A important point was made here on how easily wording can shape the way research is understood. If an association is presented as causation, the claim can become progressively stronger as it moves through press releases, news articles and social media. That can affect public understanding, business decisions and even policy, despite the original study not being able to support such a strong conclusion.
For those who are unsure of what the difference is, I have a good example from my statistics class:
There is a strong correlation between ice cream sales and deaths by drowning. Does that mean that ice cream causes drowning? Of course not - the lurking variable here is the season/temperature outside.
It’s one reason why I think people should always flip the correlation throughout an article.
People forget that correlation, like many statistical measures, is symmetric: if A correlates with B, then B correlates with A.
But humans love stories. Facts, not so much.
So “A correlates with B” very quickly starts to sound like “If A, then B.”
The other thing people forget is that aggregate statistics describe aggregates. They do not mechanically determine individual behavior.
For example, “on average, women prefer men who are over six feet tall” does not mean every woman prefers men over six feet tall. It certainly does not mean that a specific woman does.
Averages hide an enormous amount of variation. People forget that it's simply the "balance point" for the data set" - that's it. tt does no mean, in any mathematical sense, common or typical.
There could be a correlation between how much chocolate people eat and their likelihood of being a serial killer, meaning the two variables appear to be associated. However, that would not mean that eating more chocolate causes someone to become a serial killer. The relationship could be coincidental or explained by other factors. Correlation can show that two things move together, but it does not, on its own, establish causation (Rohrer, 2018).
Rohrer, J. M. (2018). Thinking clearly about correlations and causation: Graphical causal models for observational data. Advances in Methods and Practices in Psychological Science, 1(1), 27–42.
This might be correlation and not causation, but I'd have to think the way research is being funded more and more often by interested parties rather than unbiased grants might have something to do with it.
[OP] CandyFangs | 17 hours ago
A important point was made here on how easily wording can shape the way research is understood. If an association is presented as causation, the claim can become progressively stronger as it moves through press releases, news articles and social media. That can affect public understanding, business decisions and even policy, despite the original study not being able to support such a strong conclusion.
Mr-Booty-Inspector | 7 hours ago
For those who are unsure of what the difference is, I have a good example from my statistics class:
There is a strong correlation between ice cream sales and deaths by drowning. Does that mean that ice cream causes drowning? Of course not - the lurking variable here is the season/temperature outside.
lolexecs | 10 hours ago
Ha, word placement is key.
It’s one reason why I think people should always flip the correlation throughout an article.
People forget that correlation, like many statistical measures, is symmetric: if A correlates with B, then B correlates with A.
But humans love stories. Facts, not so much.
So “A correlates with B” very quickly starts to sound like “If A, then B.”
The other thing people forget is that aggregate statistics describe aggregates. They do not mechanically determine individual behavior.
For example, “on average, women prefer men who are over six feet tall” does not mean every woman prefers men over six feet tall. It certainly does not mean that a specific woman does.
Averages hide an enormous amount of variation. People forget that it's simply the "balance point" for the data set" - that's it. tt does no mean, in any mathematical sense, common or typical.
feetnotes | 11 hours ago
I keep hearing the words "correlation" and "causation" mentioned together. I think this close association must mean they're related.
[OP] CandyFangs | 9 hours ago
There could be a correlation between how much chocolate people eat and their likelihood of being a serial killer, meaning the two variables appear to be associated. However, that would not mean that eating more chocolate causes someone to become a serial killer. The relationship could be coincidental or explained by other factors. Correlation can show that two things move together, but it does not, on its own, establish causation (Rohrer, 2018).
Rohrer, J. M. (2018). Thinking clearly about correlations and causation: Graphical causal models for observational data. Advances in Methods and Practices in Psychological Science, 1(1), 27–42.
Brrdock | 14 hours ago
That's concerning. I thought the wording in papers had been fine.
Wording in articles often, too, but the framing sensationalized.
Mostly people just don't seem to understand what a link or association is or what science or a paper is able to logically conclude and what not
Khatib | 11 hours ago
This might be correlation and not causation, but I'd have to think the way research is being funded more and more often by interested parties rather than unbiased grants might have something to do with it.
DHFranklin | 11 hours ago
If anything this is finally a really good usecase for LLMs. The same computers we're using to do the mapping for mRNA and Protienfolding.
We can have a neutral arbiter count weasel words and things. It would be really useful software.