It’s June, and June is Learning Styles month for the Debunker’s Club. Now, I’ve gone off on Learning Styles before (here, here, here, and here), but it’s been a while, and they refuse to die. They’re like zombies, coming to eat your brain!
Let’s be clear, it’s patently obvious learners differ. They differ in how they work, what they pay attention to, how they like to interact, and more. Surely, it make sense to adapt the learning to their style, so that we’re optimizing their outcome, right?
Er, no. There is no consistent evidence that adapting to learning styles works. Hal Pashler and colleagues, on a study commissioned by Science in the Public Interest (read: a non-partisan, unbiased, truly independent work) found (PDF) that there was no evidence that adapting to learning styles worked. They did a meta-analysis of the research out there, and concluded this with statistical rigor. That is, some studies showed positive effects, and some showed negative, but across the body of studies suitably rigorous to be worth evaluating, there was no evidence that trying to adapt learning to learner characteristics had a definitive impact.
At least part of the problem is that the instruments people use to characterize learning styles are flawed. Surely, if learners differ, we can identify how? Not with psychometric validity (that means tests that stand up to statistical analysis). A commissioned study in the UK (like the one above, independent, etc) led by Coffield evaluated a representative sample of instruments (including the ubiquitous MBTI, Kolb, and more), and found (PDF) only one that met all four standards of psychometric validity. And that one was a simple one of one dimensions.
So, what’s a learning designer to do? Several things: first, design for what is being learned. Use the best learning design to accomplish the goal. Then, if the learner has trouble with that approach, provide help. Second, do use a variety of ways of supporting comprehension. The variety is good, even if the evidence to do so based upon learning style isn’t. (So, for example, 4MAT isn’t bad, it’s just not based upon sound science, and why you’d want to pay to use a heuristic approach when you can do that for free is beyond me.)
Learners do differ, and we want them to succeed. The best way to do that is good learning experience design. We do have evidence that problem-based and emotionally aware learning design helps. We know we need to start with meaningful objectives, create deep practice, ground in good models, and support with rich examples, while addressing motivation, confidence, and anxiety. And using different media maintains attention and increases the likelihood of comprehension. Do good learning design, and please don’t feed the zombie.