Extending the ‘testing effect’ to self-regulated learning
RememberMore (RM) is in development because or two things. One the powerful impact of retrieval and need to optimise the spacing the account for numerous learning cognitive biases that impede learning and retention. For just over a year now, we have been focusing on two tenets, optimising the direct benefits of Successive Relearning (retrieval practice and optimised spacing) and second, harnessing the metacognitive and self-regulation indirect benefits of the testing effect. In fact, it is our position that RM utilises both the direct and indirect benefits by reduces or mitigates the inherent inefficiencies Self-Regulated Learning (SRL) or cognitive biases that impede learning and retention.
Fernandez & Jamet (2017) denote both direct and indirect effects also, with their interest clearly in understanding the indirect effects of testing. That makes three of us! Like these researchers, I am also particularly interested in the effect of practice testing and the “improvement in monitoring accuracy” that Fernandez & Jamet, (2017) suggest is gained from corrective feedback. SRL processes are often put forward to at least part explain the relationship between monitoring accuracy and learning performance, however the details of how this happens remains unclear. Fernandez & Jamet, (2017) set out to shed light on matter.
To be clear, Fernandez & Jamet (2017) define SRL as an active learning process whereby students select, organise and integrate new information by consciously monitoring and controlling their cognition, motivation, behaviour and environment, in order to meet a pre-established goal (Winne and Hadwin 1998, 2008, 2013). And as any teacher will tell you, that process is fraught with distraction and error but get it right, and it can amplify the learning gains available. I would say that, we designed RememberMore to optimise that process.
Fernandez & Jamet (2017) start by setting out the four stages of Winne and Hadwin’s SRL model (1998, 2008, 2013).
- defining the task; (educators define the knowledge/ task / RM deck).
- fixing goals and plans; (the aim is to learn the deck – however we have metrics and corrective feedback built in).
- enacting tactics and strategies; and (trust RM, invest the time).
- adapting the learning activity. (RM is adaptive – mitigating various misconceptions and applying research-informed learning strategies.).
Fernandez & Jamet (2017) – Results
Participants were randomly assigned to either the practice testing group or a control group. The testing group underwent multiple practice tests, whereas the control group only underwent the multiple practice tests after the course. The practice testing group outperformed the control group. To be expected. We know the testing effect works. See Yang et al., (2021) or Argwal., (2021).
Monitoring accuracy is assessed from monitoring calibration (the mean magnitude of over or under-confidence based on Judgement of Learning). Again, as expected the control group were over-confident. Conversely, participants in the practice testing group seemed to be slightly under confident. However, this did not impact study time allocation.
These two key insights are well reported and adjusted for in RememberMore. What was interesting to learn, was four microlevel processes significantly mediated the relation between practice testing and learning outcome. Two monitoring processes (JOL, monitoring use of strategies) and two high-level strategies (summarisation and knowledge elaboration). Estimating their learning performances more accurately and exhibiting better learning performances. This study clearly identifies the impact of practice testing on SRL processes. These results are substantial.
Interesting in that RememberMore uses both JOLs and elaboration and that we have for some time now be discussing with Cognitive Scientists that these indirect benefits play a key role in the field – our classrooms.
RM – draws heavily of learner SRL / metacognitive practice, learners are informed of the challenge level, learners see the knowledge pairing every card – that is itself corrective? Possibly. Opportunities to elaborate are available. Learner report their confidence after every card (JOL), receive session feedback and learning metrics. There is a splash carefully placed reward too. Yes, I’d say the SRL is very much a part of the protocol that drives RM adoption.
Fernandez, Jonathan & Jamet, Éric. (2017). Extending the Testing Effect to Self-Regulated Learning. Metacognition and Learning. 12. 131-156. 10.1007/s11409-016-9163-9.