Interesting series of documentaries about feminism - Page 38
Forum Index > General Forum |
kwizach
3658 Posts
| ||
sibs
635 Posts
What I question is the correlation between greater population of girls playing on certain areas to better girl performance on average. The graph seems all over the place, they just arbitrarily choose 4 data points to fit their narrative, if they chose 7 shit breaks down, even with the adjusted data. ![]() | ||
gruff
Sweden2276 Posts
| ||
IgnE
United States7681 Posts
| ||
killa_robot
Canada1884 Posts
On April 14 2014 16:21 IgnE wrote: Let's say you had a job where you wanted to find the very smartest person you could find to perform it. If you knew that the smartest people in the world were all male you could restrict your search to only men. Logic: You're doing it wrong. | ||
Crushinator
Netherlands2138 Posts
On April 14 2014 16:02 gruff wrote: Even if you come to a consensus that males in general is smarter (or whatever), that doesn't change the goal of feminism one bit. It's about have equal opportunities. If there is a woman good enough to do a job, then the general consensus that women isn't as well suited for the job shouldn't play a part in her not getting the job. And vice versa. Unless you can prove that one gender is unequivocally better at something why does this matter? Statistical equality in the more prestigious jobs does seem to be a goal of feminism, so such a finding would definitely have some significance. | ||
NotJumperer
United States1371 Posts
| ||
kwizach
3658 Posts
On April 14 2014 12:44 sibs wrote: You can tell it's the adjusted ratings (for age/playing frequency) by looking at the graph males have 200ish average higher ratings instead of 500~. What I question is the correlation between greater population of girls playing on certain areas to better girl performance on average. The graph seems all over the place, they just arbitrarily choose 4 data points to fit their narrative, if they chose 7 shit breaks down, even with the adjusted data. + Show Spoiler [Fig. 4] + ![]() No, you are wrong, they are not "the adjusted ratings (for age/playing frequency)", because 1. they adjust for age later on with that sample (as indicated at the end of p. 1044) and 2. that sample is made of people who only received a rating at the end of the year, and the authors do not control for the variables "number of games played in the previous three years" and "number of games played in the current year" as they do earlier in the article. In fact, the first paragraph of the section "sex differences initial ratings of new tournament players" contains the sentence "On average, the sex difference in ratings for these groups was 110 to 200 points". There is no mention of this being an adjusted value, and they write that adjusting for age did not change the result. The authors do not speak of a linear correlation but of a threshold effect. They do not arbitrarily choose four data points: they look at the areas for which women make up at least half of the players and see that they feature no gap in average ratings between male and female players. Beyond these specific results, however, they explain that it is statistically unsurprising to find a lower ratings average for women among the competitive players population. | ||
HellRoxYa
Sweden1614 Posts
On April 14 2014 16:02 gruff wrote: Even if you come to a consensus that males in general is smarter (or whatever), that doesn't change the goal of feminism one bit. It's about have equal opportunities. If there is a woman good enough to do a job, then the general consensus that women isn't as well suited for the job shouldn't play a part in her not getting the job. And vice versa. Unless you can prove that one gender is unequivocally better at something why does this matter? Agreed. However, when you say "the goal of feminism" you should be aware of the fact that not all people who call themselves feminists think the way you do. The very fact that we're discussing the claim of whether men and women are truly created equal on the most basic level (which it would seem they are not) is evidence of this. Because if your starting point is that there are no biological differences what-so-ever then you're left with trying to explain these variations due to outside forces, and outside forces alone. Even if that isn't the explanation. And so you take incorrect action which in the long run will end up hurting people. | ||
sibs
635 Posts
On April 14 2014 20:03 kwizach wrote: No, you are wrong, they are not "the adjusted ratings (for age/playing frequency)", because 1. they adjust for age later on with that sample (as indicated at the end of p. 1044) and 2. that sample is made of people who only received a rating at the end of the year, and the authors do not control for the variables "number of games played in the previous three years" and "number of games played in the current year" as they do earlier in the article. In fact, the first paragraph of the section "sex differences initial ratings of new tournament players" contains the sentence "On average, the sex difference in ratings for these groups was 110 to 200 points". There is no mention of this being an adjusted value, and they write that adjusting for age did not change the result. The authors do not speak of a linear correlation but of a threshold effect. They do not arbitrarily choose four data points: they look at the areas for which women make up at least half of the players and see that they feature no gap in average ratings between male and female players. Beyond these specific results, however, they explain that it is statistically unsurprising to find a lower ratings average for women among the competitive players population. Nah, you are wrong , they adjust it for age later because they use the 6 to 12 subset only for the graph so no need for that, but you can just look at the graph and see, look at the little dots, do they seem to average 200 or 500? There's only a few possibilties from that: 1.)girls are way better at average than other girls in relations to boy on that specific age bracket. 