In 1955, US President Dwight Eisenhower suffered a heart attack. Eisenhower insisted on making details of his illness public instead of pretending it didn’t happen. So, the next day, his chief physician, Dr Paul Dudley White, gave a press conference at which he instructed Americans on how to avoid heart disease – stop smoking and cut down on fat and cholesterol. In a follow-up article, White cited the research of a nutritionist at the University of Minnesota, Ancel Keys.
Keys’ “diet-heart hypothesis” (or “fat hypothesis”) stated that excess saturated fats in the diet – from red meat, cheese, butter, and eggs – raises cholesterol, which congeals on the inside of coronary arteries, causing them to harden and narrow, until the flow of blood is staunched and the heart seizes up.
Keys was brilliant, charismatic, and combative. When faced with opposition, he used a 5,000 subject study he had conducted in 7 countries that proved his hypothesis. With support from the President and his physician, he destroyed any opposition to his hypothesis. His work was central in the 1980 dietary guidelines issued by the US government that made fat the enemy.
There was just one problem – Ancel Keys was wrong.
Keys was the original big data guy – a contemporary remarked: “Every time you question this man Keys, he says, ‘I’ve got 5,000 cases. How many do you have? – Ian Leslie, The Guardian
Thanks to source and health.gov for the image
—
Source and thanks to: The Sugar Conspiracy by Ian Leslie in the Guardian – a fantastic piece of journalism that inspired this 4 part series.
A Learning a Day 
But 5000 is a big number! No way he could be wrong…
exactly..
—
http://www.ALearningaDay.com – *Never failure, only learning and never older, only better..*
Rohan:
(This response was originally to Luke Miles. You responded to his post but I am not sure whether you will otherwise see my response to him so I have taken the liberty of replying directly to you as well.):
Why not? If he was looking for something specificand found them, he would continue to find them whether the sample sixe was 1, 100, 1,000, 10,000, 100,000 or a million.
Keys’ mistake was a logical fallacy of the “All dogs have four egs. My cat has four legs therefore my cat is a dog” type.
He found high levels of cholesterol in the bodies of those who were susceptible to or had suffered from heart attacks. He therefore concluded that high cholesterol levels were the primary cause of heart attacks. As scientists say, “correlation is not the same as causation”.
The famous hypothetical example is the “sales of ice creams cause murder” one. In summer, the sales of ice cream increase dramatically. In summer, the number of murders also increases dramatically. Therefore, the sale of ice creams is responsible for the increase in the murder rate.
An example like this is obviously ridiculous. Although there is a correlation (a relationship) between the sales of ice cream and the number of murders, you would need to supply an awful lot of supporting evidence if you seriously wanted to argue that the one thing caused the other.
The danger with this type of reasoning is that it can be harder to see that your thinking may be fallacious when the two sides of the correlation appear to be reasonable. Adding to the sample size without challenging how you are thinking about what you are finding simply leads to confirmation bias/ You see more and more of what you expect to see which reinforces your belief that your hypothesis is right.
The difference between bad science and good science is that the bad scientist stops as soon as he thinks he has an answer which fits the observed facts. A good scientist tries to destroy his own hypothesis by tackling it from other viewpoints and experiments. The more his hypothesis stands up to destructive testing, the more likely it is that his hypothesis models reality – at least for a given set of circumstances.
The charge against Keys is that he was a bad scientist. He stopped thinking as soon as he thought he had the right answer.
Thanks. I think Luke was being sarcastic about 5000 responses. :)
Rohan :
Ah! Possibly. So difficult to tell in print!
Very true. :)
Luke Miles :
Why not? If he was looking for something specificand found them, he would continue to find them whether the sample sixe was 1, 100, 1,000, 10,000, 100,000 or a million.
Keys’ mistake was a logical fallacy of the “All dogs have four egs. My cat has four legs therefore my cat is a dog” type.
He found high levels of cholesterol in the bodies of those who were susceptible to or had suffered from heart attacks. He therefore concluded that high cholesterol levels were the primary cause of heart attacks. As scientists say, “correlation is not the same as causation”.
The famous hypothetical example is the “sales of ice creams cause murder” one. In summer, the sales of ice cream increase dramatically. In summer, the number of murders also increases dramatically. Therefore, the sale of ice creams is responsible for the increase in the murder rate.
An example like this is obviously ridiculous. Although there is a correlation (a relationship) between the sales of ice cream and the number of murders, you would need to supply an awful lot of supporting evidence if you seriously wanted to argue that the one thing caused the other.
The danger with this type of reasoning is that it can be harder to see that your thinking may be fallacious when the two sides of the correlation appear to be reasonable. Adding to the sample size without challenging how you are thinking about what you are finding simply leads to confirmation bias/ You see more and more of what you expect to see which reinforces your belief that your hypothesis is right.
The difference between bad science and good science is that the bad scientist stops as soon as he thinks he has an answer which fits the observed facts. A good scientist tries to destroy his own hypothesis by tackling it from other viewpoints and experiments. The more his hypothesis stands up to destructive testing, the more likely it is that his hypothesis models reality – at least for a given set of circumstances.
The charge against Keys is that he was a bad scientist. He stopped thinking as soon as he thought he had the right answer.