Eric Mazur on Assessment

Eric Mazur is well known in science teaching circles for the innovative teaching approaches that he has popularized. He pays close attention to student behavior and data on student performance, and he also thinks deeply about what is going on in our classrooms. Most importantly, he isn’t afraid to try new things and advocate for them. So when he says something is wrong with the way we assess student work, one should listen. Or better yet, watch and listen because his 2013 lecture* “Assessment: The Silent Killer of Learning,” can be found on YouTube.

*Presented on Oct 29, 2013 as the Dudley Herschbach Teacher/Student Lecture

Mazur’s argument can be summarized this way: 1) students study for the test and 2) most tests are constructed so that they assess only low-level thinking skills (Mazur’s talk leans heavily on the Bloom Taxonomy of Thinking Skills which starts with Remembering as the lowest level skill and then leads upward through Understanding → Applying → Analyzing → Evaluating → Creating). The result of this one-two punch is that students study and practice the lowest level thinking skills and never get prepared to deal with the ambiguities of real-world problems. Mazur is not against testing per se, but rather what tests look like and how they are administered.

Mazur articulates the importance of higher-order thinking skills by pointing out that ‘thinking machines’ already do a very good job of operating at the Remembering and Applying levels. Robots can build a car. Google can find Planck’s constant for you. Mathematica will calculate your integrals. If a college education is to provide value that goes beyond the ‘thinking machine’ stage, it should educate students for thinking tasks that won’t eventually be performed by machines. This means raising the bar on what tests cover, how test questions are framed, and thinking more creatively about how tests are administered and graded.

Mazur marshals arguments like these, and also evidence drawn from a variety of teaching institutions (including his own classes at Harvard), to arrive at four suggestions regarding tests:

  1. Write test problems so that they mimic real life (his analysis of real-world problems vs. textbook problems is not to be missed) (see IF-AT testing and team-based learning below)
  2. Focus on feedback, not ranking (perhaps he would agree with this statement: the majority of assessment should be formative, not summative)
  3. Focus on skills, not content (see Understanding by Design below)
  4. Where possible, transfer the job of “grading” to someone other than the instructor so that instructor’s role is more that of a “coach” and less that of a “judge” (see CPR below)

Some links to the materials that Mazur mentions during his lecture:

This entry was posted in Education and the Mind. Bookmark the permalink.