Chocolate. Like many of us today, chances are you already ate some. Dark, milk or white, filled with praline, vanilla cream or marzipan, almost everyone has a favorite type. Albert, my cat, never really understood why humans are so fond of the stuff. Even though most of them will deny it, the large amount of sugar in chocolate is one of the main reason it has grown so popular. And Albert, like all cats, cannot taste sugar. While we are at it, what exactly is chocolate made of? Milk chocolate, by far the most consumed type of chocolate, is, per 100g, made of 51.5g of sugar, 29.7g of fat, 7.9g of carbohydrates that are not sugar, 7.6g of protein, 1.5g of water, 0.37g of potassium, 0.21 of phosphorus, 0.20 of theobromine, 0.19 of calcium and, finally, of many other things that do not weight much: vitamins, caffeine, etc. The benefits of chocolate, you often hear about them in the press. A few years ago, it was not possible to miss something that was on most press covers: eating chocolate would not only improve your health, but also your intelligence. And while chocopiles were delighted by this news, that chocolate magnates around the world were rubbing their hands due to this unexpected free publicity, statisticians were busy hitting their head against the wall and taking pills. What happened? While chocolate is something a lot of people are fond of, the research paper that got the press excited a few years ago NEVER proved that eating chocolate would improve your intelligence. The researchers behind the paper just discovered a correlation, that they published in 2012. Let's take a look at their research. What they analyzed, is the chocolate consumption per capita in a set of countries and the number of Nobel prizes received per capita in the same set of countries. And what happens when you make a graph of all this? Something impossible to miss appears. The more a country's inhabitants consume chocolate, the more Nobel prizes you get in this country. Albert does not understand what's the problem. Eating chocolate helps improve one's intelligence, aren't the data clear enough? No Albert, it does not work that way. When a correlation between two variables A and B is observed, most people think, naturally, that A causes B. But what most people miss, is that B might actually cause A. Regarding chocolate and Nobel prizes, maybe that each time a country gets a Nobel prize, its inhabitants celebrate the event with a chocolate orgy. In this scenario, the large consumption of chocolate per capita is the consequence and not the cause of the many Nobel prizes received in the country. But there is yet another possibility in order to explain the observed correlation between A and B. Maybe that neither A nor B is the consequence of the other but that a third variable, let's call it C, causes both A and B. Statisticians call C a confounding factor. Are you lost Albert? Let's try with an example. A team of researchers approaches people in the street. To each person they meet, they offer a candy and ask two questions. One: did you sleep with your shoes on last night? Two: did you wake up with a headache this morning? After having gathered hundreds of answers, the researchers start analyzing the data and then observe something remarkable: those who went to bed with their shoes on were almost always the ones waking up with a terrible headache. For those who went to bed without shoes, waking up with a headache was much rarer. A strong correlation between shoes on and headache in the morning is thus observed. This correlation is published in a research journal, the press quickly gets interested and we end up with cover pages such as "Proven: if you sleep with your shoes on, you'll get a headache!" Of course, sleeping with your shoes on is safe, you won't get a headache from it! Sleeping with your shoes on and waking up in the morning with a terrible headache are both the result of partying too hard during the night. This is our third variable, C, the confounding factor, about which both researchers and journalists should have thought. But everyone was so obsessed with publishing, no one took the time to think. Let's come back to chocolate. Can we figure out an obvious confounding factor that would explain the observed correlation between chocolate and Nobel prizes? What do you think, Albert? Think hard, I'm sure you can figure it out! You still have a few seconds... Well done, Albert. What matters here is wealth! The wealthier a country is, the more it will be able to spend on education and research, thus doing its best to support its brightest citizens towards a Nobel prize. At the same time, in such a country, people can usually afford chocolate, which is still a luxury for most people in the world. What should you remember from all this? That the world is an insanely complex place and that, unfortunately, discovering a correlation between two things is not always enough to conclude anything. Too often though, we forget about this. Remember, Albert: when two things are correlated, it does not necessarily mean that one is causing the other. This video was made possible thanks to the passion of the people behind Albert. If you love Albert as much as we do and want to see more of him, there is something fairly simple you can do to keep us motivated: talk about him around you. And subscribe to his YouTube channel!