“Numbers don’t lie?” No. Numbers don’t even speak.
A child is being treated at a hospital. The parent turns to their friend and says: “The doctor said the test says to replace the kidney.” The friend pushes back: a test can’t tell you to do that. The parent defends their position with “Well, numbers don’t lie,” as if that settles it.
It doesn’t settle it. It doesn’t even address it. Because the objection wasn’t that the numbers were false. It was that numbers don’t say anything in the first place. A number is a measurement, a piece of evidence. It sits there. What it means, what follows from it, what should be done about it, that is always, without exception, a matter of interpretation. And interpretations can be questioned, challenged, and sometimes overturned. The moment you forget that, you have stopped thinking scientifically.
This is not a minor technical point. It is arguably the most important idea in all of scientific literacy, and one that most science education never explicitly teaches. Understanding it changes how you read a study, how you listen to a doctor, how you evaluate any claim that arrives dressed in data.
What “Numbers don’t lie” Actually Means, and Where It Goes Wrong
People say “numbers don’t lie” to mean something reasonable: I’m not making this up, there is something real and measurable here that doesn’t bend to my wishes. That’s worth defending. The difference between guessing and measuring is enormous, and measurement matters deeply.
But the phrase claims something more than that. Measurements themselves are fallible; a miscalibrated instrument, a contaminated sample, a procedural error can all produce a wrong number. But even a perfectly accurate number doesn’t tell you what to conclude from it. Numbers don’t make that leap. Interpretation does. And that is where the phrase goes wrong.
Consider: a patient’s kidney function is measured at a certain level. That number is real, produced by a real test on a real biological sample, nobody invented it. And yet what does it mean? Does it mean the kidney should be replaced? Monitored? Treated with medication first? Does it mean the same thing in a 35-year-old as in a 75-year-old? Does it mean the same thing alongside these other test results as it would alongside different ones?
None of those questions are answered by the number. They are answered by a physician, drawing on years of training, a body of medical literature, knowledge of this particular patient, other test results on the same patient, and clinical judgments that reasonable doctors sometimes disagree about.
Why Evidence Always Requires A Theory To Mean Anything
Here is a sharper way to see the same point. Run the same test on two patients and get identical results. Do they both need the same treatment? Not necessarily, because the number means something different depending on what else is true about each patient. Run two different tests on the same patient and get conflicting results. Which one tells you what to do? Neither, by itself. A physician has to interpret them together, inside a framework of understanding about how bodies work.
This is what it means to say that evidence requires interpretation. Evidence does not arrive with instructions. You need a theory, a mental model of the world, to make sense of it. And here is the crucial implication: if the theory is wrong, the interpretation will be wrong too, even if every number is perfectly accurate.
This is an idea the philosopher Karl Popper developed with great precision. The short version: you cannot get from observations alone to conclusions. There is always a layer of theoretical thinking in between, assumptions about what you’re measuring, what it means, how it connects to everything else. That layer is invisible when things are going well. It becomes visible when something surprises you, or when two experts look at the same data and reach different conclusions.
So the next time someone tells you the numbers say something, or that a test result dictates a course of action, pause. Not to be difficult, and not to deny that the numbers are real. But to ask the question that scientific thinking always requires: who is interpreting this evidence, with what framework, and could that interpretation be wrong?
It could be. Every interpretation can be. That is not a weakness of science. It is the source of its strength.
→ If you’re interested in developing this kind of reasoning, see the Private Sessions page.
→ Other Articles