## Curve calculation help

(Study Tips, Dealing With Stress, Maintaining a Social Life, Financial Aid, Internships, Bar Exam, Careers in Law . . . )
Wildcat

Posts: 8
Joined: Wed Jan 20, 2010 8:42 am

### Re: Curve calculation help

rayiner wrote:Depending on your school (eg: NU) your curve may not be normally distributed. For most schools, the upper tail is very spread out and the lower tail is very bunched-up.

Our smaller school has a normal looking distribution for about the middle 70% of students, but it starts to break down at the top and bottom 15%. Reason is that once you hit a certain GPA there are so few kids around it that even two or three kids can throw it off. Statistics become less and less useful at the outer limits, especially with a smaller class size.

As an example, the current 1L class has several very smart kids who all ended with a 4.0 or higher, and a few others who ended above a 3.8 as well. The current 2L class to the best of my knowledge never had more than 2 or 3 kids above a 3.8. The curve stayed the same.

Using statistics, the predicted top 5% at our school is around a 3.77. For the 1L class it will in fact be around a 3.9, but for the 2L class it was a 3.7. If you combined the 2 classes however it would be almost dead on the money, funny how stats work like that.

steindle

Posts: 116
Joined: Sat Apr 19, 2008 9:39 pm

### Re: Curve calculation help

It's been a long, long time since I've taken Stats...

...Top 20% at my school is 3.528. Can anything more be extrapolated from that?

omg

Posts: 253
Joined: Sun Feb 08, 2009 7:32 pm

### Re: Curve calculation help

steindle wrote:It's been a long, long time since I've taken Stats...

...Top 20% at my school is 3.528. Can anything more be extrapolated from that?

I think you need at least one more data point. Do you know the median?

steindle

Posts: 116
Joined: Sat Apr 19, 2008 9:39 pm

### Re: Curve calculation help

omg wrote:
steindle wrote:It's been a long, long time since I've taken Stats...

...Top 20% at my school is 3.528. Can anything more be extrapolated from that?

I think you need at least one more data point. Do you know the median?

That's what I figured...unfortunately that's all we're given.

throughthewetgrass

Posts: 36
Joined: Sat Nov 29, 2008 6:15 pm

### Re: Curve calculation help

What about if the curve is 3.18? How would that break down %wise?

amila143

Posts: 8
Joined: Sat Aug 16, 2008 12:16 am

### Re: Curve calculation help

school median is 2.8 - got a 3.2 . anyone can do the math plzzzzzzzzzzzzzz
Last edited by amila143 on Thu Jan 21, 2010 3:02 pm, edited 1 time in total.

rayiner

Posts: 6145
Joined: Thu Dec 11, 2008 11:43 am

### Re: Curve calculation help

Alyosha wrote:
rayiner wrote:Depending on your school (eg: NU) your curve may not be normally distributed. For most schools, the upper tail is very spread out and the lower tail is very bunched-up.

So can somebody translate this for me? Does this mean that at NU, you can't use the standard deviation to determine the top 5% grade cutoff? Also, how do you know if you are at a school in which the curve is not normally distributed?

True normal distribution:

thesealocust

Posts: 8523
Joined: Mon Oct 20, 2008 8:50 pm

### Re: Curve calculation help

nm
Last edited by thesealocust on Wed Jun 30, 2010 11:25 pm, edited 1 time in total.

rayiner

Posts: 6145
Joined: Thu Dec 11, 2008 11:43 am

### Re: Curve calculation help

thesealocust wrote:rayiner: Those are for each class though, right? Once you have a lot of grades, the overall distribution of GPAs is likely to normalize (if not become normal) even if each class distributed some other way.

Let's consider the data we have from NU (thanks to TTT-LS):

NU class of 2008 degree honors*:

3.97+ - top 6%
3.65+ - top 43%

Also:

xoxo wrote:Graduated w/ a 3.89 in '08. Got cum laude; order of the coif.

There is really no way to fit a normal distribution to this data. A mean of 3.50 and an SD of 0.5 puts the 3.65 cut-off in almost the right place, but puts a 3.89 just outside top 20%, not the top 10% required for coif. Using a mean of 3.35 and an SD of 0.4 puts the 3.89 data point in almost the right place, but puts the 3.65 cut-off just outside the top 20%, not around top 43%.

Now this data has uncurved classes factored in and whatnot, but does show that final GPAs at NU aren't even close to normally distributed, even after 6 semesters.

thesealocust

Posts: 8523
Joined: Mon Oct 20, 2008 8:50 pm

### Re: Curve calculation help

nm
Last edited by thesealocust on Wed Jun 30, 2010 11:23 pm, edited 1 time in total.

omg

Posts: 253
Joined: Sun Feb 08, 2009 7:32 pm

### Re: Curve calculation help

sealocust,

besides the fact that seminars can potentially be curved to each seminar's students' median/mean GPA, are there uncurved courses at UVA? would it make any difference to the scale of distributions you showed earlier, given that those must include seminars (since obvioiusly they would be included in the law school's own statistic of where the top 25% is)?

thesealocust

Posts: 8523
Joined: Mon Oct 20, 2008 8:50 pm

### Re: Curve calculation help

nm
Last edited by thesealocust on Wed Jun 30, 2010 11:23 pm, edited 1 time in total.

omg

Posts: 253
Joined: Sun Feb 08, 2009 7:32 pm

### Re: Curve calculation help

thesealocust wrote: I actually have seen no evidence of that being true. That doesn't mean it doesn't exist, but I have looked and have not seen it.

