How involuntary subordination and social support influence the association between self-esteem and depression: a moderated mediation model

The ethics committee of the School of Public Health of Shanghai Jiao Tong University approved the study (SJUPN-201302). This cross-sectional study was conducted from June 2017 to December 2017 with young migrant workers at industrial factories in Shanghai, one of the biggest cities in China. One factory located in the city center and another located in a suburban area were selected with the assistance of local centers for disease control and prevention. With the help of factory managers, convenience sampling was conducted to select individuals who met the inclusion criteria from each factory. Individuals were eligible if they were aged 18–25 years and did not have local household registrations.

We reviewed previous literature and found that reported standard deviations of scores on the Center for Epidemiologic Studies Depression Scale (CES-D) for young Chinese migrants varied from 7.2 to 9.8 [6]. Assuming an estimated standard deviation of 9.5, a distance from the mean to the limits of 0.8, and α of 0.05, we calculated a required sample size of 545. We recruited participants until the required sample size was reached. A total of 830 young Chinese migrants were recruited; 214 of those declined to participate. Finally 572 participants completed the study questionnaire adequately for further analysis.

Data were collected using an anonymous, self-report questionnaire. We obtained written informed consent from all participants following clarification of the study objectives and procedure, as well as the potential risks and benefits of participation. Participants completed the questionnaire in a private room. One investigator was present to answer any questions. The questionnaire was completed in approximately 30 min and respondents were given 50 Chinese yuan (approximately 7 United States dollar) for participation after satisfactorily completing the survey.

SPSS 22.0 (IBM Corp. Released 2013. IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY, USA) was used for data cleaning, coding, and preliminary analysis. In cases where data from a participant were missing for a particular variable, the conservative approach of listwise deletion was used, and the participant’s data were completely excluded from the analysis. Descriptive statistics including medians and interquartile ranges (Q) were used for continuous variables, as the main study variables were not normally distributed. Numbers and percentages were used for binary and categorical variables. Spearman correlation was used to evaluate the correlations between variables. The mediation model and moderated mediation model were analyzed using the PROCESS macro for SPSS [29]. All continuous variables were standardized and the interaction terms were computed from these standardized scores. The bias-corrected 95% confidence interval (CI) was calculated with 5000 bootstrapping samples. To identify the statistically significant region, or range of scores for which social support acted as a moderator, the Johnson–Neyman method suggested by Hayes [29] was used. The conditional effects and CI were plotted.

In total, 616 participants completed the questionnaires; 572 questionnaires remained after deletion owing to missing variables. Table 1 shows the participant characteristics. As discussed previously, all participants were young adults, with a mean age of 21.61 years (SD = 0.07, range 18.30–23.40). Most were male (61.89%); 64.34% had a highest education level of senior high school; 82.17% were unmarried; most of the sample had a monthly income of 3200–4800 Chinese yuan (64.51%); and 61.01% had been working in Shanghai for less than 1 year. There were no significant differences in depression scores by participant characteristics.

The mean (SD) scores for self-esteem, involuntary subordination, social support, and depression were 18.30 (3.99), 83.48 (15.66), 63.90 (12.38), and 19.18 (10.06), respectively. The median (Q) scores for self-esteem, involuntary subordination, social support, and depression were 18 (4), 82 (18), 66 (16), and 19 (14), respectively (Table 2).

Migrants with higher levels of self-esteem and higher levels of social support were more likely to have lower levels of involuntary subordination and depression. Those who had higher levels of involuntary subordination were more likely to have depressive symptoms.

Table 3 shows the mediating role of involuntary subordination in the association between self-esteem and depression. In Model 1 and Model 2, self-esteem had a negative effect on both depression (β = − 1.1465, p < 0.001) and involuntary subordination (β = − 2.5153, p < 0.001). In Model 3, self-esteem had a negative effect (β = − 0.4394, p < 0.01) on depression, and involuntary subordination (β = 0.2811, p < 0.001) had a positive effect on depression.

The size of the mediation effect of involuntary subordination was − 0.7071. The direct effect from self-esteem to depression was − 0.4394. The indirect effect/direct effect was 1.6092.

Model 4 showed that self-esteem (β = − 0.9041, p < 0.001) and social support (β = − 0.1871, p < 0.001) were significant predictors of depression. Model 4 showed no moderation of the overall effect (β = 0.0122, p > 0.05).

