Quick thoughts and questions about research

I’m increasingly fascinated by it and try to use it to inform my decisions at school more and more, but I think using research, even quantitative research, can be tricky if you don’t look at the bigger picture. Andrew Sabisky got me thinking about it in more depth when he tweeted this a few days ago:

So class size doesn’t matter much. The Sutton Trust Teaching and Learning Toolkit is an obvious place to start when looking at different measures and their impact on attainment and it agrees: reducing class size has a low impact, but high cost. Is it that simple though?

Take my phase this year: we employed an extra teacher, which reduced class sizes. This enabled me, and the other teachers, to spend more time on something which does have a very high impact: meaningful feedback. If we had carried on doing what we’d always done, just with smaller class sizes, then I doubt there would have been much of a difference, but using a reduced class size to focus on something that could make a real difference did just that. So, for me, albeit indirectly, class size has mattered.

You could go on to say the same for other measures that have little or no impact according to the Toolkit, such as TAs. However, if a TA is deployed in such a way that he or she feeds into or enhances a measure that does have impact (feedback again is a good one) then having a TA then does become a resource that has an impact. Doesn’t it?

How do you use (or try to use, like me) research to inform you practice? How do different strategies impact on one another?

Number wheel by HeavyWeightGeek
Attribution License


  • I was somewhat constrained by the limits of 140 characters, but yes, we do measure outcomes in overly crude ways. Even if class size and TAs do not have much of/any impact on pupil test scores, reductions in teacher stress (if that is so) represent a real benefit, particularly considering the ridiculously high teacher burn-out rate. Not only is this desirable in itself, but lower teacher burnout may have positive benefits for pupils in terms of mental health, attachment, confidence, and security. All these things are not best measured by standardized tests. You could also make the argument that lower teacher burnout will make teaching a more attractive profession in the long run, slowly pulling in higher quality candidates (vide Finland where teaching is just about the highest-status thing you can do). That should have an impact in the long run on pupil achievement.

    TAs should have an impact on pupil achievement, by the way – it’s just that across the system they are used so badly (as a pacifier for the troublesome kids, basically), that they don’t. If we used TAs better, they would work. So Peter Blatchford argues, and after some thought I’ve come round to agreeing that he’s probably right.

    These issues of measurement are not straightforward, particularly when you consider that you really want to think about what outcomes you want to measure and directions of effect you predict before beginning your research – otherwise it’s all too easy to capitalize on the occasional statistically significant findings that any dataset will throw up. None of it, to my mind, is an argument against quantitative research, but an argument for better designed, larger-scale, and well-resourced studies.

  • Pingback: Quick thoughts and questions about research | The Echo Chamber

  • Good post and reply on why we need to treat research findings with care. They almost inevitably over-simplify simply in order to make the investigation practicable. You could argue that T.A.’s would be more effectively used if teachers had less to do because their classes were smaller. It’s all just too complex and inter-connected to call. In my experience, it is also uncannily rare ever to see any educational research reported where the findings did *not* support the original hypothesis. I wonder why that might be…

Leave a Reply

Your email address will not be published. Required fields are marked *