Making the juice worth the squeeze: Retrieval Practice (again)

In this new series of blogs I aim to look at common aspects of classroom practice and describe the different ways they are done. I then want to suggest why some methods are better than others and make a case for my preferred method.

This blog is not about why you must do retrieval practice. For information on that go here.

I know that you already know that retrieval practice (RP) is good. After all, as teachers, we all know it and do it all the time. 

But do we do it in an optimal way? 

I feel like there are a lot more ways to mess up retrieval practice than I would have imagined. I’ve only realised they exist as I’ve started working with more and more teachers from different contexts. It’s not that these things won’t help the students, it is that they won’t help the students as much in the same amount of time. 

Here is my quick list of ways I have seen people make RP less effective.

Distracting formats.

Memory is the residue of thought. If your RP is a clock, a cloud or a watering can you are reducing the students time thinking of the content. Less time thinking, less learning gained.

Student choice.

Having a grid of points-based questions to choose from is a bad idea. Firstly, it means you have given students a varying volume of work to do as some are answering 6 and others are answering 2. Secondly, it allows students to avoid questions they do not know so their retrieval strength of those areas gets even weaker. Thirdly, it provides disparate data on student performance so doesn’t inform future planning.

Teachers choosing questions.

This is a really common problem and I apologise if I cause offense here, but you are a bad decision maker. Don’t worry we all are. If you are allowed to pick the questions you want the students to retrieve on a given day, then you will do a bad job of it. This is because humans are cognitively flawed. For example, your unconscious bias will change the difficulty of the questions chosen according to your opinion of the class you are teaching. The availability heuristic will make you write questions based on the other lessons you have taught that day or your personal preferences. This leads to adhoc coverage of the questions and therefore the curriculum resulting in areas being neglected and forgotten by the students.

Bulk retrieval instead of spaced retrieval.

I have seen teachers plan retrieval practice in blocks. So when teaching electrolysis we will do RP on bonding, then when teaching rate of reaction we will do RP on electrolysis. This is not effective because while the frequency of RP will be high in the short term, leading high levels of fluency, the time between blocks will be too long, leading to a potential reduction in retrieval strength. This means the time spent is less effective than using a more distributed model that brings up any prior content at any time. 

Getting students to answer in full sentences.

If I want to avoid thinking then I will just copy. Copying is the pinnacle of proxies for learning. All the effort, none of the learning. It’s the Coke Zero of education. One of the most awkward feelings for a teacher is looking around the room and seeing a student with a blank page. We think “If i can just get them going they won’t be intimidated by the barren desert that is their book and will spontaneously write the correct answer” This might have a place in certain parts of a lesson but RP is not it. It needs to be quick and so a student who wants to avoid the effortful work of retrieval will calmly opt out by copying down the sentences and waiting out the clock. They find that peaceful nirvana of not doing any work but not being in trouble. It is a hard thing to accept sometimes but RP is not about what is in the book. The written answers are just there as a way of making the students’ thinking visible. If a student wrote nothing but recalled all the answers they would still be benefitting from RP. It is the secret world inside their head that we are trying to cultivate so just getting them to write the answer will suffice, even if it looks like they haven’t done much ‘work’.

Letting students look up answers from the get go.

Yes it shows them the value of their book or knowledge organiser but it undermines retrieval for the same reasons as above. Maybe allow it after half of the time limit has passed?

Using retrieval practice as the check of prior knowledge.

We all know it is important to check and activate prior knowledge before delivering new content. My point is that both RP and prior knowledge are so important that you dilute both of them by trying to do them at the same time. RP needs its time and its breadth to ensure curriculum coverage and correct spacing of retrieval. Checking prior knowledge needs to be flexible in its format and length. You might need 12 questions, you might only need 2. You might want to use miniwhiteboards, you might want MCQs. It doesn’t matter. The point is RP needs a regular routine whereas prior knowledge needs to be adaptive to the content about to be introduced. In most scenarios doubling them up reduces both strategies effectiveness, while only saving you a small amount of time. That time would be better spent on getting both right.

What works really well.

These are the things that I think work really well. Are they perfect? Probably not. But they do sit in the sweet spot between effort and impact. They are my best bets.

Randomly picking questions.

Having randomly chosen questions drawn from an agreed selection delivers better question coverage over time. This prevents cherry-picking aspects of the course based on various conscious and unconscious factors. It’s not perfect but its much easier than creating a algorithm to perfectly space each question. Use a tool like retrieval roulette or carousel. What about if it throws up a question you don’t like? Well that brings me on to…

Be fanatical about the questions you use.

Pre-plan the questions based on the core knowledge you want all students to have when planning the SoW. These core questions need to be well crafted and unambiguous. Make sure someone proofreads them so they have clarity. If you later come across a question you think is unfair for the students to answer that tells you one of two things: Either you didn’t teach it well enough or the question is in the wrong place in the curriculum sequence. Of course you don’t have to have the same list for every class. You can have a longer list for higher tier classes for example, or a really paired down list for an entry level class. It’s not about the number of questions, its about the forethought and consideration of what questions you have.

Don’t be afraid to provide scaffolding.

We want all students to be asked questions that make them retrieve the important information from their course. But there is a logistical issue of desirable difficulties. We need our weakest learners to be able to access the same questions as the rest of the class. To do this we might introduce some scaffolding to support students. We might verbally give some options for an answer. For example if the question was “What does the horizontal line on a distance-time graph show?” we might first remind students that horizontal means flat and that the options are “accelerate, decelerate, constant speed or not moving”. We can also give scenarios that are concrete examples. For example if the question was “Describe the effect of increasing the temperature on the rate of reaction” we might say “What reacts faster; magnesium in cold or hot acid?”

Have a routine and stick to it.

Students are creatures of habit and keeping a consistent RP routine is vital to achieving the maximum impact in the shortest time possible. While retrieval occurs at many points throughout the lesson, explicit RP is often the first thing students do. It’s a great, calm way for a lesson to start. Carefully think about your routine. How will you use the clock and ensure students have an ordered and efficient start?

Retrieval practice is vital for our students to build up a solid foundation of knowledge. We owe it to them to ensure we use the most impactful strategies possible. I hope you read this and thought ‘that’s what I do already’ If not I hope I’ve given you some food for thought.

Adam Boxer has a great chapter in his book on it if you wish to read more about how to do RP and how to activate prior knowledge.