Piotr Mitros, Anant Agarwal, Vik Paruchuri
Assessment in traditional courses has been limited to either instructor grading, or problems that lend themselves well to relatively simple automation, such as multiple-choice bubble exams. Progress in educational technology, combined with economies of scale, allows us to radically increase both the depth and the accuracy of our measurements of what students learn. Increasingly, we can give rapid, individualized feedback for a wide range of problems, including engineering design problems and free-form text answers, as well as provide rich analytics that can be used to improve both teaching and learning. Data science and integration of data from disparate sources allows for increasingly inexpensive and accurate micro-assessments, such as those of open-ended textual responses, as well as estimation of higher-level skills that lead to long-term student success.