The magic of statistics on tylervigen.com

[Christophe]

I just came across an "informative" site which makes correlations between statistics which have nothing to do with each other: http://www.tylervigen.com/

Example: Divorces in Maine correlated to the consumption of margarine



And there are hundreds of others!

What is the take-home message? We can do everything to statistics!

[elephant]

Very fun!

The assemblages are reminiscent of Jacques Prévert's "procession".

[Christophe]

Yes fun and we can do our own statistics too

[Janic]

christophe hello

What is the take-home message? We can do everything to statistics!

or more simply that the statistics lack additional information on the narrow role between dietetics and the social behavior of individuals. The removal of a product can decrease the aggressiveness of individuals (for example sugar or any stimulant for that matter) and therefore have an effect on couples or even the non consumption of additives in these margarines as hydrogen catalysts with nickel used to solidify the product, plus preservatives and other endocrine disruptors, etc ... or even information campaigns like in France on the non use of palm oil and advice like five fruits and vegetables per day.

[sen-no-sen]

A very serious American study has shown that the number of obese people in the USA increases in proportion to the number of vehicle registrations!

Junk food is not the main cause of the wave of obesity in developed countries. In fact, western populations tend to consume fewer calories and less fat in 2000 than they did in 1960, while they are gaining weight regularly. But the continuous decline in physical activity was very strong during this period. The only factor perfectly correlated with the increase in the prevalence of obesity in the United States is the number of cars in circulation. And this is probably true in France too.

http://www.manicore.com/documentation/articles/froguel.html
All this is after all very logical!

[Gaston]

Nor should we look for logic everywhere ...

Sometimes chance does things right, so on the original site we also discover thatmiss america's age is correlated with number of murders by steam and hot objects and that the number of drownings per fall in a swimming pool pretty closely follows the number of films by Nicolas Cage and to stay in the econological domain that the number of drivers killed in train collision is very correlated with oil imports from Norway

[sen-no-sen]

Nor should we look for logic everywhere ...

Sometimes chance does things right, so on the original site we also discover thatmiss america's age is correlated with number of murders by steam and hot objects and that the number of drownings per fall in a swimming pool pretty closely follows the number of films by Nicolas Cage and to stay in the econological domain that the number of drivers killed in train collision is very correlated with oil imports from Norway

the site in question is not at all intended to seek the slightest scientific correlation, a graphic is above all a geometric figure.
The interest is to find subjects directly correlated and to bring them out.

[Christophe]

The interest is to find subjects directly correlated and to bring them out.

Of course and that's what you just did with obesity and physical activity ... but that's not at all (I think) the reason for the site that I presented in this topic!

[Christophe]

A causality and correlation file in Pour La Science: http://www.pourlascience.fr/ewb_pages/a ... -32960.php