If assessment is about evidence of learning, the big question is: what are we learning? What are we measuring? The term construct is a fancy word in assessment that means the big idea behind what is being taught. You can't see this thing directly. You need to infer it from what learners do, say, make or write.

Teachers use constructs all the time, formally and informally. Defining the construct first, and what evidence of learning it would look like, is a really important step in assessment. It's a little bit of a Goldilocks situation. A good construct is not too big and not too small. It's not an entire subject, but it's not just one skill either. A construct is abstract, an idea. It's latent, so you can't see it directly. It's made up of connected elements like knowledge, skills, and dispositions that improve together as you learn.

Once you've got a construct, you want to define it, describe what it looks like at various levels of quality, and then disaggregate the sub-components that make it up and describe increasing quality of performance in them all separately.

Creating a map of the construct first can help direct teaching. What kind of teaching would get students to a high level of ability with whatever is in the construct? It's like your North Star.

In the dry world of university, we say constructs should be assessed separately, but we know in practical classroom situations that's not the case. Sometimes teachers really do want to teach a student geographical thinking skills but also their handwriting or spelling at the same time. We can't let these theoretical assessment concerns get in the way of practical classroom needs.

Getting clear on what the construct is creates a direction for our teaching. It helps us to interpret evidence of learning. It can tell us where on a construct map the learner is and what they need to do next. In that sense, there are similarities with Hattie's suggested feedback questions: Where am I going? Where am I now? What next? Patrick's developmental angle would add that to answer those questions we need evidence, not inference. Our measurement tools have match that observed evidence as much as possible.

[Photo by Vlad Hilitanu on Unsplash]