We live in a world where nutrition advice is everywhere. One person tells you to avoid seed oils. Another says focus on protein. Someone else insists that ultra-processed foods are the root of all disease. Then there’s fasting, meal timing, food quality, calories, hormones, inflammation. It never stops.
And yet, for all the noise, there’s a deeper question most people never stop to ask.
How do we actually know what’s true?
That’s the question Dr. David Allison has spent his career trying to answer. Not just what works for obesity or nutrition, but whether the evidence used to support those claims actually holds up. And once you start looking at the field through that lens, something becomes very clear.
A lot of what sounds like science… isn’t really science at all.
Why Nutrition Feels Like Common Sense (But Isn’t)
Nutrition is one of the few scientific fields where almost everyone feels like an expert.
Why?
Because everyone eats.
You’ve felt hunger. You’ve gained or lost weight. You’ve experimented with different diets. You’ve seen what “works” for you. And that creates a powerful illusion that personal experience equals understanding.
But here’s the problem.
Personal experience is incredibly limited. It’s shaped by bias, memory, environment, and a thousand invisible variables you’re not tracking. You don’t know your exact energy intake. You don’t know your exact energy expenditure. You don’t know how your sleep, stress, movement, and food environment are interacting on any given day.
So what feels like a clean cause-and-effect relationship is usually anything but. And when you multiply that across millions of people, you don’t get clarity. You get chaos.
This is exactly why nutrition becomes a breeding ground for strong opinions and weak conclusions.
People aren’t just interpreting evidence. They’re interpreting their experience. And then they’re layering meaning on top of it. Sometimes it’s innocent. Sometimes it’s driven by identity, ideology, or incentives. But either way, it creates a landscape where confidence is everywhere and clarity is rare.
A simple example makes this very clear.
Researchers have studied individuals with obesity who strongly believe they are eating very little, often reporting intakes under 1,200 calories per day while struggling to lose weight. Instead of relying on self-reports, researchers used a method called doubly labeled water to measure total daily energy expenditure, which allows you to infer how much energy must be coming in over time.
When they compared reported intake to what physiology said had to be true, the gap was enormous. On average, energy intake was underestimated by about 1,050 calories per day. That’s roughly a 47% underestimation.
In other words, what people thought they were eating and what they were actually eating were not even close. And it gets even more interesting.
When these same individuals were given a controlled test meal and later asked to recall what they ate, they still underestimated intake by about 20% just one day later. That’s not long-term memory distortion, it’s an immediate misperception.
And when researchers compared these individuals to others with obesity who did not report “mysterious” weight loss resistance, another pattern emerged. The group struggling with weight loss was more likely to attribute their condition to genetics or a “broken metabolism,” even though their measured physiology told a different story. Their total energy expenditure, resting metabolic rate, thermic effect of food, and response to exercise were comparable.
Meanwhile, the group that didn’t see themselves as metabolically broken still underreported intake, but by about 20% on average. That’s still meaningful, but far less extreme than the nearly 50% seen in the other group.
So what’s going on here?
This isn’t about people lying. It’s about perception.
Humans are remarkably poor at estimating what they eat. Not occasionally. Systematically. And not randomly, either. The error often moves in a predictable direction.
This is exactly the kind of thing that makes nutrition so confusing. Because if your internal sense of what you’re doing is off by hundreds or even thousands of calories per day, then your conclusions about cause and effect will also be off.
- You might believe your metabolism is broken when it isn’t.
- You might believe a specific diet is uniquely effective when it simply reduces your intake in ways you didn’t notice.
- You might believe something “doesn’t work” when the reality is more complicated.
And once those beliefs take hold, they’re hard to challenge. Not because people are irrational, but because their experience feels real and convincing.
This is where science becomes essential. Not to override personal experience, but to check it. To measure what we can’t see. To quantify what we misperceive. To separate what feels true from what actually is.
Because without that, we’re all just guessing.
Science Isn’t About Trust, It’s About Methods
You’ve probably heard the phrase “trust the science.” It sounds reasonable. But it’s also misleading. Because science, at its core, isn’t something you trust blindly. It’s something you examine.
Real science comes down to three things:
- The data that were collected
- The methods used to collect them
- The logic connecting those data to the conclusion
That’s it. And each of those layers matters more than most people realize.
The data tell you what was observed. The methods tell you whether those observations are reliable. And the logic tells you whether the conclusion actually follows.
