Severn at Sailer’s wrote “Educated women tend not to have high IQ’s. The subjects they take in college do not require it. I’d say the typical “communications major” is not any more intelligent than a plumber, and is probably a good deal less so.” That sounded like something that could be investigated in the GSS.
Row: WORDSUM
Column: EDUC(r: “LT HS” 1-11; “HS” 12; “SC” 13-15; “C” 16; “GT C” 17-22)
Control: SEX
First men
| Statistics for SEX = 1(MALE) | |||||||
|---|---|---|---|---|---|---|---|
| Cells contain: –Column percent -Weighted N |
EDUC | ||||||
| 1 LT HS |
2 HS |
3 SC |
4 C |
5 GT C |
ROW TOTAL |
||
| WORDSUM | 0 | 2.3 58.4 |
.7 22.3 |
.4 10.1 |
.4 7.1 |
.5 6.0 |
.9 103.9 |
| 1 | 4.7 117.7 |
1.5 49.9 |
.7 20.0 |
.7 11.9 |
.2 2.7 |
1.7 202.2 |
|
| 2 | 8.1 203.8 |
3.8 126.8 |
1.4 38.5 |
1.4 22.2 |
.6 7.9 |
3.5 399.3 |
|
| 3 | 15.7 395.4 |
7.3 244.2 |
3.7 104.7 |
1.8 28.0 |
1.2 15.5 |
6.8 787.8 |
|
| 4 | 19.1 479.9 |
13.9 462.4 |
8.6 239.3 |
3.8 60.6 |
2.7 35.1 |
11.0 1,277.3 |
|
| 5 | 19.8 497.5 |
21.9 732.4 |
16.6 463.5 |
7.6 121.0 |
5.3 69.5 |
16.3 1,883.9 |
|
| 6 | 15.8 398.8 |
25.8 861.5 |
25.2 703.6 |
20.6 328.3 |
11.4 149.4 |
21.1 2,441.6 |
|
| 7 | 8.7 219.3 |
14.2 473.9 |
19.7 552.5 |
20.9 334.0 |
17.0 222.0 |
15.6 1,801.7 |
|
| 8 | 3.3 83.4 |
6.5 217.0 |
12.6 353.3 |
17.3 275.9 |
17.8 233.0 |
10.1 1,162.6 |
|
| 9 | 1.7 43.5 |
3.4 112.8 |
7.2 201.2 |
14.4 230.7 |
22.8 298.0 |
7.7 886.1 |
|
| 10 | .8 21.1 |
1.0 34.9 |
4.0 110.7 |
11.1 177.6 |
20.6 268.9 |
5.3 613.2 |
|
| COL TOTAL | 100.0 2,519.0 |
100.0 3,338.1 |
100.0 2,797.3 |
100.0 1,597.2 |
100.0 1,308.0 |
100.0 11,559.6 |
|
| Means | 4.49 | 5.43 | 6.26 | 7.11 | 7.83 | 5.93 | |
| Std Devs | 1.98 | 1.79 | 1.81 | 1.92 | 1.86 | 2.15 | |
| Unweighted N | 2,380 | 3,131 | 2,648 | 1,537 | 1,341 | 11,037 | |
| Summary Statistics for SEX = 1(MALE) | ||||||||
|---|---|---|---|---|---|---|---|---|
| Eta* = | .50 | Gamma = | .49 | Rao-Scott-P: F(40,19040) = | 72.37 | (p= 0.00) | ||
| R = | .50 | Tau-b = | .41 | Rao-Scott-LR: F(40,19040) = | 68.89 | (p= 0.00) | ||
| Somers’ d* = | .43 | Tau-c = | .42 | Chisq-P(40) = | 3,570.58 | |||
| Chisq-LR(40) = | 3,399.13 | |||||||
Now for women
| Statistics for SEX = 2(FEMALE) | |||||||
|---|---|---|---|---|---|---|---|
| Cells contain: –Column percent -Weighted N |
EDUC | ||||||
| 1 LT HS |
2 HS |
3 SC |
4 C |
5 GT C |
ROW TOTAL |
||
| WORDSUM | 0 | 1.8 53.9 |
.6 27.3 |
.5 18.2 |
.3 5.8 |
.7 8.1 |
.8 113.2 |
| 1 | 4.6 135.8 |
1.5 73.9 |
.7 24.2 |
.4 6.6 |
.2 2.5 |
1.7 243.1 |
|
| 2 | 8.9 263.2 |
2.4 117.1 |
1.2 41.8 |
.5 9.0 |
.2 2.9 |
3.1 434.0 |
|
| 3 | 12.2 361.3 |
4.8 230.8 |
3.2 113.2 |
.9 16.1 |
1.8 21.2 |
5.2 742.7 |
|
| 4 | 17.5 517.3 |
10.1 485.6 |
6.8 240.1 |
3.1 53.0 |
2.8 33.0 |
9.4 1,329.1 |
|
| 5 | 20.2 599.8 |
19.9 956.7 |
16.1 567.6 |
8.7 148.2 |
5.6 66.1 |
16.5 2,338.4 |
|
| 6 | 18.9 559.2 |
25.7 1,233.1 |
25.6 901.8 |
20.5 351.0 |
13.0 153.4 |
22.6 3,198.5 |
|
| 7 | 8.5 250.9 |
17.6 845.7 |
20.1 708.0 |
19.7 337.8 |
16.8 198.2 |
16.5 2,340.5 |
|
| 8 | 4.0 119.2 |
9.2 442.6 |
12.8 450.6 |
17.5 300.1 |
17.4 205.1 |
10.7 1,517.6 |
|
| 9 | 2.7 79.3 |
5.3 253.3 |
8.0 281.0 |
16.0 273.3 |
20.6 242.1 |
8.0 1,129.1 |
|
| 10 | .8 23.5 |
2.8 136.7 |
4.8 169.7 |
12.2 209.4 |
20.7 243.9 |
5.5 783.2 |
|
| COL TOTAL | 100.0 2,963.5 |
100.0 4,802.8 |
100.0 3,516.2 |
100.0 1,710.3 |
100.0 1,176.5 |
100.0 14,169.3 |
|
| Means | 4.67 | 5.88 | 6.38 | 7.28 | 7.74 | 6.07 | |
| Std Devs | 2.01 | 1.86 | 1.83 | 1.80 | 1.92 | 2.10 | |
| Unweighted N | 3,186 | 4,818 | 3,535 | 1,745 | 1,226 | 14,510 | |
| Summary Statistics for SEX = 2(FEMALE) | ||||||||
|---|---|---|---|---|---|---|---|---|
| Eta* = | .44 | Gamma = | .43 | Rao-Scott-P: F(40,19040) = | 68.22 | (p= 0.00) | ||
| R = | .43 | Tau-b = | .35 | Rao-Scott-LR: F(40,19040) = | 63.52 | (p= 0.00) | ||
| Somers’ d* = | .38 | Tau-c = | .36 | Chisq-P(40) = | 3,601.20 | |||
| Chisq-LR(40) = | 3,353.14 | |||||||
