One comment after a story about a new study on innovation in American films over time reminds journalists that scientists do not “prove” things in studies.
The front page title is “Scientist Proves…”
I’m willing to bet the scientist said no such thing. Rather it was probably more along the lines of “the data gives an indication that…”
Terms in science have pretty specific meanings that differ from our day-to-day usage. “Prove” and “theory, among others, are such terms. Indeed, science tends to avoid “prove” or “proof.” To quote another article “Proof, then, is solely the realm of logic and mathematics (and whiskey).”
To go further, using the language of proof/prove tends to relay a particular meaning to the public: the scientist has shown without a doubt and that in 100% of cases that a causal relationship exists. This is not how science, natural or social, works. We tend to say outcomes are more or less likely. There can also be relationships that are not causal – correlation without causation is a common example. Similarly, a relationship can still be true even if it doesn’t apply to all or even most cases. When teaching statistics and research methods, I try to remind my students of this. Early on, I suggest we are into “proving” things but rather looking for relationships between things using methods, quantitative or qualitative, that still have some measure of error built-in. If we can’t have 100% proof, that doesn’t mean science is dead – it just means that done correctly, we can be more confident about our observations.