The Effects of Parental Support on Science Achievements in G7 and East-Asian Countries

• In Event: Chinese roundtable I (Sponsored by East Asia SIG)
  In Refereed Round-Table Session: Educational inequality

Proposal

Background: Programme for International Student Assessment (PISA) examines various aspects of 15-year-olds’ education internationally focusing on mathematics, science and reading rotationally. An increasing number of researchers have worked on PISA research topics (Hopfenbeck et al., 2018). Parental education and parental involvement in school activities play an important role in the inequality, academic achievements, self-efficacy, engagement and motivation (Gamboa & Waltenberg 2012; Sebastian et. al, 2017; Fan & Williams, 2010). However, there are few studies covering the influence of parental support on academic achievements.
Research Question: This research is interested in assessing the effectiveness of parental support on students’ science performance using a propensity score matching approach.

Data: 12 out of 72 countries and economies were sampled from the 2015 PISA. 1. These chosen 12 economies are among Group of 7 (G7) countries (Canada, Germany, Italy, Japan, the United Kingdom and the United States) and the East Asia countries (China, Hong Kong, Macau, Taiwan, Korea and Japan), with Japan as both a G7 and East Asian country. France as one of the G7 countries is not included because some of the important variables are missing in the data.
The outcome variable is average science score. The independent variable is the parental support index. The independent variable is a composite indicator of parental support. These variables and covariates are shown in Table 1 in the Appendix.

Data Analysis: Table 2 describes the sample size and outcome per country. Correlations among the variables of interest are shown in Figure 1 and Table 3 & 4. Figure 2 and Figure 3 display the distributions of the categorical and continuous variables in this data set, suggesting the categorical variables do not have attributes of normality or problem of multicollinearity.
Employing covariates, propensity scores were computed via logistic regression. After propensity score estimations, propensity score was employed to match the treatment and control groups.

Results: Before and after propensity score matching, the parental support effects on science performance were estimated. As listed in Table 5, propensity score matching did make a difference on the effect. The effect of parental support was not significant for Germany before propensity score matching (PSM), but it has a significantly negative effect after PSM. Though having positive effects, the effects weakened after PSM for UK, Japan and Singapore. For all the other economies, except Hong Kong, whose effect is not significant before and after PSM, these 7 countries and economies do not demonstrate any significant effect of parental support on science achievements after PSM, although their effects are significant at various level before PSM.

Conclusions: Results indicate that propensity score matching is useful to reduce bias and that these 12 economies have different effects of parental support. Germany has a negative effect of parental support after PSM, while United Kingdom, Japan and Singapore demonstrate positive effects before and after PSM, indicating the effect of parental support is not universal and it changes according to different economic, political and cultural backgrounds.
However, to build causal relationships, randomized control experiments are desired.

Author

• Wan Ren, University of Pennsylvania