Family migration and well-being of Chinese migrant workers’ children

This article aims to explore the effects of parental migration on the well-being of children and how to adjust social cognitive well-being through the interrelations among family relations and social cognitive well-being indicators using structural equation modelling. Two modified social cognitive well-being models were tested in 1682 Chinese migrant workers' children to examine the pathways among social cognitive well-being and family relation characteristics. The modified models are based on the social cognitive well-being model and the characteristics of Chinese migrant workers' children. The results show that caregiver-child communication frequency, caregiver-child regulation, caregiver-child conflicts, caregiver-child trust and communication, and coactivity positively impact children’s social cognitive well-being. In contrast, caregiver-child alienation negatively influences children’s social cognitive factors through caregiver-child trust and communication. Additionally, this research revealed that family-related characteristics (caregiver-child regulation, caregiver-child coactivities, caregiver-child communication frequency, caregiver-child alienation, caregiver-child conflicts, and caregiver-child trust and communication) are interconnected with social cognitive well-being indicators (academic satisfaction, outcome expectations, goal progress, lifelong satisfaction, environmental support, positive affect, negative affect, and self-efficacy). This suggests that family migration and relationships with caregiver(s) can significantly affect the well-being of migrant workers' children.


SEM in our study
Structural equa+on modeling in AMOS 22 was used to test the structural validity of these models within the target popula+on.The bootstrap method has been used to test indirect effects related to environmental support, self-efficacy, and caregiver-child coac+vity; 2,000 bootstrap samples were generated to es+mate the indirect effects and bias-corrected 90% confiden+al intervals.Here, three fit indices were used to assess the data fit: CFI (the compara+ve-fit-index), RMSEA (the root mean squared error of approxima+on), and root mean squared residual (SRMR).Values of RMSEA ≤ 0.08, CFI ≥ 0.90, and SRMR ≤ 0.08 are considered an adequate fit, and RMSEA ≤ 0.06, CFI ≥ 0.95, and SRMR ≤ 0.05 are considered a good fit (Hu & Bentler, 1999).Addi+onally, researchers have determined that models mee+ng only one of the three values (CFI, RMSEA, and SRMR) can be considered to have acceptable fit [38].The results of confirmatory factor analysis (CFA) of the above model with the current data set show an acceptable fit, with  != 8829, df =1621, p < 0.0001, CFI = 0.846, RMSEA = 0.051 and SRMR = 0.076.
Posi+ve affect and caregiver-child coac+vity have indirect effects on lifelong sa+sfac+on, the indirect (mediated) effect of caregiver-child coac+vity on the posi+ve affect is 0.114 and with sta+s+cally significant (P < 0.001).Moreover, environmental support and self-efficacy have significant indirect effects on outcome expecta+on; self-efficacy and outcome expecta+on have significant indirect effects on academic sa+sfac+on.The significant standard indirect effects for lifelong sa+sfac+on ranged from -0.287 to 0.283, whereas those for academic sa+sfac+on ranged from 0.100 to 0.310.
Figure 1 shows the research hypothesis that most paths are with sta+s+cal significance.Moreover, most of the paths between SCWB variables are strongly correlated with each other, for example, the paths' coefficient from environmental support to self-efficacy, self-efficacy to goal progress, goal progress to academic sa+sfac+on and academic sa+sfac+on to lifelong sa+sfac+on are both above 0.4.Furthermore, the family rela+onship variables are closely correlated with each other; the paths' coefficient from caregiver-child coac+vity to communica+on frequency is above 0.4, communica+on frequency to regula+on is above 0.2 and caregiver-child coac+vity to regula+on/house rules is above 0.1.Most paths are sta+s+cally significant as it was predicted.This finding is consistent with the results of other studies [2,38].Moreover, the path from communica+on frequency to environmental support is bigger than 0.2, which is easy to understand as students with beher communica+on with caregiver(s) or parents receive more support from their caregiver(s)/parents.What is unexpected is that