One Halloween I went out in a wig with a name tag that read “Cora Lashun” and my t-shirt had the word “Causation” on it with a red circle around it and red line through it.
Correlation is not Causation.
Very few people got the joke. And ok, it wasn’t Halloween it was last Saturday for drag brunch. Stop judging me.
Anyway, everyone repeat after me: CORRELATION IS NOT CAUSATION!
This pithy little edict is one to live by. And I know this contradicts my previous post about not living by pithy little edicts but this one is different. Stop judging me.
What happens when people assume causation from correlation is they tend to incorrectly reverse engineer a statistic associated with a therapy then blame, or avoid, that therapy. This is the kind of bad science that makes me ugly cry until my mascara runs. Stop judging me.
An example of this kind I came across recently was with a severe refractory asthma attack that came into the ER. A fellow RT said casually “Studies show that mortality rates are really high with ventilated asthma patients. We should avoid intubating.”
The first half of this statement is true.
“mortality rates in patients receiving intubation from 10% to 20% in this patient population” – https://www.atsjournals.org/doi/full/10.1513/pats.P09ST4
The second half is where the problem comes in. The “we should avoid intubating” part. This person took a dicey leap from “mortality rates are high in ventilated patients” to “ventilating patients causes a high mortality.”
You know what else has a really high mortality rate? Avoiding therapy because of feared statistical outcome, like not ventilating patients who can’t breathe. And yes, I just indirectly claimed that mechanical ventilation is therapy. Stop judging me.
This is known as a ‘post hoc ergo propter hoc’ fallacy often simplified to just ‘post hoc’ fallacy. This fallacy comes in the form of A occurred then B occurred therefore A caused B. Then to add insult to injury, it was combined with the false logical inverse that assumes avoiding A will prevent B. This makes Cora Lashun very, very angry. Hoc iudicate me. That means “stop judging me” in Latin.
These kinds of fallacies are easy traps to fall in. There is a correlation in ARDS patients that shows the higher the FiO2 requirement, the higher the mortality rate. But you aren’t going to save the patient by willy-nilly reducing their oxygen. Their O2 needs must be met regardless of the statistical association with death. Lurking somewhere between the post and the hoc that makes one assume A then B therefore A caused B, is that devious little letter C. C stands for Confounding Variable which influences the dependent and independent variables (A and B) giving rise to spurious associations between the two and the incorrect elimination of the null hypothesis and then science fails us and suddenly bigfoot is real because I’m a Libra and I don’t want to live on this planet anymore.
In this case C is the disease state. The worse the disease state (C), the higher the oxygen requirements (A). The worse the disease state (C), the higher the mortality (B). So yes, the higher the oxygen requirements (A), the higher the mortality (B). Correlation. Not Causation.
Perhaps instead of hypocritically contradicting my previous diatribe against axiomatic absolutes I should soften the blow of this maxim and say “Correlation does not imply Causation.” Because correlation can also be causation. If we had started with “The worse the disease state the higher the mortality rate” we’re left with A then B and – also, not therefore – A caused B.
The crux of the fallacy lies in the ‘therefore’. So when faced with statistics and you find yourself thinking “therefore”, think of me in my wig and my Cora Lashun name tag in my NOT CAUSATION t-shirt, cringe a little, and be cautious. I’ll always be therefore you.
See what I did there. ‘Therefore you’. I can’t help it. Stop judging me.