SoylentNews is people
SoylentNews
In a landmark study involving over a million students, it appears that the reason boys dominate girls in STEM fields is not that they are better than girls at it (the reverse seems to be true) but, perversely, that gender differences are lower in non-STEM fields.
About the STEM grades, which are often abused as an explanation:
A classroom with more variable grades indicates a bigger gap between high and low performing students, and greater male variability could result in boys outnumbering girls at the top and bottom of the class.
“Greater male variability is an old idea that people have used to claim that there will always be more male geniuses – and fools – in society,” O’Dea says.
The team found that on average, girls’ grades were higher than boys’, and girls’ grades were less variable than boys’.
But girls' and boys' variability were much closer in non-STEM fields.
(Score: 5, Informative) by pTamok on Monday December 10 2018, @12:38PM (4 children)
OK, so what the study confirms is that:
(i) Girls on average get better grades than boys.
(ii) Boys grades exhibit greater variability from the boys average than girls variability from the girls average.
(iii) The difference between the average grade results is smaller in STEM subjects than non-STEM, but still exists.
(iv) The range of variability is smaller in STEM subjects than non-STEM subjects, but still exists.
Graph (c) in Figure 3 of the paper shows that in STEM subjects, once you get into the top 10% of achievement, the sex ratio skews towards boys.
This means that if you have a purely meritocratic hiring policy which successfully hires only those in the top 10% of grades in STEM subjects, (all other things being equal) the sex ratio of the hired staff will be skewed towards males. The higher you set the grade achievement barrier, the greater the maleward skew. If you want a 50/50 sex ratio, you will need affirmative action, preferentially hiring females on some other criterion/criteria.
And, you will see the same effect in non-STEM subjects, but the crossover point to where the sex-ratio skews male is the top 2% of grades. So across all subjects, you can expect the sex distribution of the people at the absolute pinnacle of achievement to be highly skewed towards males. I would suggest that most recruitment strategies do not aim to get one of the '2%', but probably do aim to get one of the 10%. In that case, in non-STEM subjects, you would expect an unbiased grade-based meritocratic recruitment strategy to skew towards females.
Whether this should be changed is an ideological question. Whether it can be changed is an open question, as the effect appears stable across many studies.
(Score: 0) by Anonymous Coward on Monday December 10 2018, @04:03PM (3 children)
I think you mean Overall has top 2% skew towards men, not Non-STEM. I'm just looking at the graphs, https://www.nature.com/articles/s41467-018-06292-0/figures/1 [nature.com] ,but my interpretation is that for non-STEM, the top grades are heavily skewed female. The overall numbers have the top 2% skewed male, but the non-STEM fields skewed female in the top 30-40% of grades. (A pretty massive difference)
(Score: 1) by pTamok on Monday December 10 2018, @04:11PM (2 children)
I'm looking at graph (c) on Figure 3 [nature.com], which is clearly labelled.
(Score: 0) by Anonymous Coward on Monday December 10 2018, @05:57PM (1 child)
I think I may be misinterpreting what the graphs I was looking at are supposed to indicate. Although, reading the text more thoroughly isn't helping me a whole lot either. I must be one of those below average boys. The one you're looking at are more detailed at least and match what you were saying. Thanks for the clarification!
(Score: 1) by pTamok on Monday December 10 2018, @06:38PM
No problem.
I spend a reasonable amount of time reading this sort of stuff, and I don't always get it right.
If you want to get better at extracting the information from a scientific paper (and sometimes the papers are so bad there isn't any, it is all just smoke-and-mirrors misuse of statistics), then I would recommend starting with reading some of the website articles that pop up if you put 'how to read a scientific paper' (without the quotation marks) into your search engine of choice.
Unfortunately, understanding a lot of good scientific work will require that you have at least a familiarity with the statistical methods used. You can't really avoid it, because without it, incorrect conclusions are easily reached. This doesn't mean you need to have majored in statistics at university, but you should aim to know what the terms used mean.
There is often the problem that the people who produce the publicity material for universities and research institutions don't understand the papers, and put entirely incorrect spins on the. This irritates the researchers greatly, but there isn't a lot they can do about it. So be wary of taking publicity materials at face value.
The social sciences is one area where statistics are often misused, or even abused. Even well regarded highly-cited papers in the field can have fatal flaws. Just like other human endeavours, some people make mistakes through ignorance; others set out deliberately to mislead. Telling the difference between the two can be difficult.
Good luck with your further reading.