Research to date on the effects of between-school tracking on inequalities in achievement and on performance has been inconclusive. A possible explanation is that different studies used different data, focused on different domains, and employed different measures of inequality. To address this issue, we used all accumulated data collected in the three largest international assessments—PISA (Programme for International Student Assessment), PIRLS (Progress in International Reading Literacy Study), and TIMSS (Trends in International Mathematics and Science Study)—in the past 20 years in 75 countries and regions. Following the seminal paper by Hanushek and Wößmann ( 2006 ), we combined data from a total of 21 cycles of primary and secondary school assessments to estimate difference-in-differences models for different outcome measures. We synthesized the effects using a meta-analytical approach and found strong evidence that tracking increased social achievement gaps, that it had smaller but still significant effects on dispersion inequalities, and that it had rather weak effects on educational inadequacies. In contrast, we did not find evidence that tracking increased performance levels. Besides these substantive findings, our study illustrated that the effect estimates varied considerably across the datasets used because the low number of countries as the units of analysis was a natural limitation. This finding casts doubt on the reproducibility of findings based on single international datasets and suggests that researchers should use different data sources to replicate analyses.

中文翻译：

---

早期追踪和不同类型的成就不平等：来自 20 年大规模评估的差异证据

迄今为止，关于校际追踪对成就和表现不平等的影响的研究尚无定论。一种可能的解释是，不同的研究使用不同的数据，关注不同的领域，并采用不同的不平等措施。为了解决这个问题，我们使用了三个最大的国际评估——PISA（国际学生评估计划）、PIRLS（国际阅读素养研究进展）和 TIMSS（国际数学和科学研究趋势）——中收集的所有累积数据。近20年来在75个国家和地区。继 Hanushek 和 Wößmann（2006 年）发表的开创性论文之后，我们结合了总共 21 个中小学评估周期的数据，以估计不同结果测量的差异模型。我们使用元分析方法综合了这些影响，并发现了强有力的证据表明跟踪社会成就差距的增加，它对离散不平等的影响较小但仍然显着，并且它对教育不足的影响相当微弱。相比之下，我们没有发现跟踪提高绩效水平的证据。除了这些实质性发现之外，我们的研究表明，由于作为分析单位的国家数量少是一个自然限制，因此所使用的数据集的效果估计值差异很大。这一发现对基于单一国际数据集的研究结果的可重复性提出了质疑，并建议研究人员应该使用不同的数据源来重复分析。