https://www.smh.com.au/business/small-business/bill-gates-backed-edtech-founder-has-a-plan-for-standardised-tests-20190904-p52noj.html
I saw this article about the Khan academy running data analysis on the standardized testing and I wondered what kind of analysis they are performing on this data. This is more of a comment than a question. Something that might be interesting to those who are interest in what analytics can be used for.
Nice reference! The advantage of doing this with standardized testing is that all the questions are clearly tagged for topics and the "performance" of each question for the testing population is well understood, so it should be possible to connect the test results with specific lessons. One way they might do this is to match records of students who have completed specific modules in KA to their standardized test subscores for each topic. In Moodle terms, this would be like matching tags on questions (or question categories) to tags on learning activities and resources (possibly using ML to match keyword and other linguistic elements of questions and resources, rather than manually tagging).
I am one of the people who believe that high-stakes standardized testing takes up too much time that could be spent on learning, but switching to more frequent assessment that is used formatively (as Khan is suggesting) would be both more helpful to students and more accurate as a way to understand how well students are learning. But the danger of organizing a whole learning program based on any kind of testing subscores is that it can only be as accurate as the test. As I've often written, people have many different priorities in learning, e.g. preparation for scholarship or the workplace or citizenship, etc., but I don't know of anyone who thinks the purpose of school is to prepare students to pass standardized tests. (If that were a real world goal, I'd have a Nobel Prize by now.) The evidence linking standardized test results to real-world outcomes has been pretty weak. Many universities are moving away from using them in admissions, for example, because they are poor predictors of success at university.
This is why I try to encourage our learning analytics community to think about how real-world goals can be measured and used to drive learning analytics, rather than just assuming that passing grades or regular attendance/participation are goals in themselves.
I am one of the people who believe that high-stakes standardized testing takes up too much time that could be spent on learning, but switching to more frequent assessment that is used formatively (as Khan is suggesting) would be both more helpful to students and more accurate as a way to understand how well students are learning. But the danger of organizing a whole learning program based on any kind of testing subscores is that it can only be as accurate as the test. As I've often written, people have many different priorities in learning, e.g. preparation for scholarship or the workplace or citizenship, etc., but I don't know of anyone who thinks the purpose of school is to prepare students to pass standardized tests. (If that were a real world goal, I'd have a Nobel Prize by now.) The evidence linking standardized test results to real-world outcomes has been pretty weak. Many universities are moving away from using them in admissions, for example, because they are poor predictors of success at university.
This is why I try to encourage our learning analytics community to think about how real-world goals can be measured and used to drive learning analytics, rather than just assuming that passing grades or regular attendance/participation are goals in themselves.