PDF(Pubmed)
Abstract：
Modest correlations between teacher-assigned grades and external assessments of academic achievement (r = .40-.60) have led many educational stakeholders to deem grades subjective and unreliable. However, theoretical and methodological challenges, such as construct misalignment, data unavailability and sample unrepresentativeness, limit the generalisability of previous findings. We overcome these challenges by exploiting rich, population-wide data from the National Registries in Norway (n = 511,858), where state regulations require close construct alignment between grades and external exams. Correlations between lower-secondary education final grades and external exam results (r = .64-.86) suggest that grades are better measures of academic achievement than previously acknowledged. Dominance analyses and multivariate regression analyses indicate that external exam results are the best predictor of grades in the same subject. However, our results also indicate that state regulations and quality assurance systems cannot completely eradicate potential sources of discrepancy.
摘要：
教师分配的成绩与学业成绩的外部评估之间的适度相关性（r=.40-.60）导致许多教育利益相关者认为成绩主观且不可靠。然而,理论和方法上的挑战，例如构造错位，数据不可用和样本不代表性，限制了以前发现的普遍性。我们通过剥削富人来克服这些挑战，来自挪威国家登记处的全人口数据(n=51858)，州法规要求在等级和外部考试之间建立紧密的一致性。初中教育最终成绩与外部考试成绩之间的相关性（r=.64-.86）表明，成绩是比以前公认的更好的学业成绩衡量标准。优势分析和多元回归分析表明，外部考试结果是同一科目成绩的最佳预测指标。然而,我们的结果还表明，国家法规和质量保证体系不能完全消除潜在的差异来源。
