Empower CrossFit Group Class members with PVC pipes

Shark Attacks and Ice Cream

June 05, 2026•5 min read

Weekly Wisdom - Episode 45

June 6, 2026

We have a real problem.

It stems from our general inability to understand science. Not just you and me, but policymakers too. We assume that medical professionals, researchers, and scientists must have a solid handle on scientific reasoning, but that isn't necessarily the case. It is entirely possible to build a successful career in those fields without developing a particularly strong understanding of how evidence works.

Dietary fibre recommendations are a good example.

Before anyone reaches for a pitchfork, I'm not claiming fibre is bad for you. Quite the opposite. Fibre-containing foods like fruits, vegetables, legumes, and whole grains belong on just about everyone's plate. Vitamins, minerals, antioxidants, polyphenols, and countless other compounds found in plants offer a pharmacy's worth of potential health benefits, regardless of what role fibre itself may play.

So what's my beef with dietary fibre recommendations?

It's not that fibre is harmful. It's that the evidence used to justify many of the claims is weaker than most people realize.

The Problem with Nutrition Research

The first problem is that much of the fibre research relies on self-reported dietary intake. Researchers ask participants to recall what they ate yesterday, last week, or even last month. Unfortunately, human memory is a notoriously unreliable scientific instrument. Researchers themselves acknowledge substantial measurement error in self-reported dietary data.

Think about your own breakfast yesterday.

How many grams of fibre did you consume?

What about three days ago?

Last week?

I have no idea. Most people don't either.

But let's pretend, for the sake of argument, that every participant reports their fibre intake perfectly. We still run into a second problem.

To establish causation, we need to isolate variables. Randomization is one of the best ways to do that. If we're trying to determine whether fibre improves health outcomes, we need some way to separate fibre from all the other healthy behaviours that tend to travel with it.

Correlation Is Not Causation

Consider two study participants.

Tom is thirty pounds overweight. He drinks six beers a day, smokes, works shift work, exercises rarely, and lives largely on fast food and frozen meals.

Doris doesn't smoke, rarely drinks, exercises five times per week, sleeps on a regular schedule, cooks most of her meals, and walks to the market several times a week.

Without looking in their pantries, who do you think consumes more fibre?

Who do you think will have better health outcomes?

More importantly, if Doris turns out to be healthier, is it because of the fibre?

Or is it because she sleeps better, exercises regularly, doesn't smoke, drinks less, and generally takes better care of herself?

The observational studies can't fully answer that question.

They tell us that people who eat more fibre tend to have better health outcomes. They do not necessarily tell us that fibre caused those outcomes.

This is a difficult concept for many people to grasp.

Correlation is not causation.

The Ice Cream and Shark Attack Problem

The classic example is ice cream and shark attacks.

Did you know that eating ice cream is associated with a higher risk of shark attacks?

It's true.

Not because ice cream causes shark attacks. The two are causally unrelated but highly correlated. Both happen more often in the summer.

Summer is the hidden variable.

The challenge with nutrition science is that there are often dozens of potential "summers" lurking in the data.

Now, to be fair, fibre has stronger evidence behind it than ice cream has for causing shark attacks. We have randomized trials showing benefits for things like bowel regularity, cholesterol levels, and some markers of metabolic health.

But when it comes to many of the sweeping claims about long-term disease prevention and longevity, much of the evidence remains observational and therefore vulnerable to confounding.

What Stronger Evidence Looks Like

To illustrate what stronger evidence looks like, consider the shingles vaccine.

Several observational studies have suggested that shingles vaccination is associated with a lower risk of dementia. The problem, again, is that observational studies cannot fully eliminate confounding. Perhaps people who choose to get vaccinated are simply more health-conscious than those who don't.

Maybe we're looking at another ice cream-and-shark-attack relationship.

So why did I book my shingles vaccine?

Because this story has something the fibre story largely lacks: a natural experiment that approximates randomization.

In Wales, when the shingles vaccine was introduced, eligibility was determined by an arbitrary birthdate cutoff. People born just before the cutoff were ineligible, while those born just after it were eligible. That created two groups of people who were nearly identical in age and demographics but differed substantially in vaccination status. Researchers were then able to compare their health outcomes.

The results were striking.

Individuals who received the vaccine were approximately 20% less likely to receive a dementia diagnosis over the following seven years.

No single study proves anything beyond all doubt, but this type of natural experiment provides far stronger evidence for causation than a typical observational survey.

That ability to isolate variables is what makes the findings so compelling.

The Takeaway

The takeaway is not that you shouldn't eat fibre.

Most foods rich in fibre are unquestionably nutritious and belong in a healthy diet.

The takeaway is that we should be cautious about confusing association with causation. Nutrition science often relies on imperfect observational data, and that limits the certainty of the conclusions we can draw.

The shingles vaccine story is different. It benefits from a rare natural experiment that provides unusually strong evidence of a causal effect. While further research is still needed, the current evidence is compelling enough that I decided to get vaccinated.

And that's the lesson.

Not all evidence is created equal.

Sometimes the difference between good science and bad science is simply whether you've managed to separate the sharks from the ice cream.