Good schools make good teachers

What kind domains, wicked traps and school culture mean for improvement

It’s become an axiomatic truth that “the quality of an education system cannot exceed the quality of its teachers”.¹ To what extent this is actually true is arguable, but it’s fair to say that the quality of teaching in a school is both important and something that can be improved. If we believe that bad teachers are dragging down our education system, this might make us keen to identify which teachers are good and which are bad and justify campaigns to rid the system of the bad. Of course, this is overly simplistic. More charitably, if we could reliably identify good teachers, then we might also be clearer about what it is they do and make better efforts to train less effective to teachers to do what the most effective do well.

This is important because, as Rob Coe and colleagues put it, “what teachers do, know and believe matters more to the achievement of students than anything else we can influence”.² So, when thinking about how to go about improving a school, two questions are important:

1. Do you believe all teachers can improve?

2. Do you know how to get the best out of all teachers?

The answer to the first question feels straightforward: either you do or you don’t. But what about that teacher? What about the one you’d really rather wash your hands of and shunt off on to someone else? Are some teachers lost causes or can every teacher improve? In typically provocative style, Dylan Wiliam has said, “The most important question we should ask teachers is, ‘Do you think you need to improve?’ If they say, ‘Yes,’ work with them. If they say, ‘No,’ fire them.”³ While this may feel unduly harsh, it contains an important element of truth. If teachers don’t believe they need to improve they may be a lost cause. Teachers who don’t believe they can improve are a different kettle of fish. Sometimes we have to believe for them.

Surprisingly, perhaps, teachers don’t necessarily get better with experience. Various studies have demonstrated that when we compare students’ outcomes to teachers’ years of experience, we tend to see sharp rises in the first few years but these flatten off and even decline over time.⁴ Other studies have produced similar findings.⁵

Teachers’ years of experience correlated with students’ mathematics test scores. Source: Kraft and Papay (2014)

Wicked and kind domains

Why should this be? One way to understand this pattern is through Robin Hogarth’s distinction between kind and wicked learning environments.

In a kind domain, feedback is reliable, immediate and unambiguous. Think of a chess player, a surgeon, or a meteorologist. The chess player sees the outcome of every move in real time; the surgeon knows from the body’s response whether a procedure has succeeded; and the meteorologist, though dealing with complexity, receives regular, measurable feedback from forecasts that can be objectively checked against reality: either it rains or it doesn’t. Each operates in a system where results are transparent enough to support genuine learning.

In contrast, wicked domains muddy this loop. Feedback is slow, partial, or misleading, so errors persist uncorrected. A striking example comes from a classic study of radiologists. Researchers found that even after years of experience, doctors were no more accurate at identifying signs of cancer in X-rays than their less experienced colleagues.⁶ Because real cases rarely offer clear, timely feedback - patients may never return, or diagnoses may remain uncertain - radiologists, like teachers, can spend decades practising without ever knowing precisely when they were right or wrong. The problem is not effort or intelligence, but environment: without accurate signals, experience merely reinforces existing habits.

Experience only breeds expertise in kind domains. In wicked ones, it can just as easily reinforce bad habits. Teaching contains elements of both. Classroom management is relatively kind. The feedback loop is short and visible istudents provide real-time, unambiguous feedback on the effectivenees of your actions. But fostering learning is wicked: the real effects of teaching often emerge months or years later, filtered through the noise of curriculum, assessment, and home life.

This helps explain why experience does not automatically yield improvement. Teachers receive plenty of feedback, but much of it is wicked: confounded by unreliable proxies such as compliance, apparent engagement, or short-term performance. Without structures that make learning outcomes visible and trustworthy, experience merely hardens intuition rather than sharpening judgment. In short, teachers get better at teaching as it looks, not necessarily teaching as it works.

Where you work matters

All this being said said, the type of school you work in matters. When schools are ranked on how supportive they are, we can see that teachers working in more supportive professional environments tend to become more effective over time than colleagues working in less supportive environments. Over a five-year period, teachers working in the top 25% of schools improve, on average, 20% more (in terms of student outcomes) than teachers working in the least supportive quartile.⁷

John Papay and Matthew Kraft identified four features that distinguish schools where teachers continue to improve, ones that follow the green-line trajectory.

