Statistics? How do they work? Forum

[Samara]

Anyone out there good enough with statistics to figure this one out? I'm sure as hell not.

So, word on the street is that your LSAT score has a 0.40 correlation to your law school grades. I have two questions about this.

1) What exactly does that mean?

2) Can it be used to more accurately gauge my "odds" of getting biglaw? For T14, considering that 10-15% of students do prestigious clerkships or PI and 45% of students get biglaw, essentially the top 60% will get biglaw if they want it, right? Say my LSAT is at the 80th percentile for my school. My "odds" for getting biglaw are higher than 60%, no? What would the statistical likelihood be?

I ask this because I am a splitter who will be looking at sticker for T14 if I get in. I would like to more accurately understand the financial risks involved in paying sticker for T14. Thanks for any help!

[5ky]

Jesus christ. No, just no.

[Deuce]

lol

[Blessedassurance]

Has it come to this?

[Bildungsroman]

First, GPA and LSAT together are .4; LSAT by itself is a tiny pink .25

Second, nobody here is going to be able to predict your grades with any accuracy. If you want a thread where fellow splitters can come in and pat you on the back and say "yeah, totally, we do better at law school" then you've built a great foundation here, but this thread will be totally worthless for anything else.

[Samara]

Law school, y u no conform to my quantitative hyper-analysis???

Seriously though, the mockery is good to know. I don't expect anyone to predict my grades, but it would be good to get a more accurate picture of my "chances." There's a big difference between 50% and 75% chance of success. But even if all this thread amounts to is moral support for people like me, I'll count it in the win column.

Resume the mockery!

[bk1]

First, GPA and LSAT together are .4; LSAT by itself is a tiny pink .25

Second, nobody here is going to be able to predict your grades with any accuracy. If you want a thread where fellow splitters can come in and pat you on the back and say "yeah, totally, we do better at law school" then you've built a great foundation here, but this thread will be totally worthless for anything else.

Is the reverse splitter mad?

[bjsesq]

First, GPA and LSAT together are .4; LSAT by itself is a tiny pink .25

Second, nobody here is going to be able to predict your grades with any accuracy. If you want a thread where fellow splitters can come in and pat you on the back and say "yeah, totally, we do better at law school" then you've built a great foundation here, but this thread will be totally worthless for anything else.

Is the reverse splitter mad?

awwwwwwwwwwwwwwwwwwwwwwwwwwwww, sheeeeeeeeyaaaaaaaaaaaaaaaaaaaaaaaaaaaaat.

[Bildungsroman]

First, GPA and LSAT together are .4; LSAT by itself is a tiny pink .25

Second, nobody here is going to be able to predict your grades with any accuracy. If you want a thread where fellow splitters can come in and pat you on the back and say "yeah, totally, we do better at law school" then you've built a great foundation here, but this thread will be totally worthless for anything else.

Is the reverse splitter mad?

Hey man, I'm not a reverse splitter, my numbers are tits all around.

But seriously, reverse-splitters do the same thing too. This isn't me indulging in some anti-splitter, "they took my acceptances" rage, it's me trying to forestall a surge of neuroticism from another year of applicants.

[bk1]

it's me trying to forestall a surge of neuroticism from another year of applicants.

AKA: The impossible. (But it's not like I don't try to do the same thing.)

[dkt4]

wait, so this wasn't a flame?

[KevinP]

First, GPA and LSAT together are .4; LSAT by itself is a tiny pink .25

???
--ImageRemoved--

[Dany]

lol

[Bildungsroman]

First, GPA and LSAT together are .4; LSAT by itself is a tiny pink .25

???
--ImageRemoved--

Good catch. My memory was a little off on the numbers, but the sentiment is the same. Quoting from the same LSAC research report where you got that chart, "LSAT score alone continues to be a better predictor of law school performance compared to UGPA alone. The median validity for LSAT score alone is 0.32 for 2007 and 0.33 for 2008, compared to mean validity values of 0.28 for both 2007 and 2008 for UGPA alone." Given the small difference in correlative value between the two, neither splitters nor reverse-splitters should view their one above-median (for the school) number to mean they're going to do better than their peers at that school.

[KevinP]

Good catch. My memory was a little off on the numbers, but the sentiment is the same. Quoting from the same LSAC research report where you got that chart, "LSAT score alone continues to be a better predictor of law school performance compared to UGPA alone. The median validity for LSAT score alone is 0.32 for 2007 and 0.33 for 2008, compared to mean validity values of 0.28 for both 2007 and 2008 for UGPA alone." Given the small difference in correlative value between the two, neither splitters nor reverse-splitters should view their one above-median (for the school) number to mean they're going to do better than their peers at that school.

I agree with your sentiment. I was genuinely curious as to where you were getting the numbers since I thought maybe they released the 2009-2010 study (They're supposed to release it in October 2011). It is also kind of interesting to note that the correlation used to be .41 in 2001.