As the overall effect of self-esteem on depression was not moderated by social support, Model 6 did not include the moderation effects of the interaction between self-esteem and social support. The analysis indicated a significant effect of self-esteem (β = − 0.3870, p < 0.01), social support (β = − 0.1221, p < 0.01), and involuntary subordination (β = 0.2406, p < 0.001) on depression. The effect of involuntary subordination on depression (the latter path of the indirect effect) was moderated by social support (β = − 0.0041, p < 0.05).

Overall, involuntary subordination mediated the effect of self-esteem on depression and social support moderated the effect of involuntary subordination on depression. The R square of the moderated mediation model was 0.3426. The final pattern is shown in Fig. 2 and its test results are summarized in Table 4.

Following determination of a significant interaction, the estimated conditional effects of involuntary subordination on depression at different levels of social support were examined. The interaction effect was significantly moderated throughout the observed range of social support (12–84). As social support scores increased, the conditional effect of involuntary subordination on depression decreased (range: 0.1594–0.4504, effect of involuntary subordination on depression at mean level of social support: 0.2406). Figure 3 shows the conditional effect at different social support scores.

To further validate the moderated mediation relationship, we examined conditional indirect effects. As shown in Table 5, the conditional indirect effect of self-esteem on depression through involuntary subordination was weaker among participants who scored higher on social support (social support = 51.5217, indirect effect = − 0.5815; social support = 63.9000, indirect effect = − 0.5159; social support = 76.2783, indirect effect = − 0.4359).

Our findings confirmed the vulnerability of young Chinese migrants to depression. Their Center for Epidemiologic Studies Depression Scale score (19.18 [10.06]) was higher than the norm for the general Chinese population (13.24 [10.33]) and the norm of a relatively young Chinese population aged 18–30 years (12.32 [10.22]) (mean[SD]) [32]. These findings are consistent with the conclusion of a meta-analysis that the depression status of migrants is more serious than that of the general population in China [6].

As expected, the findings showed that involuntary subordination mediates the relationship between self-esteem and depression. Involuntary subordination was negatively associated with self-esteem, a finding in accord with several previous studies [17, 41]. Research indicates that self-appraisal, which includes beliefs about self-worth, may have a direct impact upon situational appraisal [42]. For young migrants, stressful events during migration are strongly associated with the development of negative appraisals [43]. Migrants with high self-esteem are likely to form positive self-appraisals, which reduces the likelihood that stressful events will be negatively appraised (situational appraisal). Conversely, when self-appraisals are negative, stressful events may be negatively appraised and cause feelings of defeat and entrapment [42]. Social rank theory suggests that a sense of being of a lower rank than others is likely to cause depression [44]. The present findings confirm this assumption and demonstrate a positive association between involuntary subordination and depression. Most young migrants are unable to alter unfavorable situations in big cities and have to face challenges such as unsuccessful acculturation, discrimination, and low socioeconomic status. Therefore, many are likely to experience involuntary subordination regarding their current situation [45]. Previous research indicates that this type of submissiveness is moderately associated with depression in normal populations and strongly associated with depression in depressed populations [46]. Depression may represent the activation of an internal inhibitory system under conditions of perceived involuntary, subordinate status and self-evaluation [47, 48]. The present findings may help to highlight the long-neglected role of involuntary subordination in the relationship between self-esteem and depression.

In the mediation model, both self-esteem and involuntary subordination emerged as significant predictors of depressive symptoms. Early studies reported that involuntary subordination is a full mediator [17]. However, this study indicated that involuntary subordination partially mediates the relationship between self-esteem and depression, indicating an independent role of self-esteem. The current findings support the notion that self-esteem is negatively associated with depression. Price has explained this relationship by suggesting that low self-esteem may both trigger an episode of depression and increase vulnerability to depression [49]. Price has argued that mood changes can be regarded as a self-esteem management mechanism in which self-esteem is regulated to fit with current social circumstances [50]. The present findings suggest that the mediation model can explain the effect of self-esteem on depression, and that more focus is needed on the role of both self-esteem and involuntary subordination.

