影響兒童社會能力因素之解釋力比較：SOCIAL模型取向

Comparison of Explanatory Powers among Factors Affecting Social Competence in Young Children: A Sociocognitive Integration of Abilities Model Approach

本研究以社會—認知整合能力模型（SOCIAL）為架構，以「臺灣幼兒發展調查資料庫（Kids in Taiwan: National Longitudinal Study of Child Development & Care, KIT）」中48月齡幼兒為研究對象，整體性的比較衆多因素對社會能力解釋力如何。KIT中48月齡幼兒初始樣本有2,031人，隨機擷取100人進行驗證性因素分析，即重新預試KIT中的問卷，結果表明，所有量表均具有良好的信、效度。在剩餘1,931份樣本中，採用整列刪除法，刪除有遺漏值樣本，再刪除發展異常與父母親之一是外籍配偶之樣本，共刪除86人，最後，有效樣本剩1,845人，其中男童929人（50.4%），女童916人（49.6%）。依據文獻分析結果提出本研究結構方程模型，該模型描述了不同變量對社會能力之解釋力，由於模型複雜，採用適合模型複雜的偏最小平方法結構方程模型（PLS-SEM）來分析。結果發現：情緒能力的解釋力最大，其他依序為：氣質、執行功能、退縮為、細動作、家庭環境與攻擊為等變量。就此等發現，針對家長與幼兒園提出以下之建議：首先，增強兒童社會能力的關鍵是優先關注和培養其情緒能力；其次，需要強化兒童的奮力控制性和外放性氣質；此外，也應注重培養兒童的執行功能。未來研究可選擇單一領域的因素，就其所含的變量對兒童社會能力影響進行更深入的解釋力之比較。

Social competence—the ability to achieve interpersonal success in various social contexts—is developed during the socialization process in childhood. It influences one’s ability to adapt socially, interpersonal interactions, emotions, personality, and academic performance. Individuals lacking social competence may be aggressive, be destructive, and have difficulty adapting.
Several factors influence the development of social competence during childhood; however, studies on the effects of these factors on social competence have only explored the effects of individual factors or combinations of two or three factors. Furthermore, few studies have focused on comprehensive investigations. The framework for the present study was the sociocognitive integration of abilities (SOCIAL) model, which was proposed by Beauchamp and Anderson. The SOCIAL model combines cognitive functions and social cognitive skills that influence social competence, along with internal and external factors that mediate these skills, within a single model. This framework provides a comprehensive approach to studying social competence in young children. Data were collected from the Kids in Taiwan: National Longitudinal Study of Child Development & Care Database (KIT), which aligns well with the SOCIAL model. We identified factors affecting social competence in Taiwanese young children; these factors were compared in terms of explanatory power. The results of the present study may assist families and educational institutions in enhancing young children’s social competence.
The KIT data set contains data from young children aged 48 months (N = 2,031). After applying listwise deletion, we selected 100 young children for partial least squares (PLS) confirmatory factor analysis. The remaining valid sample comprised 1,845 individuals (boys, 929 [50.4%]; girls, 916 [49.6%]).
Data were obtained from various domains within the KIT data set: cognitive development, language development, social and emotional development, physical motor development, and family-related domains. The following variables were extracted from these five major domains: memory, executive function, family environment, parental involvement, language development, temperament, secure attachment, emotional competence, withdrawal behavior, aggressive behavior, shame, gross motor skills, fine motor skills, Chinese parenting, and shyness. The reliability and validity of these variables were confirmed through PLS confirmatory factor analysis. The results indicated that factor loadings (λ), average variance extracted, and composite reliability all met the standard cutoff values, indicating good reliability. Discriminant validity was also evaluated; the value indicated good construct validity.
Because the hypothetical model involves 15 variables and is regarded as a complex model, the PLS structural equation modeling (SEM) software was used to test the model.
The first step involved assessing the appropriateness of the structural model in terms of four indicators: collinearity—the variance inflation factor must be≤5.00; coefficient of determination (R2)—thresholds of .25, .50, and .75 were used to categorize explanatory power into low, moderate, and high levels; effect size (f2)—thresholds of .02, .15, and .35 were used, as proposed by Cohen (1992), to categorize effect size into low, medium and high effects, respectively; and predictive relevance (Q2)—a Q2 value of <0 indicates a predictive relevance between the path and the construct.
During the testing phase, the association of each exogenous variable with social competence was analyzed. Upon identifying significant associations, we integrated the relevant variables into a single SEM to compare them in terms of explanatory power. The13 variables (memory, executive function, family environment, parental involvement, language development, temperament, secure attachment, emotional competence, withdrawal behavior, aggressive behavior, shame, gross motor skills, and fine motor skills) were significantly associated with social competence. However, no significant association was noted between Chinese parenting or shyness and social competence.
Subsequently, the 13 variables were incorporated into an initial SEM model to compare explanatory power among these exogenous variables. However, someλvalues in the initial model were too low, which negatively affected the model fit. For instance, the conditional discipline dimension (λ= .19) within the family environment variable did not meet the standard cutoff range of .50 to .95. Moreover, the negative emotionality temperament dimension (λ= .14) within the temperament variable led to an inadequate model fit. Hair et al. reported that factors withλvalues between .40 and .70 can be removed to improve model fit. After the removal of these factors, we tested the modified model again; the results indicated a good model fit. The variance inflation factor, R2, f2, and Q2 all met the standard cutoff values. Therefore, explanatory power was compared among the factors. The size of the explanatory power can be inferred fromβandΔR2 (R2 change); larger values indicated stronger explanatory power. To understand theΔR2 for each exogenous variable, the analysis was repeated after removing that variable from the modified model. Then, the reduction in R2 was subtracted from the original R2 to calculateΔR2. The effect size ofΔR2 was assessed using f2.
Within the modified model, the following exogenous variables contributed to the explanatory power of social competence: emotional competence, temperament, executive function, withdrawal behavior, fine motor skills, family environment, and aggressive behavior. Emotional competence had the highest explanatory power, followed by temperament and executive function. The remaining four variables hadβvalues of < .10, indicating significant but relatively low explanatory power. Secure attachment, gross motor skills, language development, parental involvement, shame, and memory did not reach significance in terms of explanatory power.
Among the aforementioned 13 variables, emotional competence had the greatest explanatory power for social competence; this variable was followed by executive function and temperament.
On the basis of our findings, we propose the following recommendations.
First, we should prioritize and cultivate emotional competence to enhance social competence in young children. In young children, emotional competence exhibits the greatest explanatory power for social competence. Therefore, to enhance social competence in young children, we should cultivate emotional competence in addition to emotional awareness, expression, understanding, and regulation. This should to be supported by an appropriate educational environment and an effective channel for peer communication. Thus, we must create an environment that facilitates emotional learning.
Second, effortful control should be strengthened, and an extroverted temperament should be developed. We found that temperament exhibits secondary explanatory power for social competence. To increase young children’s social competence, we should cultivate temperament, which comprises effortful control and extroversion, as noted in our study.
Third, we should enhance young children’s executive function by reminding them that their behavior may affect others. For example, to slow down and make things better, and to put toys or used items back in their original fixed storage place, or young children are asked to lower their volume while they speaking, they can immediately lower the volume and maintain it for at least a few minutes .
Finally, in the future, the explanatory power of social competence should be compared across different domains.