The correlation is higher for men, but that may in part be due high scores for relatively uneducated women. The only bucket in which men have higher mean wordsum scores is more than four years of college. Women have a higher mean overall, so that could throw things off. Still, in every bucket except “some college” the mean women’s score exceeds the mean men’s score by more than the difference between their overall means. Also note that the standard deviation for women is a bit lower than that of men, just as Larry Summers said.
April 20, 2011 at 6:25 am
Good stuff.
April 20, 2011 at 9:16 am
Idiotic generalizations like that are a big clue that you shouldn’t listen to Steve Sailer or people of his ilk.
April 22, 2011 at 9:41 pm
I guess we probably should not listen to those without the intelligence to understand that it was Severn who made the claim, not Sailer. Sailer probably does not even believe it.
Perhaps you fail to understand that blogger is not defined by the few posters whose claims you find abhorrent.
April 20, 2011 at 4:07 pm
I think you’re using the sailer blog comments + the GSS brilliantly in your last couple posts.
Too bad there’s just one of you, TGGP. I’d like to see you get some non ideologically-deforming funding for your blogging, something like whatever sustains Mr. Yglesias or Mr. Andrew Sullivan.
April 22, 2011 at 12:06 am
mtraven, Sailer may not be Andrew Gelman, but his comment section has widest discrepancies in quality relative to that of the original post. He doesn’t tend leap to generalizations quite that easily.
H.A, despite the name of the blog I don’t often feel I have an opinion worth blogging, and I’m glad to occasionally get an idea to investigate. Also, you know Yglesias is funded by an explicitly ideological organization which occasionally interrupts his blog when he goes off the reservation.
Perhaps I just need to become independently wealthy. I’ll get to work on winning the lotto.
April 22, 2011 at 1:13 pm
Perhaps Sailer’s resistance to adapting his claims to please his audience is wearing down. Someone needs to give him a loving kick in the rump until it builds back up again.
April 22, 2011 at 4:11 pm
That wordsum data covers men and women with high school, some college, a four year degree, and so on. Which is good data as far as it goes, but it says nothing about the IQ of the typical communications major relative to the average plumber.
Another problem is that the wordsum data does not line up with other, better proxies for IQ, such as the SAT test. Males outscore females on the SAT in both reading and math. Females score better on on writing section.
The term “high IQ” is open to some interpretation, but it seems plausible to say that the typical America college grad does not need a very high IQ.
April 23, 2011 at 1:15 pm
It should also be remembered that the various IQ tests weight each sub-test differently to compensate for different natural male and female average abilities on each.
April 22, 2011 at 9:45 pm
Perhaps the most we can draw from this analysis is that women tend to have higher average verbal IQs than men.
I have seen little evidence that they have higher average visuo-spatial IQs nor higher average abilities in areas like mechanics and so forth.
For example, I suspect that my abilities with computers exceeds 99% (or higher) of females out there, while it probably only exceeds 97% of males. (I have run into females with a strong knowledge of computers, but strangely, they have tended to be Chinese or Jewish.)
April 25, 2011 at 9:00 pm
melendwyr, which specific claim(s) by Sailer are you referring to?
flenser, there are some questions about occupational categories in the GSS, but that tends to thin out the data. Also, I forget where the table is that translates occupational codes to english. U don’t think they have data on majors.
Regarding the female advantage in verbal scores: I mentioned above what happens when you subtract the overall difference between men & women’s mean scores and the difference for specific educational buckets.