2.)the difference is playtime is adjusted 3.) Population is extremely lopsided? 4.) Graph is wrong. 1. It's a possibility they're very slightly better at that age bracket, considering the age adjusted net a higher difference for 4 data points, but very very unlikely they're overwhelmingly better at that age bracket. 2. The probable option. 3. Even then it doesn't look possible to get a 500point difference. 4. Seems it's just adjusted to me not wrong. Anyhow the end point is that the correlation between more girls playing and better girl performance is at very very best doubtful, there's several regions where girls "outperform" boys or perform just as well that have a great majority of boys, and at 45% to 49% you'd get 3 more data points completely contradicting their proposed explanation for the difference. The 20% to 30% bracket also has less of a rating difference than the 30 to 40% bracket, honestly the data doesn't support the magical threshold theory, it's just 4 data points on a graph that's all over the place, they just went with it IMO. | ||
kwizach
3658 Posts
On April 14 2014 23:22 sibs wrote: Nah, you are wrong , they adjust it for age later because they use the 6 to 12 subset only for the graph so no need for that, but you can just look at the graph and see, look at the little dots, do they seem to average 200 or 500? There's only a few possibilties from that: 1.)girls are way better at average than other girls in relations to boy on that specific age bracket. 2.)the difference is playtime is adjusted 3.) Population is extremely lopsided? 4.) Graph is wrong. 1. It's a possibility they're very slightly better at that age bracket, considering the age adjusted net a higher difference for 4 data points, but very very unlikely they're overwhelmingly better at that age bracket. 2. The probable option. 3. Even then it doesn't look possible to get a 500point difference. 4. Seems it's just adjusted to me not wrong. Anyhow the end point is that the correlation between more girls playing and better girl performance is at very very best doubtful, there's several regions where girls "outperform" boys or perform just as well that have a great majority of boys, and at 45% to 49% you'd get 3 more data points completely contradicting their proposed explanation for the difference. The 20% to 30% bracket also has less of a rating difference than the 30 to 40% bracket, honestly the data doesn't support the magical threshold theory, it's just 4 data points on a graph that's all over the place, they just went with it IMO. Again, you are confusing the way they treat the different samples used in the article. The sample used in the "sex differences in initial ratings of new tournament players" section is described in the first paragraph of the section (p. 1044): "we examined for each year from 1998 through 2004 the set of players of ages 6 through 12 who had established ratings at year end and who did not have a rating in any year before the previous one." This is not the same sample as the one used in the "cross-sectional analyses of sex differences" section (although this one includes the 6-12 sub-sample) (p. 1041): "The data for our study included rating information on all USCF members who were active between 1992 and 2004 and had both birth date and sex recorded in the USCF database, a total population of 256,741 tournament players." In that section, they controlled for age and play frequency "for rating lists from 1995 through 2004" and "players with established ratings in the given year". The reason you see dots on the graph you quoted around the 200 points difference region is that the non-adjusted differences in ratings between boys and girls studied in the "sex differences in initial ratings of new tournament players" section were smaller on average than the non-adjusted differences in ratings between men and women in the larger population. This is to be expected, notably since players aged 6-12 have a much smaller range of ratings than players aged 5-95. This smaller difference in ratings between young boys and girls was quantified by the authors on p. 1044: "On average, the sex difference in ratings for these groups was 110 to 200 points in favor of the males". There is no mention of any adjustment for additional variables, unlike in the earlier sections (for example, in the "sex differences in longitudinal rating changes" section, they write "For each player, we recorded four variables: 1995 year-end rating, age, number of games played in 1995, and number of games played in the previous 3 years."), and the one variable they do adjust for later on is explicitly stated, first when the adjustment is done on the 6-12 sample ("Linearly adjusting for age [...] did not change the significance or magnitude of the sex difference"), then when it is done using the data for the four areas where girls were at least as numerous as guys: "The same result was obtained in an age-adjusted analysis, which yielded a sex difference of 40.8 points (p = .53)". If they had adjusted for other variables, they would have stated so, exactly like they did in all previous sections. With regards to your final point, again, the authors do not suggest there is a linear correlation. I'm also not sure how you came to the conclusion that the "20% to 30% bracket also has less of a rating difference than the 30 to 40% bracket", since that's not how the graph looks to me - the 30-40% bracket is made of extremes but the mean doesn't seem higher than the mean of the previous one. Regardless, and whether or not you agree with the selection, the point is that in the areas with equal proportions of male and female players, there was no rating gap. And like I said, beyond these results and even if you want to discard them completely, the authors explain that it is statistically unsurprising to find a lower ratings average for women among the competitive players population. | ||
