Formative assessment, assessment for learning, and all that jazz

All teachers share a guilty secret. Our students do not learn what we teach. We teach wonderful lessons, our students seem to be alert and engaged, and yet, when we look at their work, it often bears no relationship to the content of the lesson. And because, as David Ausubel pointed out over 50 years ago, good teaching starts from where the students are, rather than where we would like them to be, we need to find out what our students did learn before we decide what to do next.

This idea that assessment should support learning, as well as judging how much of it has taken place, is not a new idea. In his work on mastery learning in the 1960s, Benjamin Bloom argued that because learning was unpredictable, it was essential that teachers found out what students were learning to inform their decisions. This was a radical idea because at the time it was widely accepted that some students would be more successful than others. Rather than accepting the “bell curve” of results, Bloom suggested that a normal distribution of results was a sign of educational failure, merely reproducing what nature provided. The job of the teacher was to destroy the bell curve: if some students needed more instructional input to become successful, then it was the teacher’s job to provide it.

The use of assessment to improve learning is sometimes called formative assessment or assessment for learning, and many writers use these two terms interchangeably, but there are important differences between the two.

Assessment for learning is any use of assessment that is primarily intended to improve, rather than measure, learning. If I tell my students on Monday that there is going to be a test on Friday, and the students study for the test, then the assessment is likely to improve the students’ learning, even if I do not actually give the test on Friday. Testing can also be highly effective as a study technique: as literally hundreds of psychological experiments have shown, taking a test improves learning even if the test is not scored, because it provides practice in retrieving items from memory, which is known to improve long-term learning.

Both of these uses of assessment could be described as “assessment for learning” because the purpose of the assessment is to promote learning, but they would not be examples of formative assessment, because to be formative, the results of the assessment need to be used to improve instruction. More precisely, an assessment functions formatively to the extent that evidence from the assessment is used – by teachers, by students, or by their peers – to make better decisions about the next steps in instruction.

It follows from this definition that there cannot be such a thing as a “formative assessment”, because the information from an assessment can be used formatively or summatively. If I find that a student knows 50% of his multiplication facts from 1 x 1 to 10 x 10, then that is a summative conclusion. However, if I look more carefully at the test results and see that the student is having particular difficulty with the seven-times-table, than that gives me, as the teacher, something to work with.

The same assessment, and even the same assessment results, can be used formatively and summatively, which is why the words “formative” and “summative” are best thought of not as kinds of assessments, but as different kinds of conclusions that might be drawn from assessment outcomes.

Dylan Wiliam is Emeritus Professor of Educational Assessment at University College London. In a varied career, he has taught in urban public schools, directed a large-scale testing programme, served a number of roles in university administration, including Dean of a School of Education, and pursued research on assessment for learning.