The topic of junk science comes up pretty often here. When you look at everything that passes for news these days, it seems that we appear to be in a Golden Age of Junk Science. There is far more junk science than at any point in my lifetime - or maybe I just notice it more, but don’t think about that.
A classic example that has a good thread going is that Harvard study that proclaimed that eating red meat just twice a week causes diabetes. To be completely neutral, that’s not what the headline says. It says it “may” cause diabetes. Yes, but junk science always uses those “wiggle words” like could, might, may, should and so on. Real science would give numbers like percentage chances, odds risks, things like that. Saying those wiggle words means they never can be held accountable for misleading the world; they can always say, “we never said it would cause diabetes, we said it may .”
I won’t get into why it’s junk, I think by now everyone has read Zoë Harcombe’s must-read analysis of the study, but the most important parts of her take down of it are that it simply shows correlation (which can’t show causation) and they omit any sort of mechanism that can explain the correlation.
In math, you learn about something called the transitive property. A simple example would be:
If A = B and B = C Then A = C
Consider then that correlation is transitive. I’ll abbreviate A is correlated to B as A corr B which turns that example into
If A corr B and B corr C Then A corr C.
That is the key to fame and fortune as a scientist (“fortune” as long as someone pays you for your study). It’s completely obvious when you think about it, so I’m surprised no one else has published this.
An easy example is climate change. Let’s say that global temperatures are increasing, and ignore the ton of questions that raises, because that would require proving causation. Transitivity means that anything else that can be found to be increasing in the time period in which temperatures are rising will be directly correlated to climate change. Conversely, anything that was found to be decreasing in that time period is inversely correlated to that temperature change. Lets say you find a decreasing number of species of some bird or flower or other example. You could then say climate change was endangering species. Never mind. That’s been done.
I’m not picking on or commenting on climate change, just using it as an example of an argument made all the time. Everything that’s changing in the world is blamed on climate change because of this transitivity.
In that Harvard study, instead of saying eating red meat causes diabetes, you can just as accurately say climate change causes diabetes. The two things have increased in the world in the same time period. Climate change causes microplastics in the Pacific ocean. Without doing the research, I bet if you went back to the 1950s, let alone the late 1800s temperature reference period they use, you wouldn’t find the word microplastics anywhere or even the concept. Today, it’s hard to go a week without seeing a microplastics story.
Think of it! No more need to waste time compiling fake data; if they’re both increasing, they’re correlated. You can publish more papers! Food Frequency Questionnaires? Fuggedaboutit. Just ask ChatGPT to fill it out or make up the data completely. You start out knowing the conclusion because anything that’s increasing is correlated to climate change by the transitive property.
You know people have been getting bigger and more obese in America. It’s going up, it may not go up exactly at the same slope as temperature but it correlates with global temperatures so just say that climate change is causing people to get bigger and more obese. If you find data that says vegetable consumption has gone up, you can conclude climate change caused it. By the way; have you ever noticed “fruits and vegetables” has become one word, fruitsanvegetables? You could conclude fruitsanvegetables consumption caused Americans to get bigger and fatter, but that’ll get you canceled, just like everyone advocating the carnivore diet.
Let’s face it: correlations are easy to find.