Joint Effect of Workplace Gender Discrimination and Gender Composition on the Mental Well-Being in Korean Workers

Abstract

Background

Despite growing evidence on the effects of workplace gender discrimination (WGD) and gender composition on workers’ mental well-being, little is known about how these two factors interact to shape overall well-being. This study aims to investigate the joint effect of WGD and gender composition on the risk of poor well-being among Korean workers.

Methods

We used data from the 6th (2020) and 7th (2023) waves of the Korean Working Conditions Survey (n= 62,642). We assessed both multiplicative and additive interactions between WGD (no vs. yes) and gender composition (gender balanced vs. gender imbalanced) in relation to mental well-being (good vs. poor). Interaction on the multiplicative scale was evaluated using the multiplicative measure along with p value, whereas interaction on the additive scale was assessed using the relative excess risk due to interaction, the attributable proportion, and the synergy index.

Results

WGD was strongly associated with poor well-being, with a higher risk observed among workers in gender-imbalanced workplaces [odds ratio = 2.96, 95% confidence interval (CI): 2.38 to 3.69] than among those in gender-balanced workplaces (odds ratio = 1.38, 95% CI: 0.92 to 2.05). Evidence of both supramultiplicative (multiplicative measure = 2.09, 95% CI: 1.33 to 3.28, p value<0.01) and supra-additive effects (relative excess risk due to interaction = 1.56, attributable proportion = 0.53, S = 4.82) was found. These statistically significant effects remained among women, whereas no significance was observed among men.

Conclusion

The interaction between WGD and gender-imbalanced workplace environments was associated with poorer worker well-being. These findings underscore the importance of addressing both discriminatory practices and structural gender composition to address worker well-being.

1. Introduction

Mental health problems among workers in Korea have become increasingly serious, with their prevalence rising sharply in recent years. According to the Korean Working Conditions Survey (KWCS), fewer than 10% of workers reported depression or anxiety in 2014, whereas nearly half reported anxiety by 2020 [1]. Mental health problems, such as depression and anxiety, are strong predictors of suicidal ideation among workers [2] and can lead to substantial productivity loss [3,4].

Mental well-being, commonly measured using the World Health Organization Five Well-Being Index (WHO-5) [5], is associated with various mental, behavioral, and neurodevelopmental disorders, including depression and anxiety [6]. Moreover, well-being is essential not only for individual mental health but also for organizational functioning as it is closely linked to productivity and overall performance [7].

This growing mental health burden underscores the need to better understand workplace factors influencing well-being, and workplace gender discrimination (WGD) is one of the key workplace determinants. The Korea continues to exhibit one of the highest gender wage gaps among Organization for Economic Co-operation and Development member states, with a median gender wage gap of 29% in 2024. WGD contributes to an unequal work environment that increases the risk of mental disorders and suicidal ideation [[8], [9], [10]], and remains a significant challenge in contemporary labor markets as it adversely affects a wide range of work-related outcomes [11].

While WGD is a complex phenomenon shaped by various organizational structures, processes, and practices [12], workplace gender composition can be an important factor contributing to its emergence and may cause adverse health effects [13,14]. Women in male-dominated workplaces may face heightened risks of gender discrimination as men may try to maintain their existing privileges and perceive a growing female presence as a threat to their power [15]. Moreover, gender composition has been linked to employee well-being, with gender-balanced workplaces associated with higher levels of job satisfaction and organizational commitment [16].

Despite growing evidence on the effects of WGD and gender composition on workers’ mental well-being, little is known about how these two factors interact to shape overall well-being. Therefore, the aim of this study is to examine the interaction effect of WGD and gender composition on the risk of poor well-being among Korean workers using nationally representative survey data.

2. Methods

2.1. Study populations

This study is based on data from the 6th and 7th waves of the KWCS as a question regarding workplace gender composition was first introduced in the 6th wave. The 6th and 7th waves of the KWCS were conducted in 2020 and 2023, respectively. The KWCS targeted employed individuals aged 15 years or older and was based on the original questionnaire of the European Working Conditions Survey developed by Eurofound.

