摘要:本文运用元分析方法,在广泛收集1990—2014年间国内外对这一问题的大量文献基础上,综合了符合元分析文章纳入标准的25篇文献,研究课外辅导对学生学业成绩的影响。元分析结果表明:课外辅导对学业成绩的效应量(EffectSize)为0.272。删掉一个极端效应值后,效应量仅为0.056。根据Cohen的标准,这一效应量较低。调节效应分析发现:保留异常值时,课外辅导与学业成绩之间的关系受到学业成绩测量方式和年级这两个调节变量的影响。用学科成绩这一指标来反映学业成绩时,课外辅导对学业成绩的效应较大。而用国际测试成绩或国家升学考试成绩作为学业成绩的测量指标时,这一效应量几乎为零。课外辅导对小学生的效应较大,对初中生、高中生的效应接近零。删掉异常值后,虽然效应量在不同群体中的影响趋势是相同的,但只有学业成绩测量方式具有显著的调节作用。
It seems to be intuitive that private tutoring has positive effects on students" academic achievement. However, such an assumption is commonly based upon studies that are qualitative and speculative in nature. On the other hand, among those studies on the effectiveness of private tutoring that are quantitative, the findings are inconclusive, and even contradictory to each other. The current study aimed to address the problem using meta-analysis approach. Results were synthesized across 25 quantitative studies on private tutoring and academic achievement in journal articles, which were published between 1990 and 2014 and met the inclusion criteria. Meta-analysis results showed that the overall effect of private tutoring on academic achievement is 0.272 and decreases to 0.056 after deleting an outlier. These effect sizes are small according to Cohen's standards. Follow-up moderation analysis revealed that different academic achievement measures and grade moderate the effect of private tutoring significantly. Specifically, when teacher-developed test is used as measure of student academic performance, the effect of private tutoring on academic achievement is significant. However, the effect size is almost zero when international test or national entrance examinations are used as measures of academic achievement. The effect of private tutoring on academic achievement is higher on primary students, compared to junior or senior school students. The same pattern of effect size variation across different groups is observed whether retaining the outlier or not. However, after the outlier is deleted, academic achievement measure is the only variable that shows siznificant moderatinz effect.