The Old Switcharoo
Gapminder.org’s Dr. Hans Rosling sent out this message on Twitter on April 1:
In Turkey 20% of teenagers want a science career compared to 5% in US http://t.co/c6M7Zber
The link goes to a graph from Google’s Public Data Explorer, which sure enough seems to say that among 15-year-old students, far more young Turks expect a science career than young Americans:
Turkey is near the top of the list (beaten only by Estonia), while the US is near the very bottom.
But wait, if you look down in the lower-left corner of the screenshot, you can see that this graph is actually only for students who are expecting careers in ‘Computer sciences and engineering’. (I don’t know if that means all forms of engineering or just computer engineering. Personally, I wouldn’t classify all forms of engineering as a subset of science, so I suspect it means just computer engineering, but who knows?)
If you click the ‘Clear’ button, you get shown a graph of 15-year-old students who expect a career in any kind of science, not just computer science.
(click image to enlarge)
This graph shows the exact opposite story. Now, the US comes out on top and Turkey is quite far down the list.
Personally, I think this is a dramatic example of how it’s possible to use statistics to tell whichever tale you’d like just by the choice of which graph you show and which you don’t.
I had a landlady once who had worked as a photographer. At a speech by a local politician, she took some photos from the back of the room, showing the speaker standing in front of a large crowd. But she also took some photos from a more oblique angle at the front of the room and this time the photos showed the politician talking to just the couple of people who were sitting at the very front of the room.
Depending on which angle you look from, you can take away whichever impression you prefer.