Re: Black differently abled gay height challenged single mothers in Computer Science!
"In light of this assertion, lack of diversity in technology recruitment becomes more difficult to excuse as a consequence of natural ability.
I know where the author is going with this but lack of any significant pattern in the student population does not justify to force "diversity" on employers for the sake of it - because employers are not hiring in the same population, especially not if that population is reframed by adding more "diversified" people."
I'd go much further than all this. I'd say the author has taken a study that shows that coursemarks across a specific population is roughly normal, as with many things in life, and then extrapolated that to claim that this means that all substrata of the population are equally adept at things, which is a no-no.
For a start, the dataset is hideously self selecting, almost the epitome of such: final-year computer science undergrads. Now come on; if there were a geek gene, you would rather hope that all of these people have it in the first place, so we are looking only at the 'cans'. This is specifically against the author of the article here, not the paper itself.
For the authors of the original paper, I assume they looked at pre-moderated marks, and not those after they have been fit to a curve? I couldn't find any mention of this in their paper, but since this is trick-cycling and I actually have real maths to be getting on with, I didn't look too hard.
Secondly, finally for the off-the-top-of-my-head-reasons-why-this-paper-is-rubbish, exams are designed to get a normal distribution. We write a few questions that everyone will get, then attempt to write a few questions that some will, some won't, and a few very hard questions, to get this nice distribution of people.
To be fair to the authors of the study, they are trying to prove that computer science results at university level are normally distributed, just like every other subject. And they succeed at that, although my reasons above are a pretty good explanation for that even if the underlying population ability is bimodal. And then the trick-cycling comes in, and it's garbage from then on. They miss out one very good reason why lecturers think distributions are bimodal: because the people they see are bimodal. Interactions with students are mostly with the very weak and the very able, with those in between being largely invisible until exam time.
Final grade for the paper: 55%. A solid upper second, but not enough critical analysis to get onto most graduate training programmes at major companies.