Cause and Correlation, or the Pirate Problem



As you can see from the above graph, global warming is pirate-based.  It's something I think we all suspected, but were hesitant to advance until the facts could be summarized in a handy graphic.

There is something about information delivered via graph that instantly lends an air of unassailable authority. The person trapped in the cube next to you, or even the guy down at the gas station couldn’t possibly carry the credibility of a simple graph.

It is an axiom of business that any presenter worth his or her salt is going to fill their PowerPoint with charts and graphs. The more the better, and the more oblique and difficult to read the best. Data delivered with a graph says "Here is the evidence, plain and simple. Let the ascending and descending lines tell you the story."

The problem with the story, as with the graph above, is that we aren't just suckered into believing correlation implies causality. We start thinking correlation is causality. Governments, businesses and individuals make this mistake on a daily basis. It’s impossible to calculate the frequency or the magnitude of the resulting financial loss, but it’s enormous.

We all know the crowing of the rooster doesn’t cause the sun to rise. But when rates of breast or prostate cancer is associated with soymilk, or some new drug, it can frequently drive us towards some definitive action, even though the connection of data points might in actuality be more rooster/sun than cause/effect.

The correlation/causality problem goes back a long way. The early human brain, confronted by the rustling of the bush, might naturally assume it was a tiger and not the wind. Erring on the side of safety could make the difference between life and death. Assuming correlation as causality was a small price to pay. This evolution based brain bias is still part of our biology today.

Here are four questions worth considering the next time you're faced with the seductive whisperings of an X-axis.

1. Where do the represented data points come from?  Groups and individuals might be selectively mining the facts based on their own private agenda.

2. What do the data points represent?  Tiny samples can lead to casual conclusions that would be dismissed in a more robust survey population

3. Was this a blind study? A control group gives you some yardstick by which to judge the rest of the information.

4. Could other factors be in play?  This is probably the most abused problem with the correlation/causality mix-up. Maybe there is some relation between the X axis and the Y axis, but they could just as easily be responding to some other, third influence.

In the case of the rooster/sun problem, you'd want to consider both planetary revolution and the circadian rhythms of diurnal animals. (Additionally, if you've ever been on a farm, you'll know that while roosters do crow at daybreak, those feathery little jerks will also sound their alarm in the middle of the night.)

So the next time some newscaster announces that eating peanut butter has "been linked to" autism, think back to our little graph. And remember: despite the insistence of Pastafarians everywhere (a group inspired by a modern-day Russell's teapot analogy), most meteorologists agree that pirates have next to no effect on the climate.

Fans of buccaneers, privateers, and skallywags can let out a "Yarr!" of relief.

Check out Robb’s new book and more 


Comments

Post a Comment

Popular posts from this blog

The Robert Frost Quandary, or How Irrational Thinking Might Save Your Life

Image
You stumble out of the wilderness, having had no contact with humans for at least ten days. You’re weak from hunger and fatigue and you find yourself at a crossroads, power lines stretching out along each of the separate roadways. It’s decision time. You think about Robert Frost’s poem, and wonder if his advice to take the road less traveled might not just lead to your demise. What do you do, or more importantly, which brain system should you use to make this crucial decision? Daniel Kahneman, famed psychologist, winner of the Nobel prize for prospect theory and author of Thinking Fast and Thinking Slow might be the one guy to call, assuming your existential crossroads gets cellphone reception. Kahneman explains that we have two systems for making decisions. He refers to them simply as System 1 and System 2. System 1 is reflexive, automatic, and impulsive. It takes a constant reading of your surroundings and generates short-term predictions, all operating on a level beneath yo
Read more

Who Are You: Does Personality Change as We Age?

Image
Are you the same “you” that you were twenty years ago? Thirty? Forty? What about fifty years ago? While science still struggles to quantify, measure, and categorize personality, some have a different question: how stable is personality in the first place? Do the high school version of yourself and the retiree version of yourself still have some common immutable core? It depends how you look at it , says a paper by lead author Rodica Damian of University of Houston and her co-authors Marion Spengler of the University of Tuebingen in Germany, Brent W. Roberts of the University of Illinois at Urbana-Champaign, and UH grad student Andreea Sutu. The paper in question goes by the catchy title “Sixteen going on sixty-six: A longitudinal study of personality stability and change across 50 years.” Damian and her colleagues compared the results for 1,795 U.S. adults who had taken a personality test as teenagers and then again, five decades years later. The team found that yes, there is a
Read more

The Best Painkiller Isn't What You Think It Is

Image
When we say that listening to the right song makes us feel better, usually we mean it in terms of emotions. However, a new study from the University of Utah Health suggests that combining certain painkillers with certain music boosts the pain-fighting effect of the medication a measurable amount. This recent bit of research was conducted specifically on models recreating two types of pain: inflammatory and post-surgical. (Nerve pain, or neuropathic pain, was not addressed.) Low doses of four medications were explored: ibuprofen, cannabidiol, levetiracetam, and NAX 5055. As for music, the researchers went with one artist only, while the control groups were given one of the four drugs and subjected to white noise. What music would you choose for optimal painkilling effect? The experimenters went with Mozart. (Interestingly, the phrase “Mozart therapy” already exists, referring to the beneficial effect ol’ Amadeus can have on patients with epilepsy. For instance, a 2018 st
Read more