A professor told me that thing about seminars, so that's good enough for me, haha.

Thanks!

lilybbloom

Posts: 118
Joined: Sat Aug 23, 2008 2:19 pm

### Re: Curve calculation help

can someone do the same math for the U of C curve?

all 1L classes are curved to a 177 median (out of 186).

according to an unofficial transcript, last year 179 was top 19%.

where would a 175, 176, 178, etc. put you in the class?

blzrchick2

Posts: 9
Joined: Tue Jul 22, 2008 4:25 pm

### Re: Curve calculation help

Hey TheSeaLocust, here's another one for you, if you have the time!

Median: 3.047
Top 33%: 3.275
Top 25%: 3.400
Top 10%: 3.615

What rough percentile is a 3.739?

Thanks!!

thesealocust

Posts: 8523
Joined: Mon Oct 20, 2008 8:50 pm

### Re: Curve calculation help

nm
Last edited by thesealocust on Wed Jun 30, 2010 11:20 pm, edited 1 time in total.

blzrchick2

Posts: 9
Joined: Tue Jul 22, 2008 4:25 pm

### Re: Curve calculation help

Appreciate you doing the math and giving the explanation, thanks.

Don't know if I needed the eye roll though. Just asking a question.

Edit: I have no idea if it makes a difference but that data was after 4 semesters, not 1. I'm a 3L and was just curious about how the calculations might go if the curve was indeed normal.
Last edited by blzrchick2 on Fri Jan 22, 2010 5:26 pm, edited 1 time in total.

zanda

Posts: 526
Joined: Fri Apr 24, 2009 9:36 am

### Re: Curve calculation help

thesealocust - was my inclination right about the NYU curve and needing to know more data?

Posts: 917
Joined: Mon Jun 02, 2008 12:53 pm

### Re: Curve calculation help

tag

Posts: 917
Joined: Mon Jun 02, 2008 12:53 pm

### Re: Curve calculation help

zanda wrote:thesealocust - was my inclination right about the NYU curve and needing to know more data?

yes, especially because they give a range that the profs don't have to stick to

OperaSoprano

Posts: 3417
Joined: Tue Nov 25, 2008 1:54 am

### Re: Curve calculation help

sealocust, rayiner, or others good at stats: At any given school, is the median for all students likely to be higher than the median for 1Ls? I'm not certain if that's what you guys are getting at.

drew

Posts: 114
Joined: Wed Oct 08, 2008 7:17 pm

### Re: Curve calculation help

seal, can you figure out anything from GULC's grade curve as to the median GPA, etc. (realizing that its completely speculative)? Its:

A 12%

A- 19%

B+ 28%

B 31%-36%

B- and below 5%-10%

thesealocust

Posts: 8523
Joined: Mon Oct 20, 2008 8:50 pm

### Re: Curve calculation help

nm
Last edited by thesealocust on Wed Jun 30, 2010 11:20 pm, edited 1 time in total.

rayiner

Posts: 6145
Joined: Thu Dec 11, 2008 11:43 am

### Re: Curve calculation help

drew wrote:seal, can you figure out anything from GULC's grade curve as to the median GPA, etc. (realizing that its completely speculative)? Its:

A 12%

A- 19%

B+ 28%

B 31%-36%

B- and below 5%-10%

Median is a B+, mean is probably around 3.3-3.33.

Splitt3r

Posts: 200
Joined: Thu Jul 12, 2007 2:37 pm

### Re: Curve calculation help

rayiner wrote:
drew wrote:seal, can you figure out anything from GULC's grade curve as to the median GPA, etc. (realizing that its completely speculative)? Its:

A 12%

A- 19%

B+ 28%

B 31%-36%

B- and below 5%-10%

Median is a B+, mean is probably around 3.3-3.33.

Drew and I were talking about this earlier, and though I see why it seems tempting to say that the median is a 3.3 exactly, I think at this point (we have 3 classes/11 units of grades) it's probably a 3.2 or so. I imagine that the As are pretty tightly grouped or at least that it's more often than not true that a person who got one A got two or more As, which would mean that we'd have a fair number of 3.7-3.8ish gpas, but relatively few 3.5s. Given the large percentage of Bs they give out, and the fact that B+/B line is relatively close to the natural median, it seems like the most likely scenario for the median person's grade distribution would be B+, B+, B. Obviously this will balance out over time and go towards a 3.3, but after just the first semester I think it's likely the case.