The study population initially included 100,733 participants. We excluded individuals unsuitable for the purposes of this study, including (1) unpaid family workers, (2) workers in one-person workplaces, and (3) participants whose occupations were classified as “armed forces” in the 7th KWCS. Participants who responded “don’t know/no response” or “refused” for any variables used in this study were additionally excluded, resulting in a final analytic sample of 62,642 participants.

2.2. Study variables

Mental well-being was measured using the WHO-5. Participants were asked to indicate the option that best described how they had been feeling over the past two weeks. The five statements were “I have felt cheerful and in good spirits,” “I have felt calm and relaxed,” “I have felt active and vigorous,” “I woke up feeling fresh and rested,” and “My daily life has been filled with things that interest me.” Each statement was rated on a 6-point scale ranging from 0 (“at no time”) to 5 (“all of the time”). The total score ranges from 0 to 25. Participants with a total score below 13 were classified as having poor well-being.

WGD was assessed with a yes-or-no question: “Over the past 12 months, have you been discriminated against at work? (If you have worked for less than one year, please respond based on the period since you started your main job.)”

Workplace gender composition was assessed with the question, “Approximately how many of them are female in your company or organization?” In the 6th wave of the survey, respondents chose from five options: “none or almost none,” “less than half,” “about half,” “more than half,” and “all or almost all.” In the 7th wave, the options were revised to three options: “more men than women,” “more women than men,” and “a similar proportion of men and women.” For analysis, gender composition was dichotomized into gender-balanced workplaces (“about half” in the 6th wave and “a similar proportion of men and women” in the 7th wave) or gender-imbalanced workplaces (all other responses).

Covariates included gender, age, education, occupation, industry, workplace size, monthly income, and actual working hours. Gender was categorized as male or female. Age was grouped into <30 years, 30–39 years, 40–49 years, 50–59 years, and ≥60 years. Education was categorized as below high school, high school graduate, and above high school. Occupation type was surveyed and classified into nine groups in the 6th KWCS—(1) managers, (2) professionals, (3) technicians and associate professionals, (4) clerical support workers, (5) service and sales workers, (6) skilled agricultural, forestry, and fishery workers, (7) craft and related trades workers, (8) plant and machine operators and assemblers, and (9) elementary occupations—and into 10 groups in the 7th KWCS, with the additional category of (10) armed forces. These occupational groups were subsequently categorized into white (groups 1–4), pink (groups 5–6), and blue collar groups (groups 7–9), while participants classified as armed forces in the 7th KWCS were excluded from the analysis [17]. Industry type, surveyed according to the Korean Standard Industrial Classification, was also categorized into white, pink, and blue collar groups. Workplace size was grouped into 2–9 employees, 10–249 employees, and ≥250 employees. Actual working hours were calculated by summing hours from the main and secondary jobs and categorized into <40 hours, 40–52 hours, and >52 hours per week.

Options such as “don’t know/no response” or “refused” were provided for the questions, and these responses were treated as missing values and excluded from the analytic sample.

2.3. Statistical analysis

Logistic regression models were used to examine the association of WGD and gender composition with poor well-being, and analyses were stratified by gender.

Multiplicative interaction was evaluated using the multiplicative interaction measure along with p values. The multiplicative interaction measure refers to the ratio of odds ratios (ORs), defined as the OR for joint exposure divided by the product of the ORs for each individual exposure. Values greater than 1 indicate supramultiplicativity, values equal to 1 indicate no multiplicative interaction, and values less than 1 indicate submultiplicativity, with statistical significance assessed using a p-value threshold of 0.05 [18].

Additive interaction was assessed using the relative excess risk due to interaction (RERI), the attributable proportion (AP) due to interaction, and the synergy index (S), with corresponding 95% confidence intervals (CIs). RERI represents the additional risk caused by the interaction compared with the baseline risk without exposure. AP reflects the proportion of the combined effect that arises from the additive interaction. S reflects the relative magnitude of the joint effect compared with what would be expected from the individual effects. Under no additive interaction (i.e., exact additivity), RERI = 0, AP = 0, and S = 1, whereas supra-additive interaction is suggested when RERI >0, AP > 0, and S > 1 [19].

