What Makes a Great Teacher, and Why It's Important

Close your eyes and imagine this. We’re in Spain. Sun, beaches and Sangria, the works. Now, we need to hire a new teacher for our English language Academy and there’s two candidates. The first is a degree educated native English speaker. They’ve just arrived having been teaching English in China for the past 3 years and don’t speak Spanish. The second candidate is a degree educated Spanish native who also speaks English proficiently (to a C1/C2 level).

Who would you choose?

Most schools hire the first. Private academies advertise for native speakers, parents ask for one, and when researchers interviewed the managers of six language schools in Catalonia they found the same preference under the managers’ denials. But when you look around at job notices, even the shop fronts of the different academies, it’s obvious that the native teacher is the most valued. Therefore hiring candidate number one feels the obvious decision, because the first candidate knows more English.

However, knowing more English is not what predicts whether students learn. It’s actually the second candidate that has the quality that does and even if the hiring process knows it, it ignores it. Nobody measures it afterwards either.

The clearest evidence comes from Germany, where a team led by Jürgen Baumert followed 181 maths teachers and 4,353 students through year ten. They tested the teachers twice. One test measured how well they knew the mathematics. The other measured what researchers call pedagogical content knowledge. Can the teacher see more than one way to solve a problem? Can they explain it with a good example? And shown a student’s work, can they spot the error, say why it was made, and predict the mistakes the class will make next?

The two went together, but only one did the work. After allowing for what students already knew and where they came from, pedagogical content knowledge explained a large share of the difference between classes in how much students learned that year. Knowing the maths predicted less, and had almost no effect on the two things that mattered most in the lessons, which were setting tasks that made students think and helping them one by one. The authors’ conclusion was that a teacher cannot have the second kind of knowledge without the first, and that the first on its own is not enough.

So the first mark of an effective teacher is knowing where students go wrong. In language teaching that knowledge comes from having learned the language yourself. Ask most people on the streets of England to dissect the grammar in a sentence and they’d struggle. I was the same until I started teaching. In fact I was learning the English language on the job - to the detriment of my students. My (Spanish) colleagues sometimes even gave me lessons on English grammar before I was due to take a class.

Who’s the better teacher there?

The Spanish native teacher would know why Spanish learners say “I have twenty years”, “the people is” or “I am agree”. These are classic mistakes Spanish learners make. They can be learned through experience of teaching in Spain. The Spanish teacher can relate to their students much more easily, and know from the off that these mistakes are coming.

There is a nuance to this debate. For higher level learners it matters less and the native knowledge around phrases, accents and culture then come into play. But the lines around the argument get blurred.

And the student perception is interesting. When 422 Hungarian learners compared their native and non-native English teachers, they rated the natives higher for conversation and pronunciation, and the non-natives higher for explaining grammar, preparing them for exams and anticipating their difficulties.

A study of 76 university students in the Basque Country found the same split. Péter Medgyes, who has spent a career on this question, puts the ability to predict language difficulties near the top of what a teacher who learned the language brings.

That knowledge shows in the work a teacher sets. In the German classrooms, the teachers whose students learned most set tasks that asked for reasoning rather than repetition, and knowing how students go wrong predicted that. Knowing more maths did not.

It shows again in how a teacher helps one student at a time. The German study measured this too: whether a teacher noticed when a student was stuck, treated the mistake as something to work with rather than something to correct and move past, and adjusted the explanation to that student instead of repeating the one the class had already heard. Thirty students at different stages need thirty slightly different explanations, and a teacher can only supply them if they already know the handful of ways the topic goes wrong. Again, that knowledge predicted the help, and subject knowledge on its own did not.

And it changes what a teacher says when the mistake arrives.

In the most cited analysis of feedback, Avraham Kluger and Angelo DeNisi pooled 607 results from 131 studies and found a solid average gain, but 38% of the interventions made performance worse. The difference lay in where the attention went. Comments that kept it on the task helped. Comments that moved it onto the person, how able they are and how they compare with others, tended to backfire.

A later analysis of 435 studies of student learning found a smaller share of harm, about one in six, and the biggest gains from comments on the task, the process and how to manage your own learning.

Feedback is among the most powerful things a teacher does, and among the easiest to get wrong. Getting it right means naming the error, and only a teacher who saw the error coming can do that on the spot.

The quality that predicts learning does not show on a CV. What does show is a degree, a confident interview and a training certificate, and none of them predicts how much students learn.

A good degree impresses, but intelligence does not predict how much students learn. Of 27 studies reviewed in 2020, most of them American, the few that used intelligence tests found no link or a negative one, and the rest at most a small positive one.

Confidence impresses at interview and does not hold up after it. Across 43 studies, a teacher’s belief in their own ability correlated with observers’ ratings of their teaching but barely at all with what their students achieved.

Growth mindset, the idea that ability grows with effort, has been the most widely adopted idea in schools over the last decade, and training in it looks good on any CV. But the problem is that it has a small effect on achievement at best, which is what two large reviews in 2023 found.

So the panel is guessing, and a wrong guess is expensive. The strongest estimate comes from the United States, where Raj Chetty, John Friedman and Jonah Rockoff linked the school records of more than a million children in a large city to their tax records as adults.

Students assigned to teachers who raised test scores more went to university more often, earned more at 28 years old and were less likely to have a child as a teenager. Replacing a teacher in the bottom 5% with an average one, they estimated, would raise the lifetime earnings of a single class by about $250,000 in today’s money.

The size of that figure is disputed. What nobody disputes is that the scores are unstable. In Florida, a teacher’s measured effect in one year predicted the next year’s with a correlation of between 0.2 and 0.5 in primary schools.

So we know what to look for. The most effective teacher predicts where a student will go wrong, sets work that makes students think, helps them one by one, and gives feedback about the task rather than the person. Now look at how teachers are chosen and checked.

In Spain a state school teacher is chosen by the oposiciones, a public competition set out in a 2007 royal decree. The first part tests the subject, in a practical exercise and a written essay on a topic drawn by lot. The second tests teaching, through a written programme for a year’s course and an oral presentation of one unit to a panel of examiners. No children are in the room. The winner then teaches a probationary year under a tutor, but by then the choice has been made, and the competition that made it counted for two thirds of the score.

The Basque Country adds a language bar on top. Every teaching post in its public schools carries a Basque profile, set by a 1993 decree, and a teacher must certify Basque at B2 or C1, depending on the post, before they can be appointed. There is no equivalent bar for English. In many schools the English classes go to whoever’s English is best, not to whoever is best at teaching it, and nothing in either system asks the question that matters.

After the appointment, often nobody checks. In the OECD’s 2018 survey of teachers, about one in four lower secondary teachers in Spain worked in a school where teachers are never formally appraised, against one in seven across the European Union, and in 2013 it had been more than one in three. Spain has no national rule on appraising teachers, and one in six Spanish secondary teachers told the same survey they had never received feedback on their teaching at all.

The private sector asks a different question and gets no closer. The academy that hires the native speaker is testing for fluency, which students notice and parents pay for, and which the German study found to matter least on its own.

We know what makes a teacher effective, and it is the one quality that no exam tests, no interview reveals and no appraisal records. Teachers build it the slow way, one confused face at a time, and most never find out whether they are getting better at it. What would change if that knowledge were written down is a question worth testing.