SBIR Phase I: Providing Tools for Richer eLearning Assessment

这个小企业创新研究（SBIR）第一阶段项目将研究创建测试构建工具的可行性，使学校教育工作者能够方便地制作和提供各种测试，从非正式的掌握程度评估，可以在飞行中进行，到根据目标对教学进展进行基准测试。关键的创新是(1)为存储的问题定义答案分析的能力，这样测试构造者就可以提前知道测试可以报告什么，考生可能知道什么，不知道什么；(2)以实际经验可以用来改进评估语料库的形式表示问题的能力。目标是创建超越目前广泛接受的完全由多项选择题组成的应用程序的测试，包括各种甚至游戏式的问题类型，包括动机和教学有效的反馈；也就是说，在评估的同时进行教学的问题类型。这些是问题类型，例如拖放、匹配、填空句子和可能有多个正确答案的表格构建，它们没有广泛一致的错误答案干扰，并且通常需要更多的经验来定义错误意味着考生的知识和技能。该项目的目的不是要与高风险的考试竞争，而是要超越目前完全由多项选择题组成的应用程序。这个项目是决定是否将战略重点放在提高低风险上的第一步。评估在商业上和智力上都有价值。长期以来，多项选择和构造答案考试被证明是州和国家高风险考试的高效工具。然而，多项选择题的一个问题是，许多题不评估学生知道什么，而只评估学生证明他们知道什么。在多项选择题测试中，某些类型的学生通常比其他人表现得更好。在问责制加强的时期，设计公平的、经得起法律挑战的测试项目的困难，使得多项选择题在代表大多数学龄人口的州，被一致认为是唯一有效、可靠的高风险评估形式。由于许多学生的学习成绩几乎完全由多项选择题测试来衡量，而这些学生并没有得到很好的服务，因此，通过提高评估的内在趣味性，以及改善学生、教师和管理人员可获得的形成性反馈，可以增加学习者的体验，从而对探索性学习做出重大贡献。

This Small Business Innovation Research (SBIR) Phase I project will study the feasibility of creating test construction tools that allow school educators to conveniently produce and deliver tests ranging from informal assessments of mastery that can be given and taken on the fly, to tests that benchmark progress of instruction against goals. The key innovations are (1) the capability to define answer analyses for stored question items so that the test constructor can know in advance what the test can report about what test-takers likely know and do not know, and (2) the capability to represent question items in a form in which actual experience can be used to improve the assessment corpus. The objective is to create tests that move beyond the current broadly-accepted applications that consist entirely of multiple-choice questions and that include varied and even game-like question types incorporating motivational and pedagogically effective feedback; that is, question types which teach while they assess. These are question types, such as drag-and-drop, matching, fill-in-the-blank sentence, and table builders that may have multiple correct answers, which do not have broadly agreed upon wrong answer distractors, and which typically require more experience to define what errors imply about the test-takers' knowledge and skills. The aim of the project is not to compete with high-stakes tests, but to move beyond current applications that consist entirely of multiple-choice questions. This project is a first step in determining whether the strategy of focusing on improving .low-stakes. assessments has merit commercially as well as intellectually. Multiple-choice and constructed answer exams have long proven highly efficient tools for state and national high-stakes exams. A problem with multiple-choice questions, however, is that many do not assess what students know but only what students demonstrate they know. Certain types of students typically perform better than others on multiple-choice tests. In a period of heightened accountability, the difficulty of designing fair test items that can withstand legal challenge has made multiple-choice, by consensus, the only efficient, reliable form of high-stakes assessment in states representing most of the school-age population. Because there are many students in environments in which their learning is measured almost solely by multiple-choice tests who are not served well, a significant contribution to exploratory learning can be made by increasing what learners can experience by making assessments more intrinsically interesting and also by improving the kinds of formative feedback available to students, teachers, and administrators.