Competition, Learning Styles and Test Anxiety

Part two in this series on why competition is good

In Part 1 in this series I discussed why competition is good and how it provides motivation, curation and feedback. I also responded to the most prominent critic of competition. This essay takes on the more challenging critique: “Competition might be good for some kids, but not all kids will excel in competitive environments, so we need to offer competition to some kids, but develop alternative plans for the kids that should not be doing competitions”

---

First, a confession: I’m very confident that well-designed competition is good for most kids (and adults). I’m less confident it’s good for all kids—there will always be outliers. But I now believe competition is good for more kids than most people assume, and that, on the margin, we should have more—not less—competition in children’s education.

I don’t have an airtight proof. What I do have are two things I learned this past year that shifted my thinking: the collapse of the learning-styles story, and the modern view of test anxiety. Together I believe they show that (1) preferences are not correlated to effectiveness, and (2) performing poorly in a competitive environment has more to do with capabilities than personality traits.

The Myth of Learning Styles

Learning Styles are a dangerous meme.

In 1984, David Kolb published Experiential Learning and described learning as a cycle:

• Concrete experience

• Reflective observation

• Abstract conceptualization

• Active experimentation

Kolb argued that people may prefer different parts of this cycle. He did not show the two claims that later got bolted onto his model:

1. that preferences are stable traits, or

2. that matching instruction to those preferences improves learning

Around the same time, Howard Gardner popularized Multiple Intelligences. Gardner’s thesis was that intelligence is plural—linguistic, logical, musical, etc. Gardner’s theories became the intellectual foundation for much of the learning style processes to come, but while Gardner believed in individual differences, he repeatedly warned that MI does not imply “teach each child only in their preferred modality.”

Those caveats didn’t stop educators from embracing a seductive idea: personalized teaching matched to student preferences.

Beliefs in the importance of learning styles jumped to a new level in the early 1990s when Neil Fleming introduced VARK—four sensory “styles”:

• Visual

• Auditory

• Reading/Writing

• Kinesthetic

VARK spread because it was simple, fast, and flattering. Fleming offered a short questionnaire. Students got a label. Teachers got a language of personalization. Students often enjoyed it—like people enjoy hearing that their horoscope “really nails them.”

But there was still no causal evidence that matching instruction to learning styles improved learning outcomes. What proponents largely had was:

• self-reports (students have preferences)

• anecdotes (“they seem engaged”)

• correlations (certain professional fields cluster with certain preferences)

No one asked the real question: Does matching teaching techniques to preferences cause better learning?

A learning-style industrial complex took shape because the story fit the system:

• it gave teachers a framework for personalization

• it explained failure without blaming anyone

• it felt progressive and egalitarian

• it didn’t threaten existing practice

• and it supported a profitable PD-and-publishing ecosystem

Then researchers finally went looking for the promised effect. They didn’t find it.

In the early 2000s, critical reviews started piling up. In 2004, Coffield and colleagues reviewed dozens of learning-style models and found weak measurement and little evidence of instructional benefit.

The landmark synthesis came in 2008/2009 with Harold Pashler and colleagues. Pashler’s move was simple: define what evidence would actually count. A valid test would:

1. classify learners into style groups

2. randomly assign different instructional modes

3. show a crossover effect (visual learners learn best visually; auditory learners learn best auditorily; etc.)

When Pashler and team looked for studies that met that standard, they found almost none—and those that did showed no benefit from matching. Pashler’s conclusion was blunt: there was no adequate evidence base to justify using learning-styles assessments in education.

By the mid-2010s, the message from learning science was consistent:

• preferences exist

• preferences do not imply improved learning

• modality should be chosen by the content, not the learner label

• forcing the “preferred” modality can reduce effectiveness

And yet belief persists.

Recent surveys of teachers find that 85-95% teachers still believe that teaching students to their preferred learning style results in more effective learning. And what’s worse: “There was no evidence qualified teachers were less likely to believe in it than pre-service teachers — suggesting this is not a belief that self-corrects with practical classroom experience.”

Some kids say they prefer to learn kinetically or visually, and those preferences are real. But that doesn’t mean that they actually learn better using their preferred style — in fact if they choose to do what they “prefer” instead of they most effective method for that particular topic, they will learn LESS.

Preferences are real. But the lesson is: preferences are not evidence of effectiveness.

Which brings me back to competition. Even if a child says, “I don’t like competition,” and even if adults have anecdotes about competition going badly, that does not mean competition can’t improve learning—especially when competitions are well-designed, tiered, and humane.

But what about “competition anxiety”? Could anxiety itself reduce learning?

That question has a literature too.

Test Anxiety: From Trait to Signal

Whenever a test or performance “matters,” your body produces arousal (more often referred to as “adrenaline” or “stress”). Sometimes that helps (running from a threat). Sometimes it hurts (explaining your strategy to a room full of executives).