the path from goal progress to academic sa+sfac+on is not significant.The results in Figure 1 show the coefficients, covariance, and p-value of all paths in the research hypothesis, which are with sta+s+cal significance, except for nine paths (  > 0.05 ).The path coefficients for many social cogni+ve variables (from environmental support to self-efficacy, from selfefficacy to goal progress, from environmental support to outcome expecta+on, from goal progress to academic sa+sfac+on, and from academic sa+sfac+on to lifelong sa+sfac+on) showed similar results with previous studies [38] related to the SCWB model.Most of the path coefficients with sta+s+cal significance were above zero, except the path coefficient of the path from outcome expecta+on to nega+ve affect (ℎ  = −0.19, < 0.001).The path coefficient of the path from selfefficacy to goal progress was the largest (ℎ  = 0.76,  < 0.001), which means that when the self-efficacy improves 1 unit, the goal progress increases 0.76 units and ceteris paribus.This is followed by the path from caregiver-child coac+vity to caregiver-child communica+on frequency ( ℎ  = 0.60,  < 0.001 ), the path from environmental support to self-efficacy ( ℎ  = 0.54,  < 0.001 ), the path from self-efficacy to academic sa+sfac+on (ℎ  = 0.49,  < 0.001), and the path from academic sa+sfac+on to lifelong sa+sfac+on (ℎ  = 0.47,  < 0.001).Meanwhile, it is not significant for the path from caregiverchild coac+vity to caregiver-child regula+on, the path from caregiver-child regula+on to outcome expecta+on, and another nine paths.Furthermore, many paths with sta+s+cal significance are between caregiver-child rela+onship variables and SCWB variables, for example, the path from caregiver-child coac+vity to posi+ve affect (ℎ  = 011,  < 0.001), the path from selfefficacy to caregiver-child regula+on (ℎ  = 0.13,  < 0.001), the path from caregiverchild communica+on frequency to posi+ve affect (ℎ  = 0.19,  < 0.001), and the path from caregiver-child coac+vity to lifelong sa+sfac+on (ℎ  = 0.11,  < 0.001).It implies that more communica+on and co-ac+vi+es between a child and caregiver(s) can improve the quality of the self-efficacy, posi+ve affect, and lifelong sa+sfac+on of Chinese VET school students.Caregiverchild regula+on was nega+vely correlated with academic sa+sfac+on but posi+vely correlated with outcome expecta+on.
Figure 2 shows the research hypothesis of the second family rela+onship SCWB model, which included two of the caregiver-child ahachment variables (caregiver-child communica+on & trust and aliena+on) and caregiver-child conflict.The paths between social cogni+ve well-being variables were based on previous SCWB studies [37,38].The paths between SCWB variables and family rela+onship variables were based on the characteris+cs of Chinese VET school students from migrant families.In the hypothesis model, self-efficacy, outcome expecta+on, goal progress, and academic sa+sfac+on bridged the correla+on paths between environmental support and lifelong sa+sfac+on.
The structural equa+on modeling analysis results for the child-caregiver ahachment and conflict social cogni+ve model with the current data shows an acceptable fit also, with  != 7576.2,df = 1740, p < 0.0001, CFI = 0.861, RMR = 0.068 and RMSEA = 0.045.The coefficient of four significant paths is above 0.5 (shown in figure 6), which are the path from environmental support to caregiver-child trust & communica+on, the path from environmental support to self-efficacy, the path from self-efficacy to goal progress, and the path from self-efficacy to academic sa+sfac+on.Moreover, coefficients of the other four paths are above 0.3, which includes the path from environmental support to academic sa+sfac+on, the path from self-efficacy to outcome expecta+on, the path from academic sa+sfac+on to lifelong sa+sfac+on, and the path from aliena+on to conflicts.Moreover, four of the significant paths are below 0; that means the correla+on between these two variables is nega+ve, for example, the path from environmental support to aliena+on, the path from outcome expecta+on to nega+ve affect, the path from trust & communica+on to aliena+on and the path from aliena+on to lifelong sa+sfac+on.Furthermore, the coefficient of other paths in this model is above 0.

Figure 1 :
Figure 1: Family Relationship and Social Cognitive Model I -Path Coefficients

Figure 2 :
Figure 2 : Family Relationship and Social Cognitive Model II -Path Coefficients