obtained statistical significance only in FFD, indicating that overall, genetic and environmental influences are similar across sexes. Table 3 presents the estimates of variance of the PS scales and their 95% confidence intervals (CIs) in the best-fitting models. For Friendship, Academic Stress, and Future Career, shared environmental influences were nonsignificant but additive genetic influences ranged from 57% (95% CI [51%, 62%]) for Future Career to 67% (95% CI [62%, 71%]) for Academic Stress. For Family Conflict, additive genetic and shared and nonshared environmental influences were all significant. However, among them, shared environmental influence was the largest (47%; 95% CI [30%, 60%]). Environmental influences entirely explained individual differences in FFD with shared environmental influence being much stronger than nonshared environmental influences for both gender groups. Although there were significant gender effects in the unstandardized variance components in FFD, the standardized variances of shared (men: 64%, women: 63%) and nonshared (men: 36%, women: 37%) environmental effects were nearly the same. For the Total PS scale, the variance components of additive genetic and shared environmental influences were similar (39% and 38%) in males and females. Nonshared environmental influences, including measurement error, accounted for 23% of variance for both gender groups.

Table 3.

Parameter estimates in the best-fitting general sex-limitation model for various Perceived Stress (PS) scales.

Note: A, additive genetic effects; C, shared environmental effects; E, nonshared environmental effects, including measurement error; M, male; F, female; –, fixed to zero.