Wednesday 22 April 2020

Polythetic Assessment: What it is and why we need it

There are fundamental differences in the ways we categorise and group things, including learners. When we design assignments and exams, the point of the exercise is to create a common set of variables against which the performance of each student can be assessed. Effectively the students form a "cluster" around the measurement of these variables.

This approach to clustering around a common set of variables is called "monothetic" clustering.

In ordinary life however, we often don't cluster things monothetically.  When we select our favourite music for one of those Facebook/Twitter challenges which everyone is doing now, the criteria for selection into the cluster of "my favourites" does not depend on common variables. However, intuitively we know that there is some kind of deep pattern which unites all of these things, without necessarily being able to put our finger on exactly what it is. In a sense, the point about these Facebook games is to provide some evidence for what these deep patterns might be, and in the process, who we are in making the judgement.

This approach to clustering is called "polythetic". Its a term used today by the data scientists, but it has a longer pedigree in phenomenology. It was first used by Edmund Husserl to describe "family resemblances" between phenomena which could only be revealed step-by-step over time. Later Husserl's follower, Alfred Schutz, used the term to describe the experience of music in a striking paper called "Making Music Together" (Social Research, 1951)

The problem with monothetic judgement or clustering is that it forces things into a rigid framework. But most things, including student work, can be good or bad in many different ways. For those who argue against Learning Outcomes, the wish is to find ways of making defensible judgements about student work whilst embracing the diversity of the goodness or badness of that work.

More deeply, the difference between monothetic and polythetic judgement is that the former is 2-dimensional and the latter is 3-dimensional. To understand this means to dig-in to the logic of patterns, and understand that no pattern can exist which doesn't feature some aspect of "nothing" within it which segments one part of the pattern from another, and it can be shown mathematically that "nothing" entails looking at the world in 3 dimensions. To move to a polythetic approach to assessment is a bit like introducing perspective and a "vanishing point" into art. Indeed, the idea of "vanishing" is extremely important to understand polythetic assessment. It is assessment about deep patterns in things, not surface features.

I've made a video introducing "polythetic assessment" which describes the importance of measuring "nothing" when we look at phenomena. This work draws on physics, mathematics and biology - but it has direct applicability to educational assessment in ways which I will illustrate in future posts.

If we want more personalised, self-directed, creative learning, then we will need to find new ways to assess. Polythetic assessment presents a starting point from which we can look at the learning process afresh.

No comments: