Male/female LAC acceptance rate differentials - Page 3

Corbett · 10-25-2009, 09:22 PM

Quote:

Corbett, now that you've flamed Keil for not doing the analysis that you suggested, should we interpret your lack of provision of that same analysis as evidence that you need remedial stats work? Or is it just that you'd rather flame someone than provide useful information?

Neither interpretation is accurate. In fact, my hope was that the OP (or other interested parties) would be challenged to play with the numbers and to figure out for themselves how percent differences are calculated. Or at least to consider the problem as an exercise in Internet research.

If these approaches seem unrealistically difficult, then yes, I can simply post the formula for calculating the percent difference between two numbers. It's just not in the LAC spirit to do it that way; one of the traditional goals of liberal arts colleges is to prepare students to think and learn on their own.

odyssey212 · 10-25-2009, 09:28 PM

Omg. I am really suprised at the Pomona stats. I had no idea it was required 50-50 by charter.

So if I'm a prospective political science major, then I'll probably have a better chance at CMC anyways?

Corbett · 10-25-2009, 10:00 PM

Quote:

Wow, the Vassar differential is huge. 13%!

In absolute terms, this seems much larger than the Pomona differential, at only 7.6%.

But in relative terms, the schools both discriminate in favor of men to nearly the same degree. At Pomona, the acceptance rate for men is about 60% higher than the acceptance rate for women; at Vassar, the figure is about 62%.

The absolute difference between male/female acceptance rates is over 14% at William and Mary (based on 2008-09 CDS). However, it's only about 50% in relative terms.

Keilexandra · 10-25-2009, 10:40 PM

^^^ I was hoping, rather, that you would contribute information to this thread and do your own calculations (I've already done the time-consuming part, collecting CDS data). It's also annoying to fix Excel formatting, which doesn't copy neatly to plain-text columns. Contrary to popular belief, I do have homework and a social life outside CC.

odyssey - As a female polysci major, you definitely have a better shot at CMC than Pomona. CMC is actually majority-male in enrollment. I'm 95% sure about Pomona's charter requirement, but haven't bothered to look up an official source.

Statistically, what causes the discrepancy between absolute and relative differentials?

Kei-o-lei · 10-25-2009, 10:54 PM

Corbett-

Thanks for the professorial guidance.

But this is not an LAC; it's a message board where people who post their opinions also ocassionaly do work to help the discussion along.

It's not "unrealistically difficult" to do what you suggest; it's that many of us have other things we are doing in their lives that get in the way of trying to provide the formulae you suggest would be more helpful. The OP already did some heavy lifting to start this discussion.

If you have some interesting statistical evaluation to add to the discussion in the form of revised tables, feel free to provide them. If not, that's OK . . . but no denigrating others for likewise declining.

Kei

middsmith · 10-26-2009, 12:01 AM

Quote:

Originally Posted by corbett

But in relative terms, the male acceptance rate at Pomona is about 60% higher than the female rate. Statistically, if you have a Y chromosome, your chances of acceptance at Pomona rise by approximately 60%. At Skidmore, on the other hand, the male acceptance rate is only about 16% higher than the female rate.

Pomona

Male AR - 20.3%
Female AR - 12.7%

Differential AR - 7.6%

I believe the 60% is Differential AR/Female AR = 60%.
This is partly why I didn't care for Pomona.

broetchen · 10-26-2009, 12:14 AM

at the risk of being taken to task for my horrendous lack of math skills, I am going to admit to being confused by this

Quote:

CMC 19.1% 22.2% 21.4% -0.8%

Why is the overall admit rate lower than both male and female rates? Am I missing something very obvious?

Keilexandra · 10-26-2009, 02:10 AM

^ Maybe because I mixed up my sources. I can't remember if I used CB for the overall admit rate b/c I was too lazy to calculate it by hand... quite possibly.

Corbett · 10-26-2009, 08:56 PM

Quote:

Statistically, what causes the discrepancy between absolute and relative differentials?

The same absolute difference can seem very significant or quite insignificant, depending on the magnitude of the numbers involved.

Suppose you are starting college and are shopping for a laptop in the college bookstore. You find two attractive models, priced at $749 and $759. You probably aren't going to be terribly concerned about the $10 price difference; instead, you would likely perceive the prices as essentially equivalent.

Now suppose that as you are standing in the checkout line with your laptop, you realize that you will need to go straight to class, and that you need a pen. The checkout counter has a bin of cheap ballpoints for $1 each, and also fancy fountain pens for $11 each. In this case, a $10 difference -- the exact same absolute differential -- probably looms much larger than it did with the laptops.

