论文一：人际关系在STEM路径中的重要性：机器学习与统计分析的联合证据

摘要 (中文):

背景：

STEM领域的参与度达到了历史新高，然而科学知识往往未能有效转化为应对紧迫社会与环境挑战的行动，导致“知识—行动”差距日益扩大。本研究旨在探究大学新生选择STEM职业的动机，是可能加剧还是弥合这一差距，并重点关注他们过往的STEM经历如何影响其未来“实践”STEM的方式。我们利用关于学生在学前至高中（PK–12）阶段接触STEM的丰富调查数据，探究在不同的大学前经历组合下，何种STEM未来发展的可能性会更高。

结果：

我们分析了来自美国119所院校15725名大一新生的调查数据，将STEM乐趣以及四种职业动机（帮助他人、赚钱、工作稳定、机会众多）作为预测STEM职业意向的指标。研究发现，对STEM的享受是强烈的正向预测因子，而帮助他人的愿望则与较低的STEM职业意向相关。在3729名有STEM职业意向的学生中，聚类分析识别出三种画像：既享受STEM又高度重视帮助他人的学生；尽管对STEM乐趣不大但仍为帮助他人而追求STEM的学生；以及享受STEM但不太重视帮助他人的学生。尽管总体上“帮助他人”与STEM职业意向呈负相关，但约有三分之二的STEM意向学生属于将“帮助他人”视为核心动机的画像。我们训练了一个机器学习模型，利用389个关于过往STEM经历的变量来预测学生的动机画像。结果显示，关系型STEM参与——即与朋友和家人讨论和参与STEM活动——成为最强的区分因素：既享受STEM又希望帮助他人的学生报告了最高的关系型参与度和鼓励度，而那些具有社群动机但STEM乐趣较低的学生则报告了持续较低的支持。

结论：

研究结果表明，许多有志于STEM的学生已经开始以“帮助他人”为中心来构想自己的未来，但支持这种社群导向的关系条件分布不均。大学前的经历模式指出，与朋友、家人及其他成年人共享STEM话题和活动，是教育者和机构可以采取行动，在STEM发展路径中维持和放大社群动机的有效切入点。

Abstract (原文):

Background

STEM participation is at record highs, yet scientific knowledge often fails to translate into effective responses to urgent social and environmental challenges, resulting in a growing knowledge–action gap. This study examines how early college students’ STEM career motivations are likely to either reinforce or disrupt this gap, focusing on the role that prior STEM experiences play in how they intend to “do” STEM. Drawing on rich survey data about students’ PK–12 exposure to STEM, we ask what kinds of STEM futures become more likely under different constellations of pre-college experiences.

Results

We analyzed survey data from 15,725 first-year students at 119 U.S. institutions, modeling STEM enjoyment and four career motivations (helping others, making money, job security, abundant opportunities) as predictors of STEM career intentions. STEM enjoyment was a strong positive predictor, while a desire to help others was associated with lower likelihood of intending a STEM career. Among the 3729 students who did intend STEM careers, cluster analysis identified three profiles: students who both enjoy STEM and strongly value helping others, students who pursue STEM to help others despite low STEM enjoyment, and students who enjoy STEM but place little emphasis on helping others. Despite the overall negative association, roughly two-thirds of STEM-intending students fell into profiles where helping others was central. We trained a machine learning model to use 389 variables on prior STEM experiences to predict their motivational profile. Relational STEM engagement—talking and participating in STEM with friends and family—emerged as the strongest differentiator: students who both enjoyed STEM and wanted to help others reported highest relational engagement and encouragement, whereas communally motivated students with low STEM enjoyment reported consistently lower support.

Conclusions

The findings suggest that many STEM-aspiring students already imagine their futures in ways that center helping others, but that the relational conditions that support these communal orientations are unevenly distributed. Patterns in pre-college experiences point to shared STEM talk and activity with friends, family members, and other adults as promising sites where educators and institutions can act to sustain and amplify communal motivations within STEM pathways.

📖原文信息：

标题：The importance of relational pathways into STEM: evidence from a combined machine learning and statistical analysis

作者：Christopher Irwin; Zahra Hazari; Remy Dou; Philip Sadler; Gerhard Sonnert

来源：International Journal of STEM Education