Parental emotional support, emotional intelligence, and subjective well-being are critical factors that affect the physical and mental development of high school students. Parenting style can determine adaptability, and affectionate parenting is conducive to emotional development. Children are more likely to develop high levels of emotional intelligence if their family members can satisfy their psychological needs and provide sufficient emotional support. Emotional intelligence is not innate; rather, it is an intrapersonal skill that can be cultivated. Furthermore, the literature has revealed that parental emotional support influences the emotional intelligence of children, which affects their subjective well-being. Thus, emotional intelligence can mediate the relationship between parental emotional support and subjective well-being. The researchers searched for the keywords “parental emotional support,” “emotional intelligence,” and “subjective well-being” in Chinese-language literature databases, such as Airiti Library and Chinese language version of Google Scholar. The results indicated that no studies had analyzed the aforementioned mediating effect of emotional intelligence. The following hypothesis was proposed: The emotional intelligence of high school students would mediate the relationship between parental emotional support and subjective well-being. The purpose of this study was to verify the mediating effect of emotional intelligence and to address research gaps. 

To verify the aforementioned hypothesis, we developed three scales to conduct a questionnaire survey. We revised the Emotional Support Subscale (seven items) from the short-form Egna Minnen av Barndoms Uppfostran Scale developed by Arrindell et al. (1999), and the revised form was used to measure parental emotional support. The original scale was proposed in 1999, and its participants were secondary school students and adults in Sweden; this relatively old scale targeting individuals from a different culture may be unsuitable for Taiwanese students. Therefore, a pretest was performed. An Emotional Intelligence Inventory was designed according to the Emotional Intelligence Scale proposed by Chiang (2001). The original scale was proposed over a decade ago and consists of 60 items; thus, we developed a condensed version to prevent participants from becoming bored and tired, which could have affected the accuracy of survey results. The shortened version was subjected to a pretest. A Subjective Well-being Inventory was developed based on two scales. The first was the Emotional Well-beings” Subscale of the Teacher’s Subjective Well-being Scale proposed by Yu et al. (2011). This scale has two dimensions: the commonly accepted notion of happiness and perceived life satisfaction. The other scale used to develop the Subjective Wellbeing Inventory was the Chinese Happiness Inventory developed by Lu (1998). This scale targeted adults aged 18–65 years in Kaohsiung City, Taiwan. The scale of Yu et al. was designed for teachers, whereas that of Lu was designed for the general public. Therefore, we selected items with high universal representativeness and that were suitable for high school students. The three inventories were rated on a 5-point Likert-type scale with scores of 1 (strongly disagree), 2 (partly agree), 3 (agree), 4 (mostly agree), and 5 (strongly agree). The confirmatory factor analysis (CFA) results indicated that the Inventories of Parental Emotional Support, Emotional Intelligence, and Subjective Well-being exhibited desirable composite reliability, average variance extracted, and discriminant validity. These inventories were employed to conduct a survey involving 918 Taiwanese vocational high school and senior high school students. The participants were mainly female (52.50%; 482 individuals), and first-, second-, and third-grade students accounted for 41.29% (397), 29.41% (270), and 27.34% (251) of the total sample, respectively. Survey responses were collected and analyzed using structural equation modeling (SEM), and a model was constructed in which emotional intelligence mediates the relationship between a father and mother’s emotional support and a student’s subjective well-being. At first, Amos (covariance-based SEM software) , manufactured by International Business Machines Corporation, New York, USA, was adopted for SEM, but the model fit indices indicated a poor fit. Moreover, the multivariate normality test result was significant, which suggested that the collected data had a non-normal distribution. Therefore, the researchers used Smart PLS 3.0, a variance-based SEM software program that is not limited by normal distribution. In addition, common method variance (CMV), a common problem in questionnaire surveys, had to be controlled. CMV is the bias caused by a research method in which the correlation between two variables is inflated or deflated. To avoid CMV, we designed a marker variable as the control variable. Theoretically, a marker variable should not be uncorrelated with the variables of interest; The latent variable “interpersonal understanding” was designated as the CMV variable.The mediating effect process was analyzed with reference to the method of Hair Jr et al. (2016) using three steps. Step 1: The researchers verified whether the path coefficient of the independent variable that predicts the dependent variable was significant. Step 2: The researchers confirmed that the path coefficients of the independent variable that predicted the mediator and those of the mediator that predicted the dependent variable were significant. Step 3: The researchers analyzed the mediation model only when the variables in Steps 1 and 2 were confirmed to be significantly correlated. In the present study, the variables in Steps 1 and 2 were significantly correlated; thus, the mediation model could be analyzed.

Before a mediation model analysis was conducted, the model fit was verified. The results indicated that the factor loading, composited reliability, average variance extracted, and discriminant validity all satisfied relevant criteria and showed that the measurement model (called “outer model” in PLS-SEM) was appropriate. Second, values for collinearity, R2, f 2, and Q2 all conformed to the standard structural model (called “inner model ” in PLS-SEM); that is, the structural pattern was appropriate.

All the paths reached significance, and the mediating effect was then measured using the variance accounted for (VAF), which was calculated by dividing the indirect effect by the total effect. According to Hair Jr et al. (2016), percentage VAF of > 80%, 20%, –80%, and < 20% represent full mediation, partial mediation, and no mediation, respectively. Accordingly, our research results demonstrated that emotional intelligence partially mediated the relationship between parental emotional support (both from a father and a mother) and subjective well-being. The researchers then verified whether this partial mediating effect was significant. The built-in bootstrapping function of Smart PLS was employed for this task, with the number of bootstrap samples set at 5,000. The results indicated that emotional intelligence significantly mediated the relationship between a father and mother’s emotional support and a student’s subjective well-being. The mediation model analysis revealed that the path coefficients of both a father and mother’s emotional support on the subjective well-being and emotional intelligence of (vocational) high school students indicated significant and positive correlations. Furthermore, the path coefficient between emotional intelligence and subjective well-being indicated a significant and positive correlation between the two. Therefore, the proposed hypothesis (i.e., the emotional intelligence of high school students would mediate the relationship between parental emotional support and subjective well-being) was supported.

According to these findings, we propose that parents should provide emotional support to their children more often, which would help children develop emotional intelligence and improve their subjective well-being. High school students should actively attempt to develop their emotional intelligence. Suggestions for future research and research limitations are discussed herein.