So intuitively, the same $10 absolute differential seems insignificant for a laptop, but quite significant for a pen. If you express the difference in relative terms, this becomes evident. The pricier laptop costs only 1.3% more than the cheaper laptop, but the pricier pen costs 1,000% more than the cheaper pen.

*****

Now consider colleges in the same light. Suppose your safety school has an 80% acceptance rate for women and a 90% acceptance rate for men. That’s a 10% differential in absolute terms, but it probably won’t trouble you, since the odds of getting in are very high regardless.

Now suppose your dream LAC has a 10% acceptance rate for women and a 20% acceptance rate for men. That’s the same 10% differential in absolute terms. But at the LAC, it’s more troubling, since the odds of getting in are low to begin with. Furthermore, that 10% differential means that a random male applicant’s chances are twice as good as yours (since 20% is twice as large as 10%). This is not the case at the safety school; 90% is higher than 80%, but it's obviously not twice as high.

In relative terms, the same 10% absolute differential would appear quite different at the two schools. At the safety school, the male acceptance rate is only 12.5% higher than the female rate. At the LAC, the male acceptance rate is 100% higher (or twice) the female rate.

Corbett · 10-26-2009, 09:09 PM

The secret formula for calculating the percent difference is as follows. This assumes that you want to know how much higher (or lower) the male acceptance rate (M) is, relative to the female rate (F):

(M / F * 100) – 100

So at Pomona, (20.3 / 12.7 * 100) – 100 = 59.8 %
Or at Skidmore, (34.7 / 27.3 * 100) – 100 = 27.1 %

The absolute differentials are nearly identical (7.6 vs. 7.4), but this is more significant at Pomona, where the acceptance rates are lower overall. I posted an incorrect Skidmore number in Post 4 above, but it doesn’t change the conclusion. If the male acceptance rate is lower for a particular school, then the calculated result will be negative.

You can also calculate the percent difference as per Post 36

Keilexandra · 10-28-2009, 10:11 PM

I will post the actual data later when I have time to fiddle with formatting, but some highlights from sorting by %Diff:
* Harvey Mudd at 44.1% highest favor to F
* Pomona at 59.8% highest favor to M
* 6 schools with double-digit favor to F: Harvey Mudd, Union, Colorado, Bucknell, DePauw, Trinity (CT)
* 12 schools with double-digit favor to M: Pomona, Middlebury, West Point, Swarthmore, Skidmore, W&L, Wesleyan, Bowdoin, Kenyon, Davidson, Amherst, Carleton
* Single-sex colleges excluded

Corbett · 10-29-2009, 01:25 AM

OK, here's another math wrinkle. Positive and negative numbers that you generate from the formula in Post #40 above aren't directly comparable. For example, a score of +10% means male favoritism, and a score of -10% means female favoritism -- but not to the same degree. A school scoring +10% is not the "opposite" of a school scoring -10%. Sp "double-digit" scores don't have the same meaning if positive and negative values are being compared.

Suppose School A has a 40% acceptance rate for men, and a 25% rate for women. Suppose School B has exactly opposite rates: 25% for men, 40% for women. If you work through the math, you will get a score of +60% for School A, but -37.5% for School B. For purposes of your rating, you would probably want School A and School B to generate the same scores, except with opposite signs.

If you want to compare schools that favor men vs. schools that favor women, it would be better to use two different formulas.

Use (M / F * 100) - 100 at schools that favor men
Use (-1) * (F / M * 100) - 100 at schools that favor women

The -1 is arbitrarily included in the second formula, so that schools that favor women will get negative scores. Schools that favor men will get positive scores.

With these formulas, Schools A and B have exactly opposite acceptance rates and exactly opposite scores: +60% at School A and -60% at School B.

If you use this approach, Harvey Mudd gets a score of -78.8%. Harvey Mudd actually favors women to an even greater extent than Pomona (+59.8 %) favors men. In fact, the most extreme differences in gender acceptance rates may well be the female favoritism shown by some tech schools. For example, MIT and Caltech would both get triple-digit negative scores (< -100%), because the female acceptance rates are more than double the male acceptance rates.

Keilexandra · 11-01-2009, 06:24 PM

Here's a list using Corbett's formula, sorted by % difference in male/female admission rates.

