It’s always the same, whenever we concentrate on the steps, our footwork falls apart, but when we work on footwork we forget how to do the steps.
Peter, tango teacher

Improving teaching is just like dancing tango…  Well, in one respect: when I refine one teaching technique, the others go by the wayside.  The overall result of my attempts to improve individual aspects of my practice is, I hope, a better lesson, but whatever I’m not working on tends to atrophy.

So while my posts on hinge questions easily outstrip the rest of the blog in popularity, my own skill in using them deteriorates.  I’m not pretending false modesty, simply stating that I was using hinge questions better six months ago than I am now.  Last term I focused on learning intentions: I evaluated changes, reflected, refined and as a result I’m probably using learning intentions better than I ever have before.  I was still using hinge questions in lessons, but I simply collected and used student responses without thinking them through.  They remained useful but this is a lazy exercise on my part and it’s not as effective as it could be.


Built into my ideas about how I’d improve this year was one about returning to specific skills I’d worked on last year and refining them.  Hinge questions were an obvious choice, doubly so because I’d undertaken to speak at TLT13 on this topic.

How do I design a proper hinge question?

What I tried to do at TLT was summarise the argument of a paper by Caroline Wylie and Dylan Wiliam which explains how to design a hinge question properly, using a framework to create what they term a ‘semi-dense item.’   This was my third full reading of the paper and I think I’ve got it now, by my exposition may still be unclear: at TLT, some people liked having the steps presented to them and I think others found it pretty unhelpful.  If you get fed up, skip to the sentence in red at the bottom which is the brief summary.

We are aiming to create a multiple choice question which we can use to check whether all students have understood a key concept immediately.  What I’ve done is outline each step and give an example of how it looks with a real question – about what makes a good source of evidence for a historian.

Responses = choices available to students

Cognitive rules = thought processes/ideas students have

1)-  Interpretable responses:  If students choose response B, I know what cognitive rule underlies this choice.

First draft of some options to a question designed to test understanding of what makes a good historical source:

A We should use the best source for the task.
B Primary sources only give us people’s opinions.
C Secondary sources give us a fair treatment.
D Both primary and secondary sources may be useful to find out what happened.

Of these, I would certainly ditch A and D because they are not interpretable.  A doesn’t tell me what they think makes something the best source and D doesn’t tell me why they think the sources are useful.

2)- Response discrimination: If students choose B, I know that they can only be thinking of one cognitive rule, not more than one.

C Secondary sources give us a fair treatment of an event

There are several reasons why a secondary source might give us a ‘fair’ treatment of a historical question, for example, the author of a secondary source is able to draw on many other sources; alternatively, they may be more detached from the events.  So response C fails this test.  Instead, I should break the response into separate points – for separate cognitive rules.  For example:

E Secondary sources can give a more complete picture of an event
F Secondary sources can be written by people who are less emotionally connected to an event, so they can tell us the truth.

So the different cognitive rules, the reasons why the student thinks a secondary source may be fairer, are clear from their choice of response.  (It’s worth noting that Wiliam points out how good for us as teachers this can be in terms of improving our subject knowledge – as I constructed these questions I was forced to re-evaluate what I know and think about evidence in history).

3) Rule discrimination: If students think Y, they will have to choose B.

F Secondary sources can be written by people who are less emotionally connected to an event, so they can tell us the truth.
G Secondary sources are unbiased.

Some students believe that a secondary source, such as a history book, must be true.  Students following this cognitive rule may choose F or G, so I will amend F.

F Secondary sources can be written by people who are less emotionally connected to an event.
G Secondary sources are unbiased.

If students believe a secondary source is completely truthful there is only one conceivable choice they might make.

4) Exhaustive set usage: Whatever cognitive rules students think of, a response fits it.

Here are a set of possible cognitive rules, which I hope is near-exhaustive:

– Primary sources are likely to offer a limited perspective
– Primary sources offer an insight into people’s actions and reactions at the time
– People who weren’t there can never know what happened
– Secondary sources may offer hindsight
– Secondary sources may offer completeness
– Secondary sources are unbiased

(What did I miss?)

The result – Semi-density: I have stuck all these rules together.

Which of these is not a valid consideration when evaluating evidence?

A Primary sources offer a view from the time they were made.
B Primary sources only offer us one person’s perspective.
C Primary sources provide us with an insight into people’s responses to events.
D Secondary sources are less useful because the authors weren’t there.
E Secondary sources offer the benefit of hindsight.
F Secondary sources can be written by people who are less emotionally connected to an event.
Secondary sources can give a more complete picture of an event drawing on many other sources.
H Secondary sources may be biased.


Whatever a student chooses, I will know what they were thinking, that they can only have been thinking that, and that there’s nothing else they might have thought of that I’ve missed.


I’m still not entirely happy with my example question!  It has too many options (one of the criteria for a hinge question is that students should be able to respond in two minutes (maximum), preferably one) and I now wonder whether it would be helpful to divide it into separate primary and secondary source questions.

Moreover, this question is prey to envelopment by the shades of grey of which history is made, some secondary sources seek objectivity and others are entirely biased.  A division between primary and secondary sources is convenient as an introduction, but ultimately unhelpful, because so many sources fit awkwardly or not at all into this framework.

I think the construction of hinge questions is easier in subjects like maths and science were cognitive rules are clearer (the laws of gravity are less susceptible to differing interpretations, for example).  On the other hand, even if they aren’t hard and fast, cognitive rules do exist in the arts – and I suspect that often I just don’t think them through as clearly as I could, and, perhaps should.

Next steps:

Further reading

I have collected all my posts on hinge questions, together with a couple of presentations and links to other teachers experimenting with them, in a ‘Hinge Questions Hub.’

This post represents an attempt to put my TLT Southampton Presentation into written form – the powerpoint may make things clearer (and contains example maths, science and English questions).

I continue to recommend Caroline Wylie and Dylan Wiliam’s paper on this; (it also contains more examples for maths).

I have previously written on introducing hinge questions to my teaching, issues with applying them to history and I have the beginnings of the start of a history collection.

Bruno Reddy has found a fantastic teacher-created, free, growing collection of hinge questions for maths created by Craig Barton – which may soon grow to include history questions too.

My own task for this half term will be to create more good hinge questions – I’ll upload them somewhere useful when they’re done.