Principle 3 - Checks for Understanding

The main purpose behind the need to check for understanding is to know the difference between "I taught it" and "They learned it".

All teachers have been in the situation where they ask a question about something they have just taught (be that via an explanation and model or through a discovery based task) and they are greeted by a sea of blank faces from the students. We have taught but the students have not learned.

This situation is actually preferable to the other alternative where the teacher does not ask the question, and the fact that students did not understand what we taught them becomes clear at a later date when checking work. In this case, students have been allowed to develop their own misconceptions, which can be very tricky to remove once they become embedded.

An important distinction to make when considering checking for understanding is the difference between Learning and Performance.

Learning - a relatively permanent change in knowledge and/or behaviour.

Performance - temporary changes in knowledge and/or behaviour that can be measured directly following instruction.

Our goal is to ensure our students learn, but this is very hard to measure, especially in the constraints of a lesson. To check for learning we need to determine whether students can still do something after a period of time. The literature on Desirable Difficulties (discussed in the Challenge page) suggests that by reducing initial performance you can actually increase learning. This links to the idea that students will need to think more about the content if there are desirable difficulties, and hence remember it better.

There is a limit to this idea, and this is where checking for understanding comes in. If students can not show they understand in the moment then they are not going to be able to show they understand at a later date. That is, if they cannot perform then they will not learn. However, the opposite is not true.

Just because a student can perform in the moment, that does not mean they have learned it.

The Dunning-Kruger Effect is a cognitive bias that we all fall victim to. Like all cognitive biases, it is something which is easier to avoid when you know about it, but there is no foolproof solution.

In it's simplest form, the Dunning-Kruger effect states that:

If we know a little bit about something, then we are more likely to over-estimate our actual abilities.

This has lots of implications in teaching and learning, and it explains why it is often those who perform the worst that are the most confident (and vice versa, as the extension of the effect is that with more knoweldge you actually become less confident in your abilities).

The reason that we fall for this cognitive bias is that we do not have any idea of the nuances or the sheer size of a topic when we first learn about it. We just feel like we know a lot more than we did. From a mathematical perspective, knowing 1% of a topic is infinitely more than know 0%, so it feels like a huge jump. But we can't even see the other 99%.

In Teach Like a Champion 2.0, Doug Lemov breaks checking for understanding into three parts, each of which we will explore here.

Culture of Error

In order to check for understanding we have to create a classroom where students are willing to make mistakes and learn from them. This links to Principle 4 so it will not be detailed here, but, in short, that means we must not ridicule students for wrong answers (or let other students ridicule them), and in fact praising having a go. In the clip below, we can see the teacher tells the students who made a mistake to be proud of them working it out, making it clear to all students that being wrong is neither a negative nor a permanent state.

As Drucker says "Culture eats Strategy for breakfast", and this can be applied to the classroom and wider school setting. Spending time deliberately developing the culture we want will pay off in the long run.

Collecting Data

This is the stage where we actually check whether students have understood what we have taught them. There are lots of ways that we can collect this data, and we explore this in further depth below. The basic principle is that we have to find ways to determine whether students have understood the concepts we have taught and can perform in the moment (as this is a prerequisite to learning).

Acting on the Data

Once we have collected some data on whether students understand or not, we then actually have to act on it. This is often the stage that gets forgotten, but without it, the collection of data was largely a waste of time. How we act on the data collected will depend on what data we get.

If all the class show they understand, then moving on to the next activity or concept is a sensible option.

If only a few (1 to 5 students) do not understand, it probably still makes sense to move on, but take a note of those who did not understand, and check in with them individually.

If a significant number of students show they don't understand, then some whole class exploration of errors could be appropriate. What are the different errors? Why were they made? How could we avoid making them again in the future? In this discussion, even those who showed an unnderstanding will benefit from thinking more deeply about the ideas.

If none of the class (or very few) show understanding, then reteaching the concept, possibly through further examples and modelling, is probably the best way forward. If there are a couple who understood, they could be pushed on to the next activity, or used to help explain to the rest of the class.

Marking is an activity that happens outside of class. We collect homework, quizzes, exit tickets or any number of other assignments, check it and usually provide feedback (Principle 11) to the students for them to improve. As marking happens outside of class, it is not a way to immediately check for understanding and provide immediate action to correct any misconceptions. It's purpose is to inform the planning of future lessons. Due to this, the action we take can be more considered and planned.

However, this also means that students have been given an opportunity to embed any misconceptions, so it should not be solely relied upon. Something that could have been fixed in a couple of minutes using the other methods of collecting data could take a whole lesson to fix after marking a piece of work. The other methods of collecting data discussed below can help address these issues more quickly.

It is still important to do some marking, though, especially since this is often the only way to adequately address issues that come with extended pieces or work (e.g. essay structure, layout of working, exam technique). But we want to try to streamlime this as marking is a very time consuming process, and does not always lead to improvement. Here are some suggestions.

Exit Tickets

Exit tickets are a question for students to answer based on what you taught that lesson. The idea is that students all answer the question at the end of class and you collect them as they leave the room. Having students answer on post it notes, or pieces of A5 paper, allows you to easily look through them. After the lesson you can then sort them into piles:

• excellent answers;
• answers that show some misconceptions;
• answers that show no understanding at all.