Estimates were adjusted for gender, age, education, occupation, industry, workplace size, monthly income, and actual working hours. Sampling weights provided by the KWCS were applied to obtain nationally representative estimates.

All data management and statistical analyses were conducted using R software version 4.5.0 (R Foundation for Statistical Computing, Vienna, Austria).

3. Results

Table 1 summarizes the general characteristics of the study population. Of the 62,642 participants, 31,947 (51.0%) were women and 30,695 (49.0%) were men. Of those reporting poor well-being, 50.8% were women and 49.2% were men.

Table 1.

Characteristics of the study population

Variables	Total n = 62,642 n (%)	Good well-being n = 43,805 n (%)	Poor well-being n = 18,837 n (%)	p∗
Gender
Female	31,947 (51.0)	22,372 (51.1)	9575 (50.8)	0.59
Male	30,695 (49.0)	21,433 (48.9)	9262 (49.2)
Age (years)
15–29	6883 (11.0)	5217 (11.9)	1666 (8.8)	<0.01
30–39	12,706 (20.3)	9405 (21.5)	3301 (17.5)
40–49	14,253 (22.8)	10,244 (23.4)	4009 (21.3)
50–59	15,669 (25.0)	10,849 (24.8)	4820 (25.6)
≥60	13,131 (21.0)	8090 (18.5)	5041 (26.8)
Education
Below high school	6558 (10.5)	3576 (8.2)	2982 (15.8)	<0.01
High school	21,059 (33.6)	14,031 (32.0)	7028 (37.3)
Above high school	35,025 (55.9)	26,198 (59.8)	8827 (46.9)
Occupation type
White	35,216 (56.2)	25,780 (58.9)	9436 (50.1)	<0.01
Pink	9252 (14.8)	6507 (14.9)	2745 (14.6)
Blue	18,174 (29.0)	11,518 (26.3)	6656 (35.3)
Industry type
White	24,971 (39.9)	17,605 (40.2)	7366 (39.1)	0.03
Pink	22,600 (36.1)	15,684 (35.8)	6916 (36.7)
Blue	15,071 (24.1)	10,516 (24.0)	4555 (24.2)
Workplace size (number of employees)
2∼9	29,976 (47.9)	20,381 (46.5)	9595 (50.9)	<0.01
10∼249	27,909 (44.6)	19,866 (45.4)	8043 (42.7)
≥250	4757 (7.6)	3558 (8.1)	1199 (6.4)
Monthly income (10,000 KRW)
<200	15,207 (24.3)	9553 (21.8)	5654 (30.0)	<0.01
200∼299	20,344 (32.5)	14,255 (32.5)	6089 (32.3)
300∼399	14,763 (23.6)	10,713 (24.5)	4050 (21.5)
≥400	12,328 (19.7)	9284 (21.2)	3044 (16.2)
Actual working hours (hours/week)
<40	12,206 (19.5)	7854 (17.9)	4352 (23.1)	<0.01
40∼52	44,844 (71.6)	32,357 (73.9)	12,487 (66.3)
>52	5592 (8.9)	3594 (8.2)	1998 (10.6)

^∗p value represents differences in proportions using Pearson χ² test.

Table 2 and Fig. 1 show the ORs of poor well-being with WGD and gender composition. The reference group was workers without experience of WGD in gender-balanced workplaces.

Table 2.

ORs of poor well-being and 95% CIs with workplace gender discrimination (WGD) and gender composition