Code:

Rank	School	        Coed AR	M AR	F AR	F-M D	% Diff
14	Harvey Mudd	31.1%	29.7%	53.1%	23.4%	78.8%
6	Pomona	        15.6%	20.3%	12.7%	-7.6%	59.8%
4	Middlebury	16.8%	20.3%	14.4%	-5.9%	41.0%
14	West Point	15.7%	16.6%	12.0%	-4.6%	38.3%
3	Swarthmore	15.7%	18.8%	13.7%	-5.1%	37.2%
46	Skidmore	29.8%	34.7%	27.3%	-7.4%	27.1%
14	W&L             16.8%	18.7%	15.2%	-3.5%	23.0%
13	Wesleyan	27.2%	30.4%	25.0%	-5.4%	21.6%
6	Bowdoin	        18.6%	20.5%	16.9%	-3.6%	21.3%
33	Kenyon	        31.3%	34.4%	29.0%	-5.4%	18.6%
8	Davidson	25.7%	28.0%	23.7%	-4.3%	18.1%
43	Union	        39.2%	36.5%	42.5%	6.0%	16.4%
24	Colorado	26.0%	23.9%	27.8%	3.9%	16.3%
2	Amherst	        14.8%	16.0%	13.8%	-2.2%	15.9%
30	Bucknell	29.9%	27.8%	32.0%	4.2%	15.1%
43	DePauw	        64.6%	61.7%	70.0%	8.3%	13.5%
8	Carleton	27.5%	29.3%	26.0%	-3.3%	12.7%
36	Trinity (CT)	41.7%	39.1%	44.0%	4.9%	12.5%
49	Reed	        32.5%	30.6%	33.9%	3.3%	10.8%
14	Grinnell	43.0%	40.8%	44.9%	4.1%	10.0%
22	Colby	        30.9%	29.5%	32.1%	2.6%	8.8%
36	Sewanee	        64.0%	61.5%	66.5%	5.0%	8.1%
46	Centre	        62.8%	60.5%	64.9%	4.4%	7.3%
25	Bates	        29.2%	30.3%	28.3%	-2.0%	7.1%
43	F&M	        35.9%	34.6%	37.0%	2.4%	6.9%
46	Dickinson	44.2%	42.7%	45.3%	2.6%	6.1%
21	Hamilton	28.1%	27.2%	28.8%	1.6%	5.9%
1	Williams	17.0%	17.4%	16.6%	-0.8%	4.8%
11	CMC	        19.1%	22.2%	21.4%	-0.8%	3.7%
22	Oberlin	        32.7%	33.3%	32.2%	-1.1%	3.4%
49	Pitzer	        22.3%	22.7%	22.1%	-0.6%	2.7%
40	Furman	        57.3%	58.0%	56.7%	-1.3%	2.3%
30	Richmond	31.7%	32.1%	31.4%	-0.7%	2.2%
49	Gettysburg	37.8%	38.1%	37.6%	-0.5%	1.3%
19	Colgate	        23.9%	24.1%	23.8%	-0.3%	1.3%
49	St. Olaf	58.9%	58.5%	59.2%	0.7%	1.2%
36	Holy Cross	33.8%	33.6%	33.9%	0.3%	0.9%
36	Whitman	        45.8%	46.0%	45.6%	-0.4%	0.9%
10	Haverford	27.0%	27.1%	26.9%	-0.2%	0.7%
29	Macalester	41.1%	41.0%	41.3%	0.3%	0.7%

4	Wellesley	36.0%	0.0%	36.0%	36.0%	n/a
11	Vassar	        25.0%	n/a	n/a	n/a	n/a
18	Smith	        47.7%	0.0%	44.7%	44.7%	n/a
19	Naval Academy	13.9%	n/a	n/a	n/a	n/a
25	Mt. Holyoke	52.6%	0.0%	52.6%	52.6%	n/a
25	Bryn Mawr	48.8%	0.0%	48.8%	48.8%	n/a
25	Scripps	        43.4%	0.0%	43.4%	43.4%	n/a
30	Barnard	        28.5%	0.0%	28.5%	28.5%	n/a
33	Occidental	39.4%	n/a	n/a	n/a	n/a
35	Lafayette	37.2%	n/a	n/a	n/a	n/a
40	Bard	        25.2%	n/a	n/a	n/a	n/a
42	Connecticut	36.6%	n/a	n/a	n/a	n/a

ghostbuster · 11-01-2009, 07:12 PM

Keil, are you a student or professional college counselor? With almost 3,000 posts and very granular analysis, it sort of begs the question.

Keilexandra · 11-01-2009, 07:46 PM

^Really? I am honored. I suspect a professional college counselor wouldn't give out quite so much information for free. XD (Or use emotes as gratituously as I do.)