class rank (statistically speaking)

correguachin

class rank (statistically speaking)
« on: January 23, 2007, 08:31:07 PM »
i took one year of statistics in college nearly five years ago, so i might be a little rusty at this... but i think you can roughly approximate your ranking based on the following assumptions and equations

(1) assume grades fall on a normal distribution

(2) z = (x - u) / s;  where x = your GPA, u = the mean GPA, and s = standard deviation

(3)ESTIMATE YOUR STANDARD DEVIATION!! s = (L - S)/4;  where L = highest possible GPA, S = lowest possible GPA (note: since all Fs are unheard of, I simply put the lowest possible GPA as a D average, which makes sense, at least at my school where all As is about as rare as all Ds); instead of 4, you can use 6 if you are more confident about your L and S estimates...

(4) calculate your z, and reference the following table http://www.sruweb.com/~walsh/witte_z_table_1.jpg; your z number should be a three digit number (negative or positive) with a decimal after the first number

If your z number is positive, you are ahead of the median!! look at the second column; it will tell you what percentage of students are ahead of you!!

If your z number is negative, you are behind the median!!
look at the second column, it will tell you what percentage of students are behind you!!

I hope this helps.. this is a rough, rough, rough approximation.. hahahaha we've made some huge assumptions here (like everything falls on a normal distribution).. many teachers have told me though that if you have a big enough sample size (i.e. class size), usually it works out that way... besides, most schools enforce a distribution, but it isn't always normal.. close enough, though

if there is some other more sophisticated way, i'd love to know..

math nerds of the world, unite!

jimmyjohn

• 141
Re: class rank (statistically speaking)
« Reply #1 on: January 23, 2007, 08:50:53 PM »
You lost me at (1).

brewha

• 140
• I drank much of this after the Bar Exam....
Re: class rank (statistically speaking)
« Reply #2 on: January 23, 2007, 10:51:57 PM »
I got to the (1   of (1) and gave up.

yvyty

Re: class rank (statistically speaking)
« Reply #3 on: January 24, 2007, 04:02:02 AM »
Thanks...I almost hope that isn't correct because it puts me at top 25%. [3.44 gpa with a 3.0 curve, hoped for higher rank]
Useful to know though.

rhombot

• 715
• hanthala
Re: class rank (statistically speaking)
« Reply #4 on: January 24, 2007, 04:43:10 AM »
problem: estimating the standard deviation by dividing the range by 4 ought to assume taking the highest and lowest *actual* grades, and if i remember my stats class correctly, only after eliminating outliers. the method suggested here takes the highest and lowest *possible* grades. unless there are non-outliers at the top and bottom of the curve, this will overstate the standard deviation, and give you a lower class rank than you actually have.

roygbiv

Re: class rank (statistically speaking)
« Reply #5 on: January 24, 2007, 07:48:42 AM »