Feedback is one of the few things in education that pretty much every agrees is important and worthwhile. The need for feedback is obvious: if you were expected to learn how to reverse park a car whilst wearing a blindfold you would be very unlikely to learn how to go about this without causing damage either to your car, or to the environment. In order to learn you would need to see where you were going and what happened when you turned the wheel. We get this sort of trial and error feedback all time; we act and then observe the effects of our actions to determine whether an action is worth repeating or adapting.

When we talk about feedback in schools, we’re much more likely to mean specific advice given by a teacher to a student on how to make an improvement in some area of their study. There’s been extensive research into the effects of feedback and the consensus is that giving such feedback has great benefits. For instance, the Education Endowment Foundation place feedback at the top of their list of interventions schools can use to close the achievement gap between more and less advantaged pupils.

The claim is that if teachers give their students feedback those students will make an average additional 8 months worth of progress for every year of instruction.

A little critical thinking casts doubt on the likelihood of such a claim. In order for this assertion to stack up you’d have to compare the condition of ‘giving feedback’ against a condition of ‘not giving feedback’, but, as far as I’m aware, I’ve never met a teacher who refuses to give feedback. I’ve certainly never heard of a school that has decided to stop teachers giving feedback. So how could we ever reliably know the effect of giving feedback?

In fairness to the EEF, if you delve a little deeper they make it clear that not all forms of feedback are the same. This is what they say:

Feedback studies tend to show very high effects on learning. However, it also has a very high range of effects and some studies show that feedback can have negative effects and make things worse. It is therefore important to understand the potential benefits and the possible limitations of the approach.

We probably all think we understand the potential benefits of feedback – students improve their performance – but what might the possible limitations be? One limitation is that giving feedback may not be as straightforward as we commonly suppose. While there’s certainly lots of research evidence on the efficacy of feedback, there have been various studies which show lower impact, which might indicate that making feedback work in the classroom isn’t as straightforward as we might assume.

Of these studies, I think Kluger & DeNisi’s 1996 meta analysis of feedback interventions is particularly interesting. Although it’s now over 20 years old, it represents a gold standard of how to conduct a meta analysis and the only studies included are those which met a very high standard of sample size and properly controlled experiments. Here’s a graph showing their findings:

After controlling for sample size, the weighted mean of this distribution, as reported by the EEF,  is 0.41, which clearly suggests that, on average, feedback has a moderate positive effect on performance. But that ignores the fact that over 38% of the effects were negative. That is to say in more than a third of cases, receiving feedback reduced performance.

This can come as a bit of shock, but as Kluger & DeNisi say in the introduction to their paper

We argue that a considerable body of evidence suggesting that feedback intervention (FI) effects on performance are quite variable has been historically disregarded by most FI researchers. This disregard has led to a widely shared assumption that FIs consistently improve performance.

They go on to make the point that such assumptions are due to the “lack of a FI theory”. They went on to propose such a theory. Their Feedback Intervention Theory (FIT) contained five arguments:

  1. How we behave when given feedback is mediated by comparing what the feedback tells us to the goals or standards we have in our heads
  2. We tend to organise these goals or standards hierarchically – easy, moderate or difficult
  3. Because our working memory capacity is strictly limited, we’re only able to some of the feedback we get and that we will only act on the feedback that we pay attention to
  4. We’re likely to ignore goals or standards we consider to be too easy or too difficult to achieve and focus on goals of moderate difficulty
  5. Being given feedback will change what we pay attention to, and therefore affect our behaviour.

This suggests that much feedback has little effect because we are unable to relate it to what an improvement in our performance would look like. For instance if you had written an essay and were told to ‘structure it better’ unless you knew what a better structured essay was like you would be unable to do much with the feedback. But not only that, we’re likely to ignore, or filter out, any feedback which seems irrelevant. We’re unlikely to devote much attention to how we could achieve something we see as either beneath us or unattainable. Our minds seem wired to seek out feedback that is ‘just right’. But despite this preprogrammed preference to work on improvements which seem attainable, the act of being given feedback diverts our attention and focuses us on something else. If our attention is drawn to something we see as too easy, we’re likely to reduce our effort, and if something we see as unachievable is pointed out we’re likely to either downwardly revise what we think we’re capable of, or give up entirely.

Dylan William summarises their theory of how feedback affects behaviour with this handy table:

All this goes some way to explaining why feedback appears negatively correlated with attainment in the latest PISA results.

The index of perceived feedback (highlighted in yellow) was derived by asking students to respond to the following five statements:

  • The teacher tells me how I am performing in this course
  • The teacher gives me feedback on my strengths in this subject
  • The teacher tells me in which areas I can still improve
  • The teacher tells me how I can improve my performance
  • The teacher advises me on how to reach my learning goals

These things might at first appear to be obviously beneficial, but maybe they’re things which fail to take Kluger & DeNisi’s FIT into account. For instance, being told how you’re performing in a course could easily result in a student thinking, ‘What’s the point?’ If you’re given feedback on your strengths you might conclude you’re better than you really are and decrease how hard you’re trying. Similarly, being told about where you could improve might lead you to decide doing better is overwhelmingly out of reach.

The final two statements should be more useful. We all need to know how we can improve, but this doesn’t mean we’ll either understand or act on such feedback. Finally, if a teacher advises a student on how to reach their goals this is theoretically likely to be effective, but how many teachers actually do this? Don’t we tend to advise students on how to reach our learning goals? If these are goals which students see as too easy or too hard, our efforts could be in vain.

All of this is further complicated by the fact that even if students decide to aim higher or try harder based on the feedback they receive, feedback that seeks to raise current performance may end up degrading learning. In this post I suggested a feedback continuum where the aims of giving feedback are different at different stages of an instructional sequence. So, at the beginning of a course giving lavish feedback to help students establish effective mental representations of what successful performance looks like may be useful. Students can be encouraged to revise their hierarchy of goals and standards upwards by being helped to produce work of a higher standard. But then, in order to prevent students from becoming dependent on such support it probably makes sense to reduce feedback both in terms of frequency and increase it in terms of complexity. If feedback is too easy to follow, then students may see no need to internalise the processes required to improve performance. If feedback is made progressively more difficult to use then students will be prompted to remember the advice so as not to go through the increasingly arduous process of working out how to apply the feedback they’re given.

In my view, the role of feedback is probably crucial in balancing the twin pressures of struggle and success, so it’s worth thinking a bit harder as to exactly what we mean when we talk about feedback.