Should we give teachers the ‘benefit of the doubt’?
Earlier in the week, Schools Minister, Lord Nash announced that schools should be more like businesses and jettison underperforming staff. According to this TES report he’s reported to have said, ““I think one of the things that it’s easy to say … is that sometimes in education there is a tendency to give people the benefit of the doubt too often.” The consequence of this well meaning woolliness is that we consign children to a sub-standard education. Much better for school leaders to be like business leaders. The “best leaders in education” are “tough”, “have a real sense of pace”, and “realise the clock is ticking fast for their children.”
I’m not at all sure this is true. That’s not to say that I’m an apologist for shonky teaching; it’s more an acknowledgement that, while there’s little doubt there are some rubbish teachers out there, we don’t really know who they are. This will sound contrary to many people’s lived experience: in every school I’ve ever visited a cohort of teachers has been identified as requiring improvement. Every school leader I’ve ever spoken to about this admits that when the subject of underperforming teachers is brought up, they can bring to mind a list of teachers to whom they’d love to bid adieu. Other teachers are usually clear on who are the slackers, and even – especially – students feel pretty clear on who the worst teachers are. So why don’t we just sack ’em?
To answer that we need to think about the problems with human bias. We’re very good at identifying who we don’t like, but are extremely poor at understanding why this might be. We don’t like uncertainty so we rush to come up with plausible post-hoc rationalisations for our preferences. Frequently, we arrive at conclusions which cannot be supported by the data we have available. The trouble is, whilst data allows us to say with certainty that some teachers are definitely better than others, for all sorts of complex reasons it doesn’t allow us to reliably identify who those teachers are. The best measures of teachers’ performance we have are like a scales which gives us an individual’s weight to within +/- 50 pounds. This is sufficient to conclude that men are, on average, heavier than women, but tells us nothing useful about the weight of an individual. Does this matter? Well, that depends on what you do with the information. The very best* we can probably manage is to say with a probability of between 0.6 – 0.8 whether a teacher is performing well. This is not good enough for high stakes decisions about pay or employment but it should be sufficient to design effective training. As Dylan William spells out in Leadership for Teacher Learning, when we go with our gut, or place our faith in unreliable data, we end up promoting ineffective teachers and sacking good ones as often as the reverse. In other words, unless we want to give into prejudice, we really should be giving people the benefit of the doubt.
Teachers are, almost without exception, well-intentioned. Few people go into teaching for the easy life or high salary. And, as you might have noticed, we’re currently experiencing a recruitment and retention ‘crisis’. Schools struggle to persuade teachers to stay in teaching as it is, so sacking the ones who are prepared to carry on at the chalk face is a profligate waste of resources. It’s not as if we have serried ranks of excellent teachers queuing up to replace all the crappy ones! If business leaders routinely squander their most precious resource they quickly go out of business. If school leaders behave this irresponsibly they’re hailed as ‘tough’ and having ‘a sense of pace’. Instead, what we need is intelligent accountability where accountability is balanced with trust, and autonomy is earned. Nash is right that the clock is ticking for children – they only get one shot at education, and this is too important not to give teachers the benefit of the doubt.
* What constitutes the ‘very best’ is so expensive and unwieldy it never – or almost never – takes place in schools. From that we can infer that the reliability of judgements will be significantly less. See the MET Project final report for details.