sibs
635 Posts
On April 15 2014 02:16 kwizach wrote: Again, you are confusing the way they treat the different samples used in the article. The sample used in the "sex differences in initial ratings of new tournament players" section is described in the first paragraph of the section (p. 1044): This is not the same sample as the one used in the "cross-sectional analyses of sex differences" section (although this one includes the 6-12 sub-sample) (p. 1041): In that section, they controlled for age and play frequency "for rating lists from 1995 through 2004" and "players with established ratings in the given year". The reason you see dots on the graph you quoted around the 200 points difference region is that the non-adjusted differences in ratings between boys and girls studied in the "sex differences in initial ratings of new tournament players" section were smaller on average than the non-adjusted differences in ratings between men and women in the larger population. This is to be expected, notably since players aged 6-12 have a much smaller range of ratings than players aged 5-95. This smaller difference in ratings between young boys and girls was quantified by the authors on p. 1044: "On average, the sex difference in ratings for these groups was 110 to 200 points in favor of the males". There is no mention of any adjustment for additional variables, unlike in the earlier sections (for example, in the "sex differences in longitudinal rating changes" section, they write "For each player, we recorded four variables: 1995 year-end rating, age, number of games played in 1995, and number of games played in the previous 3 years."), and the one variable they do adjust for later on is explicitly stated, first when the adjustment is done on the 6-12 sample ("Linearly adjusting for age [...] did not change the significance or magnitude of the sex difference"), then when it is done using the data for the four areas where girls were at least as numerous as guys: "The same result was obtained in an age-adjusted analysis, which yielded a sex difference of 40.8 points (p = .53)". If they had adjusted for other variables, they would have stated so, exactly like they did in all previous sections. With regards to your final point, again, the authors do not suggest there is a linear correlation. I'm also not sure how you came to the conclusion that the "20% to 30% bracket also has less of a rating difference than the 30 to 40% bracket", since that's not how the graph looks to me - the 30-40% bracket is made of extremes but the mean doesn't seem higher than the mean of the previous one. Regardless, and whether or not you agree with the selection, the point is that in the areas with equal proportions of male and female players, there was no rating gap. And like I said, beyond these results and even if you want to discard them completely, the authors explain that it is statistically unsurprising to find a lower ratings average for women among the competitive players population. They're studying sex differences in chess, so the first thing they do, is treat the data for other variables to try to see whats the sex difference, when they say sex difference after the initial analysis they mean the treated sex difference unless otherwise specified. Yes the difference between 6 to 12 was 110 to 200, instead of 150 to 200 for the whole population. Yea my bad, I meant from 30 to 40 women are more competitive than for 40 to 50. 30 to 40 there's 9 data points, 4 close to zero, 5 close to average. 40 to 50 there's 6 data points, 1 close to zero, 5 close to average. That graph is all over the place like I said, how they can draw conclusions from it is beyond me. | ||
sibs
635 Posts
We conclude that the greater number of men at the highest levels in chess can be explained by the greater number of boys who enter chess at the lowest levels. Fine but you wouldn't expect the same on the other end of the spectrum? It seems the higher the rating the less women are in it, which makes sense looking at the average, just having more players won't raise your average. Fide 2007 women/number of players. 5029/77144 = 6.5% (all rating levels) 3587/58179= 6.2% (over 2000, superior to strong club players) 1768/39155= 4.5% (over 2100) 697/20743= 3.4% (over 2200) 223/7971= 2.8% (over 2300) 66/2715= 2.4% (over 2400) 10/771= 1.3% (over 2500, grandmaster level) 1/151 = 0.7% (over 2600) | ||
NotJumperer
United States1371 Posts
| ||
kwizach
3658 Posts