@OP:
As LSAC notes, the correlation would be much higher if the LSAT scores at a school represented the LSAT scores of the testing population in general. However, since the variation in LSAT scores at a particular school is relatively small, the correlation of the LSAT is extremely weak. Therefore, don't use the LSAT score to predict your grades.

[Samara]

@OP:
As LSAC notes, the correlation would be much higher if the LSAT scores at a school represented the LSAT scores of the testing population in general. However, since the variation in LSAT scores at a particular school is relatively small, the correlation of the LSAT is extremely weak. Therefore, don't use the LSAT score to predict your grades.

That makes sense. It does lend some (perhaps not much) credence to the idea that you would perform higher at your safety school than your reach school, no? Probably not enough to overcome the drop-off in biglaw employment levels, but still, interesting. Thanks for the insight.

And again, I'm not trying to predict my grades, I'm trying to predict the likelihood my grades will be high enough for biglaw. Yes, at the end of the day, it's kind of a moot point, but I'd still be interested to see some more robust statistical analysis.

[Tiago Splitter]

@OP:
As LSAC notes, the correlation would be much higher if the LSAT scores at a school represented the LSAT scores of the testing population in general. However, since the variation in LSAT scores at a particular school is relatively small, the correlation of the LSAT is extremely weak. Therefore, don't use the LSAT score to predict your grades.

+1. LSAT scores (and to a lesser extent, GPAs) tend to cluster around the median at most schools. Practically there is no difference between a 166 LSAT and a 169.

[AffordablePrep]

I'd imagine that it has less of an impact on predicting grades now than ten years ago so I'd be curious to know when this study took place.

I think it is true that someone who got a 170 is going to be better at seeing different sides/holes/assumptions in an argument than someone who got a 150.

However, back in the day, being a lawyer normally led to good prospects even if someone struck out at big law. So the LSAT was probably a better way of showing natural ability before literally over half of people taking the test learned "tricks" that if mastered virtually are guaranteed to lead to the 75th-80th percentile with a little motivation, as why bother busting your butt studying if the difference between a 160 and 170 was not life changing?

But now when two people start at a 150, one gets a private tutor and one does not, is it really smart to argue the one who could afford that $300/hour tutor has better reasoning skills?

Also, the difference between t-20 and t-10 schools is often two questions on a multiple choice, which could just mean a lucky day for someone and an unlucky day for someone else. Often the difference between NYU and Fordham can be choosing to retake vs. not choosing to retake, and there's no retaking 1L.

[dailygrind]

Anyone out there good enough with statistics to figure this one out? I'm sure as hell not.

So, word on the street is that your LSAT score has a 0.40 correlation to your law school grades. I have two questions about this.

1) What exactly does that mean?

2) Can it be used to more accurately gauge my "odds" of getting biglaw? For T14, considering that 10-15% of students do prestigious clerkships or PI and 45% of students get biglaw, essentially the top 60% will get biglaw if they want it, right? Say my LSAT is at the 80th percentile for my school. My "odds" for getting biglaw are higher than 60%, no? What would the statistical likelihood be?

I ask this because I am a splitter who will be looking at sticker for T14 if I get in. I would like to more accurately understand the financial risks involved in paying sticker for T14. Thanks for any help!

It's not just grades. Personality, previous work experience, hustle, and luck factor a lot into the mix, ie the median student at a t14 with no prior work experience and a mediocre personality is not necessarily going to "get biglaw if they want it." Besides that

Jesus christ. No, just no.

[kahechsof]

Agree with all above posters.

But, what do these numbers mean? What does a .32 correlation mean?

[shoeshine]

So far, I would compare my 1L experience to a competitive pinata competition. Some people are expending a ton of energy. There are some people with meticulously planned out strategies. There are others who seem to be completely unaware of what is going on.

The point is we are all wacking it in the dark. No one knows what to study or how to be the best. You can't assume anything you have done before this will lead to good grades.

[AffordablePrep]

So far, I would compare my 1L experience to a competitive pinata competition. Some people are expending a ton of energy. There are some people with meticulously planned out strategies. There are others who seem to be completely unaware of what is going on.

The point is we are all wacking it in the dark. No one knows what to study or how to be the best. You can't assume anything you have done before this will lead to good grades.

dating the professor will probably lead to good grades.

[WSJ_Law]

Incompetent statistical analysis is incompetent. Your "odds of getting biglaw" are best measured by NLJ. No way to gauge 1L performance because peer students are peer (everyone has similar #s). In sum, there has to be a bottom and that bottom is probably you. HTH



But seriously, when I say that bottom is probably you, I mean you need to be (usually) around top 33% to be competitive at lower T14 + Vandy/USC/UCLA + BC and better than that for most T25. HYSCCNBMVP is a different story but you're a splitter so unless you cured AIDS and resurrected JFK you're looking at DNCG on your best day.

[JoeFish]

As popularized by Mark Twain:
There are lies, damned lies, and statistics...

As has been said, saying I'm 80th percentile in LSAT so I have >60% chance at big law is generally no, just no, for a number of reasons