If any one of those breaks down, the entire claim can fall apart. So when someone makes a claim about nutrition or health, the real question isn’t whether they sound credible, confident, or even well-intentioned. It’s: how did they arrive at that conclusion?
- What was measured?
- How was it measured?
- What assumptions were made?
- What was controlled for, and what wasn’t?
Two people can both “use science” and arrive at completely different answers depending on how a study was designed, how the data were analyzed, and how the results were interpreted.
That’s not a flaw in science. That’s the reality of it.
It’s also why disagreement exists even among experts. Not because one side is “anti-science,” but because they may be weighing different evidence, questioning different assumptions, or interpreting the same data through different methodological lenses.
And this is where the idea of “trust” starts to break down.
You don’t need to distrust science as a process. In fact, it’s one of the most powerful tools we have for understanding reality. But you do need to be careful about placing trust in individual claims, studies, or communicators without understanding how those conclusions were built.
Because scientific language can be incredibly persuasive. A graph, a citation, a mechanistic explanation, a confident delivery. All of it can create the feeling of certainty.
But none of it guarantees that the reasoning is sound. And once you start looking through that lens, you begin to see something surprising. Many claims that sound scientific are actually resting on very fragile foundations.
How Weak Evidence Turns Into Strong Claims
One of the most important parts of this conversation is how easily research can be misinterpreted or overstated.
Sometimes it’s obvious.
Observational studies—where researchers track what people eat and what happens to them—are often used to suggest cause and effect. But those studies can’t actually prove that one thing caused another, at least not on their own. They can show associations. They can generate hypotheses. But they can’t isolate cause from the many other variables happening at the same time.
That doesn’t mean they’re useless. It means they’re limited.
And those limitations often get lost in translation.
Other times, the issue is more subtle.
Researchers may use statistical methods that don’t match the design of the study. They may analyze grouped data as if individuals were randomized, which can make effects look stronger than they actually are. They may run multiple comparisons and highlight only the significant ones. They may design interventions that appear promising on paper but don’t hold up when examined more rigorously.
None of this necessarily requires bad intent.
In many cases, it’s a combination of incentives, expectations, and human nature. Researchers want to find something meaningful. Journals want publishable results. Readers want clear answers.
So the story gets cleaned up.
- The uncertainty gets smoothed over.
- The effect size gets emphasized.
- The limitations get pushed to the background.
And by the time it reaches the public, what started as a modest, uncertain finding can sound like a definitive conclusion. Once that happens, the claim starts to spread. It gets repeated in articles, podcasts, social media posts, and conversations. Each time, it becomes a little more simplified, a little more confident, a little more detached from the original data.
At that point, it’s no longer about the data. It’s about the narrative.
And this is where Allison introduces a powerful idea: some research doesn’t function as discovery, it functions as idea advertisement.
Not necessarily because researchers are intentionally misleading anyone. But because the goal subtly shifts.
Instead of asking, What is true? The focus becomes, How do we support this idea?
Once that shift happens, everything downstream changes. The way studies are framed. The way results are interpreted. The way findings are communicated.
And over time, those narratives can shape public belief far more than the underlying evidence ever did.
That distinction matters more than most people realize because if you don’t understand how weak evidence turns into strong claims, it’s almost impossible to tell the difference between something that’s well-supported and something that just sounds convincing.
Why Outcomes Can Be Misleading
Here’s another idea that completely changes how you interpret results. Two people can follow the same intervention and get very different outcomes. One loses weight. One doesn’t.
At first glance, that seems straightforward. One person “responded,” the other didn’t. Case closed. But that assumption breaks down when you actually look at the data more closely.
To illustrate this, consider the DIETFITS study out of Stanford, one of the largest and most well-controlled trials we have comparing low-fat and low-carbohydrate diets in the real world.
Researchers recruited over 600 adults with overweight or obesity and assigned them to either a “healthy low-fat” or “healthy low-carb” diet. Importantly, both groups were coached to eat high-quality diets built around whole foods, especially vegetables, while minimizing added sugars and refined flours.
This wasn’t a junk food vs clean food comparison. It was a well-designed test of two popular dietary approaches done in a way that actually resembles how people eat in the real world.
Participants were initially instructed to reduce either fat or carbohydrate intake to very low levels, then gradually reintroduce those macronutrients to a level they felt they could sustain long-term. They were not explicitly told to restrict calories.