• Ordered environment: Behavioural norms are clear and consistently upheld. Teachers know that leaders will support them in maintaining order, so time and attention can be directed toward teaching rather than firefighting.

• Professional development: Time and resources are deliberately focused on refining teachers’ instructional expertise. Development is sustained, relevant, and closely tied to classroom practice rather than detached workshops or compliance exercises.

• Constructive monitoring: Performance management is fair, proportionate, and developmental. Teachers receive regular, specific feedback that helps them improve rather than simply judge them.

• School culture: Relationships among staff are marked by mutual trust, respect, and professional candour. Collaboration is routine, not performative, and collective effort is anchored in a shared commitment to improving student learning.

None of this should be surprising. We know from research into a variety of domains that three facts consistently predict the working environments in which workers are most likely to thrive:

1. Supportive learning environments,

2. Concrete learning processes

3. Leadership that reinforces learning.7

Additionally, teachers who work in schools where they are encouraged to collaborate frequently, receive meaningful feedback about their teaching and are recognised for their efforts will on average improve at faster rates.8

Good bets for improving teaching

Improving teaching is not a matter of inspiration or luck. It depends on designing conditions in which better practice becomes the natural consequence of ordinary work. Some policies offer little more than slogans; others- what we might call good bets - have a credible evidence base and a plausible mechanism for improvement. They share a few common traits: they make feedback more accurate, practice more deliberate, and professional life more coherent. The points below outline several of these good bets and sketche what they look like when they move beyond rhetoric into day-to-day practice.

Supportive learning environment

• Instructional coaching which focuses on aligning teaching to the long-term retention and transfer of knowledge rather than short-term performance in the classroom.

• Teachers negotiate performance targets that align with long-term learning goals and are within their power to attain.

• Failures are treated as an opportunity to learn from mistakes rather than causes for distrust.

• Teachers earn the autonomy to decide how they should teach.

Concrete learning processes

• Professional development that focuses on sustained implementation of a few ‘high yield’ strategies (e.g. providing formative feedback).

• Coaching is based on the practice of discrete elements rather than broad principles

• Teachers earn the autonomy to focus on their own training goals.

Leadership that reinforces learning

• Accountability processes are aligned with long-term learning goals.

• Teachers are held to account for the trust they are given.

• Leaders provide opportunities for teachers to engage in reciprocal observation of colleagues.

• Leaders ‘look at’ what teachers are doing and ask probing questions rather than ‘looking for’ lists of approved activities.

• Leaders support teachers in using well-constructed systems for managing students’ behaviour.

Teacher collaboration

• Teachers are given scheduled opportunities to discuss and share successful instructional practices.

• Teachers are given scheduled time to plan and evaluate the curriculum together.

Meaningful feedback

• Feedback is focused on long-term learning goals (i.e. whether students have retained and can apply information taught in previous lessons).

• Feedback is negotiated; if teachers request feedback they are more likely to act on it.

• Feedback is given on aspects of teaching that are within teachers’ ability to control.

Recognition of effort

• Leaders acknowledge that teachers are working hard and seek to reduce unnecessary workload.

• Teachers’ efforts are shared among staff and treated as a collective source of pride.

No matter how talented an individual may be, whether teachers continue to improve depends, to a large extent, on the culture of the school in which they work. As you can see from the list above, few of these items are within a teacher’s control. If school leaders do not actively put these things in place and guard them as sacrosanct, teachers – and therefore students – are less likely to improve.

In fact, as Adam Boxer has recently argued, there are many schools in which systemic pressures are almost certainly making teachers worse! As he says,

[T]o improve every school. We need every teacher to be in a good school - to be in a social and organisational context where good teaching is the norm, and all staff have access to high quality training. And we need, relentlessly and systematically, to find the schools where this isn’t the case and support them to improve.

Let’s recast Wiliam’s question to one aimed at senior leaders: do you think all teachers can improve? Think carefully. If you say ‘yes’, there’s hope for you. If you say ‘no’, maybe it’s you who needs firing!⁸

This second question is an invitation to consider what was covered in Chapter 1. By now, you should be ready to accept that individualistic certainty tends to be harmful, whereas collective approaches to knowledge will tend to pay off. Although you may have all sorts of great ideas for how to help teachers improve – some of them may have an impressive track record of success – you can’t know how best to reach all teachers in every circumstance. Some approaches show greater promise than others.