Gender (n)	WGD	Workplace gender composition	Good well-being n	Poor well-being n	OR (95% CI)
Total	No	Gender balanced	9346	3900	1.00 (reference)
(n= 62,642)	No	Gender imbalanced	33,932	14,453	1.03 (0.97 to 1.10)
	Yes	Gender balanced	127	77	1.38 (0.92 to 2.05)
	Yes	Gender imbalanced	400	407	2.96 (2.38 to 3.69)
Female	No	Gender balanced	4788	1932	1.00 (reference)
(n = 31,947)	No	Gender imbalanced	17,234	7310	1.07 (0.98 to 1.17)
	Yes	Gender balanced	92	55	1.36 (0.88 to 2.11)
	Yes	Gender imbalanced	258	278	3.04 (2.36 to 3.92)
Male	No	Gender balanced	4558	1968	1.00 (reference)
(n= 30,695)	No	Gender imbalanced	16,698	7143	1.00 (0.91 to 1.10)
	Yes	Gender balanced	35	22	1.39 (0.62 to 3.10)
	Yes	Gender imbalanced	142	129	2.96 (2.02 to 4.33)

Gender, age, education, occupation, industry, workplace size, monthly income, and actual working hours were adjusted in the model.

CI, confidence interval; OR, odds ratio; WGD, workplace gender discrimination.

Fig. 1.

The Joint effect of workplace gender discrimination (WGD) and gender composition on mental well-being. Workplace gender discrimination was coded as 0 = no and 1 = yes. Workplace gender composition was coded as 0 = gender-balanced workplace and 1 = gender-imbalanced workplace. Gender, age, education, occupation, industry, workplace size, monthly income, and actual working hours were adjusted in the model.

Among total workers without WGD, workplace gender composition was not independently associated with poor well-being, and this association was not statistically significant (OR = 1.03, 95% CI: 0.97 to 1.10). However, the interaction between WGD and gender-imbalanced workplace environments was associated with markedly higher odds of poor well-being (OR = 2.96, 95% CI: 2.38 to 3.69). These patterns were consistent among women in the gender-stratified analyses, suggesting that the adverse impact of WGD may be manifested in gender-imbalanced workplaces.

Table 3 presents the multiplicative and additive interactions between WGD and gender imbalance in relation to mental well-being. Among total workers, the multiplicative interaction indicated a supramultiplicative effect, with a multiplicative measure of 2.09 (95% CI: 1.33 to 3.28, p< 0.01), demonstrating that the joint effect of WGD and gender-imbalanced workplaces on poor well-being was 2.09 times greater than would be expected if the two exposures acted independently on a multiplicative scale. Supra-additive interaction was also evident. The RERI was 1.56 (95% CI: 0.73 to 2.39), indicating that the joint effect of WGD and gender-imbalanced workplaces on poor well-being exceeded the sum of their individual effects by 1.56 on the additive scale. The AP was 0.53 (95% CI: 0.32 to 0.74), suggesting that approximately 53% of the joint effect of WGD and gender-imbalanced workplaces on poor well-being was attributable to their additive interaction. The S was 4.82 (95% CI: 1.19 to 19.48), further indicating that the joint effect was nearly five times greater than expected under an additive model. Overall, these findings indicate that the joint effect of WGD and gender-imbalanced workplaces on poor well-being was greater than would be expected based on the individual effects of each factor alone on both the multiplicative and additive scales.

Table 3.

Measure of multiplicative and additive interactions between workplace gender discrimination and gender composition in relation to mental well-being

Gender (n)	Multiplicative interaction		Additive interaction
	Multiplicative measure (95% CI)	p for interaction	RERI (95% CI)	AP	S
Total (n= 62,642)	2.09 (1.33 to 3.28)	<0.01	1.56 (0.73 to 2.39)	0.53 (0.32 to 0.74)	4.82 (1.19 to 19.48)
Female (n= 31,947)	2.09 (1.27 to 3.46)	<0.01	1.61 (0.67 to 2.56)	0.53 (0.30 to 0.76)	4.77 (1.10 to 20.62)
Male (n= 30,695)	2.13 (0.88 to 5.18)	0.10	1.57 (0.00 to 3.13)	0.53 (0.11 to 0.95)	5.05 (0.26 to 96.79)

Gender, age, education, occupation, industry, workplace size, monthly income, and actual working hours were adjusted in the model.

AP, attributable proportion; CI, confidence interval; RERI, relative excess risk due to interaction; S, synergy index.

