Research School Network: Effective Retrieval Practice: What should we consider?

Phil Stock, Director of Greenshaw Research School

It’s fair to say the term retrieval practice is now very much part of the mainstream educational vernacular. Some form of Dunlovsky’s practice testing or Rosenshine’s daily review is happening on a regular basis in classrooms up and down the country. Low stakes quizzing is all the rage!

It’s easy to see why. As resource-creation goes, it’s relatively straightforward to write a few simple questions, which can be easily put onto a sheet or projected onto a screen for students to answer. 

No one thinks retrieval practice is an educational panacea – we know improving outcomes is a bit more complicated than that – yet we do seem to have become more excited about the possibilities of quizzing students than other insights from cognitive science, such as spacing and elaboration.

Recently, Professor Rob Coe wrote a piece gently challenging this growing swell of support for retrieval practice. Coe was asking questions about retrieval practice many were already starting to formulate, such as what kinds of tasks work best? When is it most effective? How much time should we set aside for it in lessons?

At Greenshaw we have been using low stakes quizzing in our teaching for a number of years. Retrieval practice (though we never call it this) happens all the time, though there are two points in the school day when it is more formalised – at the start of every lesson and once a fortnight in assessment hour – when students answer multiple choice questions in each subject.

In this time, we have gained a lot of experience about when and how to use low stakes quizzing to improve student learning. I’ve tried to distil these thoughts below, outlining some of the things worth considering around effective retrieval practice: 

1. Retrieval practice can be part of a whole-school teaching and learning strategy 

I have already referred to the two main ways in which quizzing is formalised in our setting – at the beginning of lessons and in an ​‘assessment hour’. In both cases the main purpose is to return to previous learning in an attempt to consolidate it. In other words, to use testing as a learning event. There are, however, a number of other aspects to our use of regular quizzing which we find desirable.

The ​‘Do Now’ at the beginning of the lesson, for instance, enables students to get settled and on task quickly in a predictable way that establishes routine. It affords the teacher the chance to perform simple admin tasks before circulating to systematically check and respond to targeted questions. Similarly, fortnightly assessment hour provides very useful student, class and cohort diagnostic data. 

2. Retrieval practice may benefit some students more than others

Regardless of the type of questions or how accessible they are, in most classes there will always be some students who complete more of them than others. For some retrieval of prior learning is easier, though in each case there should still be benefits. Some teachers overcome these discrepancies by writing progressively more complex questions or setting additional generative tasks at the end.

Research can help explain why retrieval practice works better for some than others. For instance, a study by Agarwal et al. found that, while all students benefitted from retrieval practice, those with a lower working memory capacity benefitted the most. 

3. Learning progressions help shape the timing and design of effective retrieval practice 

A further paper by Agarwal et al. looked at different forms of retrieval practice and the benefits of each in relation to the demands of the final task. Using the revised Bloom’s Taxonomy, the study concluded that higher order retrieval practice improved performance on higher order final tasks more than factual retrieval, which is probably the approach most commonly used in schools at the moment. Better were mixed conditions, when a factual quiz was followed by a higher order quiz.

Agarwal’s paper thus highlights the importance of clear learning progressions that inform when best to deploy retrieval tasks of varying complexity. If you look at the way the questions on Adam Boxer’s original roulette wheel are structured, there is a logical and progressive sequence. Similarly, Tom Needham’s blogs on retrieval practice in English show how his department distinguish between different types of retrieval task in accordance to where his students are at with their learning.

4. The questions and tasks set for retrieval practice matter

Question quality and task type matter. As Coe himself highlights, simple factual questions are easier to generate than questions that illicit higher order thinking. It often takes a deep subject knowledge to be able to put together sharp questions and tasks that dig into areas of evaluation and analysis. Ben Rogers shows how his Trust have developed multiple choice retrieval questions that promote deeper thought.

As with all promising insights from cognitive science, a lot of thought needs to go into what the application of retrieval practice looks like in the classroom and what it means for a busy classroom teacher who doesn’t necessarily read or engage with research. Retrieval practice seems like something we should be doing, but we need to be clear about when and where it is most effective.