For instance, although the instructional coaching model has a strong evidence base and is probably a good bet, there are very many ways to get it wrong. Hundreds of leaders are finding out that handing over cash for fancy coaching systems is not - quelle surprise - magically transforming their schools.

Admitting your lack of certainty creates the possibility that you may have the humility and realism to seek advice and support from others, and it makes it more likely that you’ll approach the work of helping teachers to improve with patience and tentativity.

1. This is often attributed to the 2007 McKinsey report by M. Barber and M. Mourshed, How the World’s Best Performing Education Systems Come Out on Top but a footnote on p. 19 reveals that the quotation actually comes from an unnamed South Korean government official.

2. Rob Coe, Christian J. Rauch, Stuart Kime and Dan Singleton’s Great Teaching Toolkit: Evidence Review (2020) synthesises decades of educational research to identify the elements of teaching that most reliably improve student learning. The report groups effective practice into four broad dimensions: understanding the content, creating a supportive learning environment, maximising opportunity to learn, and activating hard thinking. It argues that what teachers do, know, and believe exerts more influence on pupil outcomes than any other controllable factor, and that professional development should focus on deliberate practice within these areas, supported by feedback and a culture of continuous improvement.

3. See Dylan Wiliam at the Schools Network Annual Conference, 20 December 2011. Comment is at 34:50.

4. S. G. Rivkin, E. A. Hanushek and J. F. Kain, Teachers, schools, and academic achievement. Econometrica, 73(2) (2005), 417–458. This landmark econometric study examines the relative impact of teachers, schools, and other factors on student performance. Using large-scale longitudinal data from Texas, the authors find that teacher quality is the single most important school-related factor influencing student achievement. They show substantial variation in teacher effectiveness, only weakly linked to experience or credentials, and conclude that improving teacher quality would yield far greater gains than structural reforms such as reducing class sizes or increasing school resources.

5. For example, D. Boyd, P. Grossman, H. Lankford, S. Loeb and J. Wyckoff, Overview of measuring effect sizes: the effect of measurement error. Brief 2. National Center for Analysis of Longitudinal Data in Education Research (2008); D. N. Harris and T. R. Sass, Teacher training, teacher quality and student achievement. Journal of Public Economics, 95(7–8) (2011), 798–812; J. P. Papay and M. A. Kraft, The myth of the performance plateau. Educational Leadership, 73(8) (2016), 36–42; J. E. Rockoff, The impact of individual teachers on student achievement: evidence from panel data. American Economic Review, 94(2) (2004), 247–252; M. Wiswall, The dynamics of teacher quality. Journal of Public Economics, 100 (2013), 61–78.

6. Kundel, H. L., & La Follette Jr, P. S. (1972). Visual search patterns and experience with radiological images. Radiology, 103(3), 523–528. Kundel and La Follette’s examined how radiologists interpret X-rays and found that years of experience did not necessarily lead to greater diagnostic accuracy. Experienced radiologists were faster at scanning images but not significantly better at identifying subtle abnormalities such as early-stage cancers. The study showed that, in the absence of reliable and immediate feedback, expertise in radiology can stagnate, since practitioners rarely receive clear evidence of when their judgments are right or wrong.

7. Papay, J. P., & Kraft, M. A. (2016). The Myth of the Teacher Performance Plateau. Educational Leadership, 73(8), 36–42. In Papay and Kraft challenge the widespread belief that teachers stop improving after their first few years in the profession. Drawing on longitudinal data, they show that teachers continue to develop when they work in supportive professional environments that promote collaboration, meaningful feedback, and sustained learning. The so-called “plateau,” they argue, is not a natural limit of human capability but a failure of organisational design.

8. This is a joke. Sort of. Obviously, firing people is rarely simple and not something most people ever find comfortable. The ins and outs of formal disciplinary and capability procedures and employment rights lie far beyond the scope of this article. There are standard protocols that schools should follow which you can find out more about in: Teaching Regulation Agency, Teacher Misconduct: Disciplinary Procedures for the Teaching Profession (May 2020).