The core finding is old and intuitive: there’s an optimal arousal range. Too little and you’re flat. Too much and you choke. That idea goes back to the classic “inverted U” model first quantified by Yerkes-Didson in a study back in 1908. He found that moderate arousal can improve performance; excessive arousal can shut you down—especially on complex tasks.

The key detail that often gets missed is this:

• Arousal helps when the skill is automatic.

• Arousal hurts when the skill is fragile.

That’s why elite performers can thrive in front of crowds while novices fall apart.¹ It’s also why even elites can choke when the stakes feel overwhelming.

Side Story: In his book, Bounce, Matthew Syed, the table tennis Olympic gold medalist, recounts how he once choked in competition. He explained how it was like all his training “disappeared” and it was like he was an amateur again having to consciously think through each of his actions on the table — which is impossible at the speeds required in table tennis.

He shared how he found a way to reduce that anxiety was to practice like what you were doing is the most important thing in the world, but when you compete, compete like it it makes no difference at all.

To be clear: choking is a problem for EVERYONE. What is different between individuals is the amount of stimulation that is necessary to trigger the “choke”, which is a function of (1) Skill at the task, (2) Difficulty of the task, (3) Perceived importance of the task, (4) Perceived skill gap in completing the task, (4) Skill in techniques for handling arousal

Historically, research treated “test anxiety” as something you had—a trait. In the 1950s and after, researchers built test-anxiety questionnaires and linked them to broader personality traits like neuroticism.

Later, in the 1970s and 80s, models became more cognitive: anxiety hurts performance by consuming limited working memory (“worry thoughts” steal mental bandwidth).

In the 1990s and early 2000s, the focus swung toward situational cues—especially stereotype threat. Studies showed that certain cues (making identity salient, framing a test as diagnostic) could depress performance for some groups. Test anxiety was framed as “socially induced and fragile”. This was widely influential.

But in the 2010s, replication efforts and meta-analyses complicated the story. Effects were often smaller, inconsistent, and less reliable outside lab conditions. Cue-based effects weren’t “nothing,” but they were not the master key many people hoped for.

So what’s the modern consensus that matters for parents and teachers?

The current picture of test anxiety

From about 2010 onward, evidence increasingly supports:

• anxiety is domain-specific (math anxiety ≠ writing anxiety ≠ reading anxiety)

• students can be anxious in one area and calm in another

• what best predicts anxiety is often knowledge gaps and lack of fluency

In plain English: many students aren’t anxious because they’re “anxious people.” They’re anxious because they don’t know the material well enough, and they know they don’t know it. That’s not pathology; it’s information.

This links back to the idea of memory slots. Anxiety is now modeled as insufficient automaticity. You only have s many working memory slots. If you know how to do single-digit multiplication, but you have not memorized your 12x12 times tables, then whenever you need to do simple multiplication, you need to use a memory slot to do it. This makes understanding and working with more advanced concepts more difficult — because you have less working memory to work with. This creates anxiety. You know you should be able to complete the task, but you are working too slowly to do well on the test and you are getting anxious because it is difficult to hold the new ideas from the test in your brain while also managing the slots needed for basic math.

For most students, the best way to reduce math anxiety, is not to use meditation or other psychological techniques — it is just to increase math fluency. Learn your times tables! The strongest test-anxiety reducing technique is not therapy, it is improved competence and familiarity with the content.

Test anxiety is “awareness of low competence”. The strongest test-anxiety intervention is competence + familiarity, not therapy.

Avoiding tests doesn’t fix anxiety. It often just hides the underlying skill deficit the test reveals.

What this suggests about competition

Learning-styles myths show how often educators mistake preferences for constraints.

Test-anxiety myths show how often educators mistake signals for pathologies.

That doesn’t prove competition benefits every child. But it raises a serious possibility:

“Competition aversion” can be a case of mistaking discomfort for harm—especially when the child is under-skilled, expects to lose, and has learned to treat “I don’t like this” as evidence it can’t help them.

I don’t have the full answer yet. But I now suspect far more kids could benefit from competition than is currently dreamed of in our philosophy—if we designed competitions that are tiered, feedback-rich, and humane.

Next essay: Now that I have all the believers in competitive athletics on board, let me find a way to alienate them as well. In the next essay I will explain why I believe the benefits of academic competition can far exceed the benefits of sports competition.

Keep learning,

Edward (and Everest)

1. Another story: Stand up comics will tell you that there are cases where a talented newcomer will kill their first set (killing is good). But then when they try and duplicate that some performance with the same script in their sophomore outing it falls flat. Too much adrenalin and you are frantic. Too little and you aren’t fully present in the moment.