Columbia University, Department of Medical Humanities & Ethics, New York, NY, United States; The Hastings Center, New York, NY, United States.
New York State Psychiatric Institute, Department of Mental Health and Data Science, New York, NY, United States.
Acta Psychol (Amst). 2024 Aug;248:104403. doi: 10.1016/j.actpsy.2024.104403. Epub 2024 Jul 14.
Although limited in predictive accuracy, polygenic scores (PGS) for educational outcomes are currently available to the public via direct-to-consumer genetic testing companies. Further, there is a growing movement to apply PGS in educational settings via 'precision education.' Prior scholarship highlights the potentially negative impacts of such applications, as disappointing results may give rise a "polygenic Pygmalion effect." In this paper two studies were conducted to identify factors that may mitigate or exacerbate negative impacts of PGS.
Two studies were conducted. In each, 1188 students were randomized to one of four conditions: Low-percentile polygenic score for educational attainment (EA-PGS), Low EA-PGS + Mitigating information, Low EA-PGS + Exacerbating information, or Control. Regression analyses were used to examine differences between conditions.
In Study 1, participants randomized to Control reported significantly higher on the Rosenberg Self-Esteem Scale (RSES), Competence Scale (CS), Academic Efficacy Scale (AES) and Educational Potential Scale (EPS). CS was significantly higher in the Low EA-PGS + Mitigating information condition. CS and AES were significantly lower in the Low EA-PGS + Exacerbating information condition compared to the Low EA-PGS + Mitigating information condition. In Study 2, participants randomized to Control reported significantly higher CS and AES. Pairwise comparisons did not show significant differences in CS and AES. Follow-up pairwise comparisons using Tukey P-value correction did not find significant associations between non-control conditions.
These studies replicated the polygenic Pygmalion effect yet were insufficiently powered to detect significant effects of mitigating contextual information. Regardless of contextual information, disappointing EA-PGS results were significantly associated with lower assessments of self-esteem, competence, academic efficacy, and educational potential.
尽管预测准确性有限，但通过直接面向消费者的基因检测公司，公众现在可以获得用于教育成果的多基因评分（PGS）。此外，通过“精准教育”，在教育环境中应用 PGS 的运动正在兴起。先前的学术研究强调了此类应用的潜在负面影响，因为令人失望的结果可能会引发“多基因皮格马利翁效应”。在本文中，进行了两项研究，以确定可能减轻或加剧 PGS 负面影响的因素。
进行了两项研究。在每项研究中，将 1188 名学生随机分配到以下四种条件之一：教育成就（EA-PGS）低百分位数的多基因评分、低 EA-PGS+减轻信息、低 EA-PGS+加重信息或对照。回归分析用于检查条件之间的差异。
在研究 1 中，随机分配到对照组的参与者在罗森伯格自尊量表（RSES）、能力量表（CS）、学术效能感量表（AES）和教育潜力量表（EPS）上的得分显著更高。在低 EA-PGS+减轻信息条件下，CS 得分显著更高。与低 EA-PGS+减轻信息条件相比，低 EA-PGS+加重信息条件下的 CS 和 AES 得分显著降低。在研究 2 中，随机分配到对照组的参与者报告 CS 和 AES 显著更高。成对比较显示 CS 和 AES 没有显著差异。使用 Tukey P 值校正进行的后续成对比较未发现非对照条件之间存在显著关联。
这些研究复制了多基因皮格马利翁效应，但没有足够的效力来检测减轻背景信息的显著影响。无论背景信息如何，令人失望的 EA-PGS 结果与自尊心、能力、学术效能感和教育潜力的评估较低显著相关。