After one year, the average result was surprisingly similar between groups. Both lost about 12–13 pounds. There was no meaningful difference between low-fat and low-carb. Measures like insulin secretion and genetic markers didn’t predict who lost more weight. Energy expenditure didn’t differ between groups either.
If you stopped there, the conclusion would be simple: Both diets work equally well.
But when you look at the individual data, the graph tells the real story.
Some people lost a substantial amount of weight. Some lost a little. Some didn’t change much at all. And some actually gained weight.
Same study. Same structure. Same general guidance. Very different outcomes.
This is where most people jump to a conclusion.
“See? People respond differently. One person is a responder, another is not.”
But this is exactly the point Allison is challenging. What if those different outcomes don’t actually reflect fundamentally different biological responses?
What if they reflect different contexts? Think about everything happening outside the study protocol.
- One person sleeps better. Another doesn’t.
- One person is under chronic stress. Another isn’t.
- One person adheres closely to the diet. Another drifts.
- One person changes their activity. Another stays sedentary.
None of that shows up cleanly in the final number on the scale. So what you’re seeing isn’t just biology. You’re seeing biology plus behavior, plus environment, plus adherence, plus randomness. And all of that gets compressed into a single outcome.
This is why Allison draws such an important distinction between outcomes and responses. An outcome is what you observe. A response is what the intervention actually caused. Those are not always the same thing.
Two people could have the same physiological response to a diet or a drug, but their real-world outcomes could look completely different because of everything else happening in their lives.
And once you understand that, it changes how you interpret almost every nutrition claim you hear. It makes you more cautious about labeling something as “working” or “not working” based on surface-level results.
It makes you more skeptical of simple narratives about personalization and “finding what works for you.” And it reinforces a broader point. Human biology doesn’t operate in controlled conditions.
It operates in real life. And that means simple conclusions about what “works” are often far more complicated than they appear.
How to Make Better Decisions in a Noisy World
So where does this leave you?
You’re not going to read every paper. You’re not going to audit statistical methods. And you don’t need to.
But you do need a better way of thinking.
Because without that, it’s easy to get pulled in every direction. One week it’s carbs. The next week it’s seed oils. Then it’s fasting, gut health, food dyes, inflammation, hormones. Every claim sounds confident. Every claim sounds urgent. And if you don’t have a framework for evaluating them, you end up reacting instead of thinking.
The goal isn’t to know everything.
It’s to make better decisions with the information you have.
That starts by recognizing that not all decisions carry the same weight. Choosing between two brands of protein powder is not the same as deciding whether to start a medication, undergo a procedure, or dramatically change your diet. The higher the stakes, the more careful you should be about where your information is coming from and how much effort you put into verifying it.
It also means getting comfortable with uncertainty.
You are not always going to have perfect information. In fact, you rarely will. Even experts operate with incomplete data. The difference is that they understand the limits of what they know. They don’t confuse confidence with certainty, and they don’t treat early or weak evidence as settled fact.
Another useful shift is learning to think in layers.
- A podcast might introduce you to an idea.
- AI might help you gather information.
- An article might summarize the evidence.
- A trusted clinician or expert might help you interpret it in context.
Each layer adds perspective. None of them, on their own, should be treated as the final answer.
And over time, you start to notice patterns.
Certain voices are consistently careful with their claims. They acknowledge uncertainty. They update their views when new evidence emerges. Others tend to overstate, oversimplify, or chase attention. Learning to tell the difference is one of the most valuable skills you can develop.
Because at the end of the day, you don’t need to become a scientist, but you do need to think a little more like one.
Actionable Checklist: How to Think More Clearly About Health Information
- Ask what kind of evidence you’re looking at
Is it observational data, a controlled trial, or just a mechanistic theory? Not all evidence carries the same weight.
- Be cautious with certainty
The more confident and absolute a claim sounds, the more carefully it should be examined.
- Don’t confuse helpful rules with true explanations
Something can help guide behavior without being the actual reason something works.
- Remember that outcomes are noisy
Different results don’t always mean different biological responses.
- Look for consistency across credible sources
If multiple knowledgeable people with different perspectives converge on similar conclusions, that matters.
- Use information in layers
Podcasts, AI, and articles are great starting points. Not final answers.
- Consider the stakes before going deep
Not every decision requires exhaustive research. Save that effort for the decisions that actually matter.
- Accept that some goals involve discomfort
The absence of ease doesn’t mean something is wrong. Sometimes it’s part of the process.
