I had the opportunity to revisit Wiliams’ 2018 book, Embedded Formative Assessment, while looking for definitions of “formative assessment.” The first two chapters are simply priceless. Beyond those, the other chapters include a general problem to be solved, and then practical techniques on solving them. Here’s an overview of what I consider the best parts…

But first, one of the lasting thoughts Wiliam makes about the case for formative assessment in Chapter 2 is “we cannot predict what students will learn as a result of any particular sequence of instruction,” then continues, “formative assessment involves getting the best possible evidence about what students have learned and then using this information to decide what to do next” (p. 56). That, right there, is something every educator needs to hear and reflect on.

Intentions/Objectives, Criteria, Rubrics
There are some great quotes about what is now seemingly the obligatory posting of objectives on the board, like “sometimes telling the students where they are going completely spoils the journey! Indeed, in conversations with students, many have told me that always receiving the objectives at the start of the day turns them off” (p. 63). This makes sense, and we should really examine a strange policy/expectation that usually just amounts to learning fatigue.

Beyond that, I’ve been thinking more about the idea that a student can learn something new in exactly one class period before moving onto a the next thing the next class. It’s kind of insane, actually, especially what we know that learning rates differ from student to student, etc. Some stuff in Wiliam’s Chapter 3 really touches upon this idea, like writing super detailed rubrics so “the clearer you are about what you want, the more likely you are to get it, but the less likely it is to mean anything” (p. 66, while citing Kohn, 2006). I also like the table on disentangling learning intentions, with the first example being “to be able to write instructions on how to change a bicycle tire” as the confused version of what should be “to be able to write clear instructions,” and that the context of learning is “changing a bicycle tire” (p. 66).

Another observation of mine that I call “doing something might not be doing anything,” has support from Wiliam, too, as “it is relatively easy to think up cool stuff for students to do in classrooms, but the problem with such an activity-based approach is that too often, it is not clear what the students are going to learn” (p. 66). This section helps teachers think through what it is students should be learning, and how to articulate that.

I also appreciate the warning that “rubrics seem to focus more on grading rather than improvement” (p. 72). This is especially obvious in rubrics with several domains up and down and levels across the top that look like a big matrix with points that add up to a final score in the bottom right.

Practical Techniques. Of the practical ideas that Wiliam shares in this section to help students understand objectives, I particularly like showing students correct and incorrect examples (i.e., an example and a non-example).

Evidence of Learning
This is a very good chapter. Not only do teachers spending too much time thinking about cool activities often forget about the main learning intentions/objectives (p. 66), they also forget about HOW they will find out where students are in their learning (p. 83). The backwards planning of Understanding by Design is supposed to address this, but in practice it seems like the process is so involved that student teachers-turned teachers abandon the approach once in their own classroom. Unfortunately, that often means they’re generally thinking less about assessing, and/or fusing that concept with grading.

This section has a lot of great stuff on developing valid questions that show learning. Wiliam states, “there are only two good reasons to ask questions in class: (1) to cause thinking and (2) to provide information for the teacher about to do next” (p. 90).

Practical Techniques. I like the various strategies that lead to equitable outcomes, like never allowing students to choose whether to participate, likely resulting in the same students volunteering over and over. I also really like the Question Shells to avoid yes/no responses. For example, instead of “is a square a trapezoid,” asking “why is a square a trapezoid” is more likely to show student thinking (p.98). Wiliam also distinguishes between discussion questions and diagnostic questions, the later of which being designed to check whether students have understood something before moving on, which rests on a hinge. Hinge-point questions are awesome, but take a bite of practice to get right, which is why Wiliam put out a call, writing “sharing high-quality questions may be the most significant thing we can do to improve the quality of student learning” (p. 121).

Feedback
This section is great, touching upon all the heavy hitters, like how “the effect of giving both scores and comments was the same as the effect of giving scores alone” (p. 125), how poor feedback can actually lower performance, and that feedback is formative “only if the learner uses the information fed back to them to improve performance” (p. 140).

Practical Techniques. I like the idea of not providing feedback unless there is class time to act upon that feedback.

Peer- & Self-Assessment
I’m drawn to the self-assessment chapter more than peer-assessment, but you should definitely read it. Otherwise, it has been suggested that self-assessment is necessary, not optional, for students to make sense of and act upon feedback. Therefore, any way to make this process better should be given some attention.

Practical Techniques. I’ve used the Preflight Checklist idea of having a classmate “sign off” on a student’s work with great success, though always tried to simplify tasks so that it wasn’t necessary. As for self-assessment, I’m less-inclined to recommend the in-class self-assessments like traffic lights, disks, or cups that students show the teacher during a lesson. Instead, portfolios are king. Read that part.