At a guess, the difference between sexes may come from the fact that men and women's bikes look very different, with men's bikes having a much more obvious functional shape.
There's one tube difference, and that one is already drawn in. Additionally, in most of the images I've seen from this sort of experiment, it's not the top tube that's in the wrong place, it's all the mechanical bits.
I get the feeling that this is a discovery well known in the art world: that you really need to study the subject you intend to draw to even get the most basic details right.
See also the art project Velopedia [1] (and its recent discusison here [2]), which consisted of images that looked like CAD renderings (but using photoshop) based on similar sketches of bicycles.
That's a seriously cool project by Gianluca :). I'm surprised by the diversity of the bikes that come out of the sample -- and what's with so many people attaching both wheels to the chain? Some of the designs look like they might work though (except for those people hell-bent on disabling steering, though...)!
If you have adjusted the seat height yourself you know that the seat tube slides within the frame. So the first/left-most bicycle is wrong because the seat tube is vertical and the frame tubes are at angles. Then you scan the four diagrams to find the one where the seat tube and one of the frame tubes form a straight line. None of them.
At this point you give up and just answer randomly.
The experimenter claims "People made lots of fundamental mistakes." but I think the experimenter made a fundamental mistake and his subjects lost interest in the experiment.
Well, if you read the description, the first task actually includes tacking on the bits of frame you think are missing.
So with all the subjects rising to your level of pedantry crowding out the other people, the drawing task should have much lower error rates than the multiple choice test. However, it seems many people got so outraged by the wrong slant on the seat tube in the stick figure drawing they didn't draw a frame at all!
(but did draw the chain and pedals in the correct place, presumably because the insult of the seat tube did not taint all of the bike)
The thing that surprised me the most was that women's error rates were as high as they were for frequent cyclists. Could it be that functional understanding of bikes comes more from bike maintenance (simple things like changing a tube) than it does from riding or daily contact, and that more female cyclists hand off their routine maintenance to men?
This is more of an "IQ test" and has the fallacies of an IQ test.
1. The sample was people that went to University Open Days.
They may not be interested to spend much time on answering.
2. It is the style of IQ question that you exclude the odd answers until you get the right one. If you are not familiar with IQ tests, you would pick one and get over with it.
11 comments
[ 2.7 ms ] story [ 36.5 ms ] threadThe final paper is here: https://www.liverpool.ac.uk/~rlawson/PDF_Files/L-M%26C-2006....
[1]: http://www.gianlucagimini.it/prototypes/velocipedia.html
[2]: https://news.ycombinator.com/item?id=11478061
At this point you give up and just answer randomly.
The experimenter claims "People made lots of fundamental mistakes." but I think the experimenter made a fundamental mistake and his subjects lost interest in the experiment.
So with all the subjects rising to your level of pedantry crowding out the other people, the drawing task should have much lower error rates than the multiple choice test. However, it seems many people got so outraged by the wrong slant on the seat tube in the stick figure drawing they didn't draw a frame at all!
(but did draw the chain and pedals in the correct place, presumably because the insult of the seat tube did not taint all of the bike)
1. The sample was people that went to University Open Days. They may not be interested to spend much time on answering.
2. It is the style of IQ question that you exclude the odd answers until you get the right one. If you are not familiar with IQ tests, you would pick one and get over with it.