On Sun, Mar 28, 2010 at 8:29 PM, wren ng thornton
Jon Fairbairn wrote:
Another (provocative) observation is that most of the women programmers I've known were good at it and thought they might not be, but most of the men claimed to be good at it but were not.
I've observed this too, but it's a bit droll. Let:
p = proportion of people who think they're good but aren't q = proportion who think they're not good but are M = number of men in CS W = number of women in CS
Given that M >> W, we'll naturally find that p*M > q*W if p and q are even remotely comparable, regardless of whether p and q are independent of gender or not.
I recall going to a PhD defense several years ago about gender differences in computer science. The dissertation is here: http://ir.library.oregonstate.edu/dspace/bitstream/1957/4954/1/FinalVersion.... A few take-away points I recall from the defense: * The difference between genders is smaller than the difference between individuals * In spreadsheet debugging tasks, women would rate their confidence lower than men * In spreadsheet debugging tasks, women would do at least as well as men (often better) * Men were more likely to jump right in without reading the instructions * Women were more likely to read the instructions and try to understand the task before starting it It's entirely possible that the cases where the women performed significantly better than the men it was largely because they took the time to read the instructions. Otherwise, it seemed like the difference in self-assessed confidence was bigger than any gender difference in measurable performance. In other words, approaches and confidence varied by gender more than results. Also, I might be completely misquoting the results. Best to read the dissertation for yourself if you find the topic interesting. Jason