States as STEM Learning Environments: Building an Indicator System to Guide Instructional Improvement at Scale

调查人员正在设计一个全州范围的、基于经验的指标系统，该系统与田纳西州（TN）的教学改进愿景相一致，该系统是通过对四年级到八年级数学课堂的教学和学习样本进行深入研究而收集的数据建立起来的。研究人员从一组已知的结构和测量开始，这些结构和测量可以预测学生在高水平数学评估中的学习（高认知要求教学），并且与教师制定雄心勃勃的教学（教师社会资本）的能力有关。我们的目标是将这些结构的测量提炼成一组核心的调查指标，这些指标可以预测学生的成绩，可以在规模上有效地管理，并对国家决策者的决策产生重要影响。采用三层方法——从密集的教师样本中深入收集数据（n=50），从中间样本中收集不太全面的数据（n=100），从规模样本中收集基于调查的数据，规模样本的数量每年都会增加，在第三年结束时，有1000名教师或大约10%的TN 4-8年级的数学教师-分析利用深入的数学教学和教师社会资本的定性检查，与基于调查的方法对话，构建一个有效性论点（Kane, 2006）， 1)教师在中心结构的代理措施上有所不同，2)这些代理措施与其他理论相关措施相关。3)在量表样本中，这些代理分别对学生学习和高质量教学具有预测作用。以往关于认知需求的研究为教学与学生学习之间的联系提供了理论和实证依据。本项目通过建立高认知要求教学的代理，并观察它们与数百个教室中学生学习的关系，扩展了这项工作。以往关于教师社会资本的研究主要包括理论构建和定性研究，这些研究显示了教师社会资本的具体特征与教学水平之间的关系。本研究在大量的课堂中测试了这种关系，并将其扩展到学生的学习中。此外，研究人员还建立了各州如何支持大规模教学改进的实践理论。虽然这方面的大多数理论和研究都集中在各州在为大量教师提供专业发展方面面临的挑战上，但很少有人关注监测这些努力的有效性和影响的挑战。这个项目是基于这样一种观点：仅仅做得更多——没有系统地从自己正在做的事情中学习——是一个失败的主张。在当前改革的时代，越来越多的州将投入大量资源来改善全州的教学。通过将整个状态概念化为一个学习环境，这项工作呼吁人们关注从该活动中学习的系统和有组织的方法，从而更聪明地找到支持大规模数学改进的有效方法。可以在整个州范围内部署的指标的发展，本身就是一个广泛影响的标志。此外，该指标体系是可移植的，并且由于它与大多数州采用的大学和职业准备标准相一致，因此可以被其他州采用，特别是那些承诺进行共同评估的州。此外，如果泰米尔纳德邦和其他州的领导人使用该指标系统来确定哪些地方的教学不足，哪些地方需要教师的支持，他们就能更好地决定在哪里以及如何帮助全州的教育工作者。反过来，这将有助于实现改善STEM教育的社会目标，从而培养出更多具有全球竞争力的高中毕业生。最后，如果这项研究是成功的，我们将建立一个比以前更长的证据链。我们将展示(a)教师获得社会资源之间的联系；(b)教师的教学实践；(c)学生表现。

Investigators are designing a state-wide, empirically based indicator system that aligns with Tennessee's (TN) vision of instructional improvement and is built from data gathered through in-depth study of teaching and learning in a sample of 4th through 8th grade mathematics classrooms. Investigators begin with a set of constructs and measures that are known to predict student learning on high-level mathematics assessments (high-cognitive-demand instruction) and that are associated with teachers' capacity to enact ambitious instruction (teacher social capital). The objective is to distill the measurement of these constructs to a core set of survey indicators that are predictive of student outcomes, can be administered efficiently at scale, and are consequential for state policy makers' decision making. Using a 3-tiered approach - in-depth data collection from an intensive sample of teachers (n=50), less comprehensive data from an intermediate sample (n=100), and survey-based data from a scale sample, a group that will increase in number each year, ending in year 3 with 1000 teachers or approximately 10% of TN's grades 4-8 mathematics teachers - the analyses leverage in-depth qualitative examination of mathematics instruction and teacher social capital in dialogue with survey-based approaches to construct a validity argument (Kane, 2006) that 1) teachers vary on proxy measures of central constructs, 2) measures of these proxies are correlated with other theoretically relevant measures, and 3) these proxies are predictive of student learning and high-quality teaching, respectively, in the scale sample.Previous research on cognitive demand has provided both theoretical and empirical warrants for the connection between teaching and student learning. This project extends that work by establishing proxies for high-cognitive-demand instruction and observing their relationship to student learning across hundreds of classrooms. Previous work on teacher social capital has consisted primarily of theory-building, qualitative studies that have shown relationships between specific features of teacher social capital and level of instruction. This study tests that relationship in larger numbers of classroom and extends it to student learning. Additionally, investigators build a practical theory of how states can support large-scale instructional improvement. While most theorizing and research in this regard has focused on the challenges that states confront in providing professional development for large numbers of teachers, less attention has been paid to the challenge of monitoring the effectiveness and impact of those efforts. This project is based on the idea that just doing more - without systematically learning from what one is doing - is a failing proposition. In the era of current reforms, increasing numbers of states will be devoting substantial resources to improving instruction across their states. By conceptualizing an entire state as a learning environment, this work calls attention to systematic and organized ways to learn from that activity, thereby getting smarter about efficient ways to support large-scale improvement in mathematics.The development of indicators that can be deployed across an entire state is, in and of itself, a sign of broad impact. In addition, the indicator system is portable and, because it is aligned with college and career ready standards adopted by the majority of states, could be taken up by other states, especially those who have committed to common assessments. Furthermore, if leaders in TN and other states use the indicator system to identify where instruction is flagging and where teacher supports are needed, they will be positioned to make better decisions regarding where and how to assist educators across their states. This, in turn, will support the achievement of the societal goal of improved STEM education that will produce more globally competitive high school graduates. Finally, if this study is successful we will have established a longer chain of evidence than heretofore has been assembled. We will show linkages among (a) teachers' access to social resources; (b) teachers' instructional practice; and (c) student performance.