When analyses were stratified by gender, both supramultiplicative (multiplicative measure = 2.09, 95% CI: 1.27 to 3.46, p< 0.01) and supra-additive effects (RERI = 1.61, 95% CI: 0.67 to 2.56; AP = 0.53, 95% CI: 0.30 to 0.76; S = 4.77, 95% CI: 1.10 to 20.62) remained statistically significant among women. In contrast, among men, no statistically significant association was observed.

4. Discussion

The objective of the study was to investigate the joint effect of WGD and gender composition on the risk of poor well-being. We found both supramultiplicative and supra-additive joint effects between WGD and gender-imbalanced workplaces, indicating that gender-imbalanced environments may manifest the detrimental impact of WGD on poor well-being. These joint effects may be explained by the unique psychosocial dynamics that characterize gender-imbalanced workplaces. Individuals in numerical minority positions often lack peers who share similar experiences, leading to reduced social support and fewer opportunities for validation or collective coping, both of which are important mediating factors between discrimination and well-being [20]. This characteristic can also be explained by the concept of tokenism, wherein workers in numerical minority positions (“tokens”) are known to experience greater work-related stress, heightened psychological symptoms, and increased role overload [21]. In addition, a discriminatory work environment can lead to reduced supervisor support, thereby exacerbating employees’ well-being [22]. The absence of supportive structures and the heightened visibility associated with being a minority may amplify the psychological burden of discriminatory experiences. As a result, gender discrimination may be perceived as more threatening, isolating, and difficult to challenge in gender-imbalanced workplaces, where minority workers may suffer more discrimination but have fewer supportive peers or structural resources, ultimately amplifying its adverse impact on well-being.

When stratified by gender, the adverse association persisted among women, while it did not reach statistical significance among men. It is important to note, however, that the lack of statistical significance in men was likely driven by limited precision, as reflected by the wide confidence intervals in the male-stratified analyses, rather than a true absence of association. The estimates among men were consistently in the positive direction, and caution is warranted in interpreting these null findings as evidence that WGD has no detrimental effect on men’s well-being. This may be attributable to the substantially smaller number of men who reported experiencing WGD than women, resulting in insufficient statistical power to detect significant effects. Additionally, workplace gender composition may operate differently for men and women. Bryngelson et al. reported that gender-imbalanced environments were associated with long-term sickness absence among women, whereas no such association was observed among men [23]. Men tend to occupy higher-status positions, interact more frequently with higher-ranking male colleagues, and are more likely to be treated as equals by their male peers, which may further buffer them from the negative consequences experienced by numerical minority groups. A similar pattern has been documented in a study examining occupational gender composition and mental health outcomes [24]. While women working in male-dominated environments often face the glass ceiling, an invisible barrier that hinders their advancement, men in female-dominated occupations may benefit from the glass escalator, which encourages the rapid and disproportionate promotion of men [25,26]. In the specific Korean occupational context, these dynamics may operate particularly strongly. Korea’s highly hierarchical and male-dominated organizational culture may allow men in female-dominated workplaces to benefit from the glass escalator effect, thereby buffering them from the detrimental impact of WGD on well-being [27]. In contrast, the glass-ceiling effect and tokenism framework predict that women in male-dominated environments experience heightened scrutiny, role overload, and isolation, consistent with our finding of significantly amplified associations among women [[28], [29], [30]]. These mechanisms are likely reinforced by Korea’s persistent gender wage gap and entrenched occupational segregation. Further empirical studies are needed to more precisely examine the differences between men and women.

