Internet Explorer vs. US homicide count
The thesis that " correlation is not equal causality " at the end of January was again illustrated by a wonderful schedule comparing the number of homicides in the US and the proportion of Internet Explorer among the browsers preferred by users.
According to the author of the schedule: “Of course, I know that correlation does not mean a cause-and-effect relationship, but it still seems to me that there’s definitely something in it ...”
')
In fact, of course, before us is an example of a funny (and provocative) correlation. Its fun is how cleverly the correlation between two completely unrelated values was noticed; the trouble is that he is trying to compare percentages and absolute values.
But isn't it interesting how all these values correspond to the truth?
The number of murders in the United States can be checked with Wikipedia (although their number per 100,000 people is indicated there), the popularity of IE can be seen in the annual reports of W3Schools . The percentages for W3Schools turned out to be slightly lower than on the original schedule, and to clarify the population size, we had to visit the US Census Bureau.
Below are two graphs, one for 2003-2011, and the other for 2007-2011, which uses an approximation of the original graph data. As you can see, the original numbers are very accurate.
However, as noted at the beginning, the comparison was initially incorrect, so a couple of graphs were made. In the first case, data were taken for the years 2003-2011, in the second - for the years 2007-2011.
As you probably already noticed - unfortunately, despite the fact that both broken curves have a similar tendency to decrease, the rates of their change are still noticeably different.
According to the author of the schedule: “Of course, I know that correlation does not mean a cause-and-effect relationship, but it still seems to me that there’s definitely something in it ...”
')
In fact, of course, before us is an example of a funny (and provocative) correlation. Its fun is how cleverly the correlation between two completely unrelated values was noticed; the trouble is that he is trying to compare percentages and absolute values.
But isn't it interesting how all these values correspond to the truth?
The number of murders in the United States can be checked with Wikipedia (although their number per 100,000 people is indicated there), the popularity of IE can be seen in the annual reports of W3Schools . The percentages for W3Schools turned out to be slightly lower than on the original schedule, and to clarify the population size, we had to visit the US Census Bureau.
New graphics
Below are two graphs, one for 2003-2011, and the other for 2007-2011, which uses an approximation of the original graph data. As you can see, the original numbers are very accurate.
In percents
However, as noted at the beginning, the comparison was initially incorrect, so a couple of graphs were made. In the first case, data were taken for the years 2003-2011, in the second - for the years 2007-2011.
As you probably already noticed - unfortunately, despite the fact that both broken curves have a similar tendency to decrease, the rates of their change are still noticeably different.
Source: https://habr.com/ru/post/174675/
All Articles