On April 15 2014 03:37 sibs wrote: They're studying sex differences in chess, so the first thing they do, is treat the data for other variables to try to see whats the sex difference, when they say sex difference after the initial analysis they mean the treated sex difference unless otherwise specified. Yes the difference between 6 to 12 was 110 to 200, instead of 150 to 200 for the whole population. No, each time they account for/take into account additional variables, they explicitly specify it, including in the "sex differences in initial ratings of new tournament players" section. Why would they suddenly fail to specify it in this section and account for every variable except for age, to only bring back age later on? Your position is not reflected in what the authors state in the paper. The average difference for the 6-12 sample of 110 to 200 is unadjusted, and, as they explicitly state, adjusting for age "did not change the significance or magnitude of the sex difference". That graph is all over the place like I said, how they can draw conclusions from it is beyond me. The first result they get from the data is the overall average ratings difference between young boys and girls, and the second result they get is that there is no ratings difference in areas where there are at least as many girls as boys. On April 15 2014 04:16 sibs wrote: Also this is really puzzling: Fine but you wouldn't expect the same on the other end of the spectrum? It seems the higher the rating the less women are in it, which makes sense looking at the average, just having more players won't raise your average. For an in-depth look at increasing proportions of men at the top, which is a different issue than that of average ratings, see the paper I cited on p. 31 of this thread: Merim Bilalić, Kieran Smallbone, Peter McLeod and Fernand Gobet, "Why are (the best) women so good at chess? Participation rates and gender differences in intellectual domains", Proceedings of the Royal Society B, 22 March 2009, vol. 276 no. 1659, pp. 1161-1165. On April 15 2014 05:07 Jumperer wrote: I suspect that they came up with that conclusion because it would be controversial to say something like "Men are superior to women at chess." Going against feminism's ideal is dangerous in today's world. It's an act of committing a social suicide. The researchers would likely get no more funding and their reputation ruined because it doesn't fit the popular narrative that women = men. They tried everything from adjusting rating to only picking 4 data points. They probably don't expect anyone to ever read the paper. What would be controversial would be to say something that is directly contradicted by the results of the paper, which I'm guessing you did not read, exactly like the previous one I provided you with, since you do not have the slightest idea of what you're talking about. On April 15 2014 05:07 Jumperer wrote: "It is a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts." - Sir Arthur Conan Doyle Poor use of that quote: http://en.wikipedia.org/wiki/Hypothetico-deductive_model | ||
sibs
635 Posts
No, each time they account for/take into account additional variables, they explicitly specify it, including in the "sex differences in initial ratings of new tournament players" section. Why would they suddenly fail to specify it in this section and account for every variable except for age, to only bring back age later on? So why not just adjust for playing time and just come out with the conclusion that girls are better at chess? Because it's already adjusted. You don't adjust for variables, then come to a conclusion without taking those variables that massively pollute your data into consideration, it just makes a terrible argument. Anyhow Yea my bad, I meant from 30 to 40 women are more competitive than for 40 to 50. 30 to 40 there's 9 data points, 4 close to zero, 5 close to average. 40 to 50 there's 6 data points, 1 close to zero, 5 close to average. That graph is all over the place like I said, how they can draw conclusions from it is beyond me. | ||
ZackAttack
United States884 Posts
| ||
NotJumperer
United States1371 Posts
| ||
kwizach
3658 Posts
On April 15 2014 06:22 sibs wrote: So why not just adjust for playing time and just come out with the conclusion that girls are better at chess? Because it's already adjusted. You don't adjust for variables, then come to a conclusion without taking those variables that massively pollute your data into consideration, it just makes a terrible argument. How do you know that adjusting for playing time would lead to such a conclusion? You don't. You don't know that it would have changed anything - accounting for age didn't. In addition, the data used in that specific sample consisted in different sets of players each year - those who "had established ratings at year end and who did not have a rating in any year before the previous one". Accounting for variables like "playing time in the three previous years", taken into account for previous samples, therefore seems a lot less relevant. In any case, as I've said repeatedly, the authors did not write the variables you refer to were accounted for, and when a variable was accounted for they explicitly specified it. Like I wrote, your position is not reflected in what the authors state in the paper. I already replied. The first result they get from the data is the overall average ratings difference between young boys and girls, and the second result they get is that there is no ratings difference in areas where there are at least as many girls as boys. In addition to the data, they explain that it is statistically unsurprising to find a lower ratings average for women among the competitive players population, which is what the argument was about. | ||
Kukaracha
France1954 Posts
Not only is the argument astonishingly weak, but also surprising if it is supposed to come from psychological studies, which are supposed to share a strong link with other social sciences and humanities that stress the importance of cultural context. It truly bothers me how some can ironically present such irrigorous arguments while discussing logic and cognitive capacities. Let's not mention the lack of a satisfying definition for the term "feminism" and voilà, armchair philosophy. | ||
| ||