Judging by the size of each pile you can make a quick decision about what follow up is needed the following lesson. This process takes about 10-15 minutes, and is very flexible as you can change the question depending on how far through the lesson plan you get.

Teachers ask a lot of questions. One estimate suggests 400 a day (Wragg & Brown, 2001). There are two purposes to questioning: Checking for Understanding and Developing Understanding (Principle 9). Depending on the purpose of our questions we will ask different types of questions. Here we focus on the former, which tend to be more closed questions.

A Simple Switch

A very common question asked by teachers is "Are there any questions?" or "Everyone got it?". These flow from our lips without much thought. But, for all intents and purposes, these questions are functionally rhetorical. Rarely do we allow enough time for students to answer the question. Even if we do, it takes a brave student to ask a question in this scenario as they will be opening themselves up to their peers thinking they are stupid for not understanding, even if the whole class does not understand. And it is also quite probable that they cannot even identify that they do not understand, as they just don't know enough yet to be capable of making this determination (they are falling prey to the Dunning-Kruger effect, and thinking they can do it).

We as teachers are also subject to the Dunning-Kruger effect, and this is a good example. When we ask one of these questions, we trick ourself into thinking the class can do it. We get a little bit of information (the fact that nobody wants to ask a question), and believe (usually incorrectly), that this means there are no problems. A little bit of knowledge makes us overconfident in our (usually subconscious) conclusions.

A simple switch is to change every time we ask one of these questions with "Ask me two questions". This subtle change makes it clear that we expect them to have questions, and that we don't expect them to understand everything straight away (helping to develop a Culture for Error). It legitimizes the fact that students have questions, and even rewards those who ask questions. This also gives you valuable information, as the questions students ask can actually highlight some misconceptions that it is hard to get at through a closed check for understanding question.

Another switch is to get students to explain it back to you, perhaps after giving them some time to write down their understanding of the idea, or engageing in a Think-Pair-Share style activity.

Hinge Questions

The final way to collect data is to observe what students are doing in class. By this we do not mean sitting at the desk and watching students, but rather moving about them whilst they work, looking at what they are doing, and picking up on any errors there and then. Doug Lemov calls this Circulate, which he stresses is different from proximity. When circulating, a teacher must first "break the plane" and force themselves to be amongst the student desks, not in the 'teacher's space' at the front. To successfully circulate, the teacher also needs to engage with the students, picking up their work, showing you are reading it, or giving quick checks on their books with your pen. By doing this you are able to pick up on errors and misconceptions of individuals as they work.

Observation often happens when students are practicing independently (Principle 7) or in small groups, and allows the teacher to get a bit more of an insight into an individual's thoughts than a whole class hinge question. Of course, whilst the teacher observes, they may use some questioning with individuals or small groups to further probe their understanding.

The hard part of observing students to collect data on understanding is that there is a lot of things going on in a classroom at any point in time, and it can be difficult to catch errors with so many distractions. Below are a couple of ideas from Teach Like a Champion 2.0 to help combat this.

Standardize the Format

Standardise the format has the teacher create resources or spaces such that they can look in the same place to collect data for all students. One way to do this is to have a booklet into which students write answers, as then all their answers will be in the same place and you know exactly where to look to check. Another example, in a more practical setting, would be to have all groups organise themselves in the same way. With exit tickets, get all students to place their post it in the same place every day. With your Do Now, keep the same format for all tasks. This allows you to focus less on finding the work you need to check, and more on actually checking it.

Due to the differences between learning and performance discussed above, some of the most useful data we can collect is when students make an error. But there are different types of errors that are made, which broadly split into two:

• Mistakes
• Misconceptions

The difference between these is important as our approach to acting on the data will differ depending on which is the case. As Craig Barton put it in his podcast, if you ask a student to check their work, they will spot a mistake, but they will not spot a misconception. We all make mistakes, either due to rushing, not putting enough thought into what we are doing, or being distracted. Mistakes are not a major cause for concern because they do not show a lack of understanding (though they can show a lack of fluency, which is important in key skills). Misconceptions, on the other hand, show that a student has not understood something, and these need to be addressed.

Misconceptions form for a variety of reasons, and it is very likely that, no matter how well we plan our explanations and modelling (Principle 6) some students will still develop a misconception (either because of faulty prior knowledge, or not listening attentively, or they are hungry, etc). The reason we check for understanding is to identify and fix these before they become a bigger issue that cause more misconceptions to develop later on.

But to do this it is helpful to think about the possible misconceptions that students might develop, and what causes them. We will have a good idea of some of the most common ones for our subject just from seeing them happen so many times, but it is also worthwhile to think carefully about other misconceptions that students might develop. This is a great development activity to do within your department: sitting down and discussing possible misconceptions, how to try to stop them from happening in the first place, how to discover them (possibly through designing hinge questions), and what to do if we discover a student has that misconception.

In The Hidden Lives of Learners, Graham Nuthall made some amazing discoveries by spending hundreds of hours analysing video and recordings of students during classtime. One of the most startling was discovered through the use of pre-testing and post-testing students on the content delivered in a particular unit. He found that, on average, a student knew between 40 and 50% of the content that the teacher wanted them to learn before the unit started. The problem was that each student knew a different 40-50%.