Several limitations should be considered. First, the cross-sectional design limits causal interpretation, and reverse causation cannot be excluded. Workers with pre-existing poor mental well-being may be more likely to perceive or report workplace experiences as discriminatory, potentially inflating the observed associations. All policy and intervention recommendations should therefore be interpreted as potential avenues for further investigation rather than direct causal conclusions. Second, variables were assessed through self-reported measures, which may be subject to reporting bias or common method variance. WGD was assessed using a single self-reported yes/no item, which may not fully capture the multidimensional nature of discriminatory experiences, including its type, frequency, severity, or chronicity. This limitation may result in exposure misclassification, which is likely to be nondifferential and could bias the observed associations. However, this measure of WGD is commonly used in large-scale working-condition surveys, such as the KWCS, European Working Conditions Survey, and the Organization for Economic Co-operation and Development Employee Well-being Survey, and has been widely used in studies utilizing these datasets [9,31,32]. Future research should utilize validated, multi-item scales that better capture the multidimensional nature of WGD. Similarly, workplace gender composition was assessed using a single subjective item. Perceptions of what constitutes a “similar” gender ratio may vary across respondents, potentially introducing measurement error. Furthermore, to enhance the robustness of the analysis by maximizing the available sample size, data from the 6th and 7th waves of the KWCS were combined. However, the response scale used to assess workplace gender composition differed between the two waves, changing from a five-point scale to a three-point scale. Although both waves were dichotomized using a consistent criterion (i.e., “about half” or “a similar proportion” as gender balanced), the change in scale may have introduced heterogeneity in how respondents conceptualized the threshold. However, wave-stratified analyses showed consistent interaction patterns across both waves, supporting the robustness of the dichotomization approach, though residual differences in respondent perception cannot be fully excluded (Supplementary Table 1). Third, to ensure sufficient statistical power and enhance the robustness of our analysis, we combined data from the 6th and 7th waves of the KWCS. Although the same individuals may have been included in both waves, this likelihood is low because the KWCS samples approximately 50,000 households nationwide through regionally stratified sampling based on the Population and Housing Census. Fourth, although we adjusted for several sociodemographic and occupational factors, key psychosocial work variables were not incorporated into the adjustment models, including high job demands, low job control, low social support, and exposure to other forms of workplace harassment. These factors are plausibly associated with both WGD and mental well-being, and their omission may result in residual confounding. These unmeasured psychosocial factors are likely positively associated with both WGD and poor mental well-being, suggesting that residual confounding would bias the observed associations toward overestimation [[33], [34], [35]]. Therefore, the effect of WGD on mental well-being after adjusting psychosocial work variables may be somewhat attenuated compared to our estimates. Fifth, the use of ORs to derive additive interaction measures requires careful interpretation when outcomes are relatively common as ORs may overestimate risk ratios in such settings. However, a sensitivity analysis using robust Poisson regression yielded consistent results, supporting the robustness of the reported interaction estimates (Supplementary Table 2). These limitations should be considered when interpreting our findings.

To the best of our knowledge, this is the first study to examine the interaction between WGD and workplace gender composition in relation to mental well-being, demonstrating both supramultiplicative and supra-additive effects. Using nationally representative data from the KWCS, our study benefits from a large sample size and a rigorous analytic approach, which strengthens the robustness of our findings. Taken together, these results highlight that improving worker well-being requires not only addressing discriminatory practices but also fostering structurally gender-balanced workplaces. If achieving gender balance is not feasible, efforts to reduce discriminatory practices, provide positive gender-relevant role models, and strengthen social support systems in traditionally male-dominated environments may be helpful [36].

5. Conclusion

WGD was significantly associated with poor well-being, and its association appeared to interact with that of gender-imbalanced workplaces on both multiplicative and additive scales. These findings suggest that attention should be given not only to discriminatory practices but also to the structural gender composition of workplaces as both may be relevant to worker well-being.

CRediT authorship contribution statement

Yeseo Lee: Writing – original draft, Investigation, Formal analysis, Data curation, Conceptualization. Jongin Lee: Writing – review & editing, Validation, Supervision, Methodology.

Patient consent for publication

None applicable.

Data availability statement

The data used in this study were obtained from the Korean Working Conditions Surveys, which are open to the public, available on https://oshri.kosha.or.kr/eoshri/resources/KWCSDownload.do.

Ethics statement

The data used in our study were drawn from the 6th and 7th waves of the Korean Working Conditions Survey, which are publicly available and provided with all personally identifiable information removed. The study design and evaluation protocols were approved by the institutional review board of Seoul St. Mary’s Hospital, the Catholic University of Korea (Approval No. KC25ZASI0850).

Funding

There are none to declare.

Conflicts of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article: