Why We Don't Talk About Summative and Formative Assessment

Are we dividing assessments the wrong way?

On this Substack, you’ll see us use many words to describe assessments, but summative and formative won’t be among them. This may surprise you. These terms are everywhere in school assessment policies and discussions.

Teachers are often told they have autonomy over formative assessment, while their department or school sets rules around summative assessment. Assessment policies separate the two, listing distinct requirements for each, as if they are entirely different entities. These policies typically state that formative assessment informs responsive teaching, while summative assessment is used for reporting.

But this distinction is flawed. Many of you will recall Dylan Wiliam stating that formative and summative should describe how an assessment is used, not the assessment itself. An assessment is formative if it informs teaching and learning. It is summative if it is used to make judgments about attainment.

We want to take this idea further. We argue that all assessments must be both formative and summative—otherwise, they cannot fulfil their purpose. And what is their purpose? To promote learning.

Why must all assessments be formative? Because formative assessment contributes to the formation of learning. As our framework in this post shows, all useful school assessments shape learning, whether through the expectations they set or the inferences they generate for students and teachers.

Why must all assessments be summative? Because assessment involves making inferences. Whether it’s a quick classroom check or an hour-long exam, every assessment leads us to surmise what a student can do. If an assessment does not allow for this inference, it isn’t an assessment at all.

Ultimately, all assessments serve the broader goal of improving learning, but they do so in different ways—through expectations before the assessment, learning during the assessment, or the consequences of inference afterwards. Some assessments primarily influence students and teachers; others also shape decisions made by parents and school leaders.

(None of this is particularly controversial among assessment experts. Professor Paul Newton has written extensively about the limitations of these terms, but his insights remain buried in academic journals, out of reach for many educators.)

At this point, some of you may be thinking, “It’s just language—who cares?”

Well, we care, because language shapes how we create and use assessments. When we divide assessments into formative (assessment for learning) and summative (assessment of learning), we introduce misleading distinctions that distort their purpose. Here’s how:

1. ‘Big tests’ aren’t seen as ‘for learning’

If we believe that only formative assessments contribute to learning, we risk treating larger, high-stakes tests as purely bureaucratic exercises—measuring attainment for a spreadsheet rather than supporting student progress. But summation is not the point. The real question is: How can the existence of this test help the student learn better?

2. We assume accuracy matters more for summative assessments than formative ones

A common claim is that summative assessments must be highly accurate and reliable, while formative assessments can be ‘rough and ready.’ But this is misleading. All assessments require a reasonable degree of accuracy because decisions—whether in the moment or at the end of a course—depend on valid inferences. Crispin Weston wrote eloquently on the importance of accuracy in tests used for formative purposes in his seminal 2017 post called ‘Curriculum Matters’.

3. We oversimplify assessment task choice

Another mistaken belief is that only large assessments should focus on extended writing or complex ideas, while smaller assessments should test isolated concepts or skills. While this approach may work in some cases, it is ultimately subject-dependent. In some disciplines, short assessments can assess complex reasoning, while in others, end-of-year assessments might optimally break down into discrete, skill-based tasks. Assessment design should be driven by disciplinary needs and complex considerations, including their effects on motivation. (We’ll write more about this in the very near future.)

4. We assume relative attainment only matters for ‘big tests’

Another myth is that only summative assessments require measurement on a consistent scale (e.g., grades or scaled scores), while formative assessments do not. In reality, relative performance is just as crucial in formative contexts—teachers constantly compare students' progress to interpret levels of understanding and decide whether support is best directed. And there are many end-point or ‘big test’ situations where absolute judgements of mastery are far more meaningful than relative or ranking approaches.

Rethinking Assessment in the Age of AI

The acceleration of AI-driven assessment will force us to rethink these distinctions. Consider platforms like Sparx Maths or Duolingo. These tools maintain an ongoing picture of a student’s attainment based on responses to questions which could equally be used to predict future performance or generate short term personalised learning goals. Should we call them formative or summative? The distinction is meaningless. The system is both assessing what students know and shaping their learning in real time.

The Conclusion: All Assessments Must Be Both

Relying on the words summative and formative leads to dysfunctional assessment practices in schools. All assessments must be both, or else how can they serve their fundamental purpose—to promote learning?