Social support significantly moderated the effect of involuntary subordination on depression (the latter path) throughout the observed range of social support. As social support increased, the effect of involuntary subordination on depression monotonically decreased. According to the buffering model, the stress response can be modulated by the social environment. Adequate rescue resources can largely mitigate negative social rank and stress response and alleviate feelings of entrapment [44]. Perceived adequate support from family, friends, and significant others provides young migrants with emotional comfort, helps them to cope with stress more easily, and facilitates more successful adjustment to the new environment [51, 52], which in turn mitigates the effect of involuntary subordination on depression. Contrary to expectations, social support did not moderate the effect of self-esteem on involuntary subordination (the former path). This may be explained by the nature of the dependent variable, involuntary subordination, in young migrants. Migrants are often considered to be of a lower rank than local citizens. This is an aspect of the societal context that may not be easily affected by an interpersonal factor like social support. Overall, the present results complement previous findings that social support moderates the association between traumatic life events and depression [28] and demonstrate that social support can further moderate the relationship between psychological problems caused by traumatic life events and depression. Although it is not easy for young migrants to withstand the harsh conditions of migration, sufficient social support can reduce the adverse psychological consequences of involuntary subordination, such as depression [53].

The present study also showed that social support affects depression directly. Young migrants with greater social support were less likely to experience depression. This is fairly consistent with previous findings that social support, rather than merely protecting an individual against the negative impact of stress, may itself be important in ameliorating depressive symptoms [53]. More attention needs to be paid to this added benefit of improving social support, as this strategy may simultaneously prevent low self-esteem in young migrants who experience involuntary subordination and subsequent depression, and also directly protect against depression.

Consistent with previous research findings [17], the present results revealed that the indirect effect of social support was larger than the direct effect, indicating a dominant role for the indirect effect. Variables that affected the strength of the indirect effect were more important: the impact of self-esteem on depression depended upon the level of social support, indicating that low social support amplified the indirect effect. According to the moderated mediation model, young migrants with lower perceived social support were more vulnerable to the negative effect of low self-esteem on depression through involuntary subordination. This finding provides a theoretical basis for developing interventions to improve social support to prevent depression among young migrants. Social support can be regarded as a defensive mechanism in the indirect effect of self-esteem on depression, and is particularly important, as involuntary subordination is difficult to prevent and to change.

The model tested here combined mediation and moderation to provide a new understanding of factors that correlate with depression. Previous researches indicated that low self-esteem played a critical causal role in the etiology of depression. On this basis, we introduced involuntary subordination, a long-neglected factor, to the association between self-esteem and depression. This is the first study to confirm its mediating role in the relationship between self-esteem and depression. We also considered the moderating role of social support on depression and determined how this factor influenced the indirect effect of self-esteem on depression. These findings may inform strategies for depression prevention both theoretically (in understanding how these variables affect depression) and practically (the development of interventions such as social support).

Although the R square of the moderated mediation model was not high, we believe it was sufficient for a model with only three main variables. R square has a close relation to the number of independent variables (the more independent variables that the model contains, the higher R square will be). We focused on the role of involuntary subordination and social support in the association between self-esteem and depression. Therefore, we did not include other factors, like stress, in our model. Furthermore, there were no differences in depression status between gender or income groups. As a result, sociodemographic characteristics were not included in the model as controls. Despite the limited number of independent variables in the moderated mediation model, the proportion of variance explained by the model was sufficient.

The dataset used for analysis excluded cases with missing data (N = 572). We tested the moderated mediation model again using imputation for missing data (N = 616). The results showed similar findings (involuntary subordination mediated the association between self-esteem and depression; social support moderated the latter path of the indirect effect), which indicated the robustness of model (see Additional file 1).

Considering the proportion of variance explained by the model and the robustness of the non-imputation and imputation datasets, we suggest that this moderated mediation model will be useful for future studies of depression.

The study strengths include the use of a sample of young migrants, the focus on the long-neglected role of involuntary subordination, further exploration of the role of social support, and consideration of the combined effects of mediation and moderation.

Our findings should also be interpreted with caution owing to methodological limitations of the study. The cross-sectional design makes it difficult to establish cause–effect relationships. The initial hypothesized cause–effect relationships were based on previous research findings. Additionally, the sample was drawn from two factories and so may not be representative of all young migrants in the region. It was difficult to obtain a random sample as migrants are generally mobile. Therefore, we used convenience sampling, which limited the representativeness of the sample. Another limitation is that self-report data are subject to possible bias. This is particularly the case for measures of psychological variables, which need to be culturally sensitive owing to differences in how psychological well-being is constructed and reported. We minimized this problem by using scales with good validity and reliability for the Chinese population, using anonymous questionnaires, and providing private rooms for the survey. Finally, there was no control group of non-migrants. Therefore, there is a possibility of bias in this study.