Splitters versus higher averages--and LSP. Forum

[Scuppers]

My apologies if this is a stupid and/or frequently posted question. I did browse the forums for a while and did not see a similar post.

At what point does a stronger LSAT/GPA combination that is still below both medians beat a splitter who is slightly above one median but way below the other one? Never?

Toying around with LSP it seems that 1 LSAT is approximately .09 GPA. (I understand that this is an approximation and also that some schools favor either LSAT or GPA so this isn't exact, but it's close enough for hypotheticals.)

Using UCLA as an example, my odds according to LSP are 43% (3.65 167) despite the conventional TLS wisdom that being below both medians is basically auto ding. (I think that's the conventional wisdom anyway, not trying to straw man here.)

LSP results: (All examples are non-URM, non-ED.)

UCLA medians = 168 LSAT 3.78 GPA

#1 3.65 167 = 43%

#2 3.47 169 = 43%
#3 3.29 171 = 43%
#4 3.11 173 = 43%
#5 3.83 165 = 43%

So 1 LSAT = .09 GPA. But is the first one at a significant disadvantage considering that it's below both medians? Yes? Right, now consider these:

#1 3.73 167 = 48%
#2 3.00 170 = 18%

The first seems much better, but it's also still below both medians. Does #2 win? Where is the line? Does LSP just fail in that it doesn't take account of one's relation to the median?

[ThomasMN]

I am going to cut this short, look at law school numbers. Most people that get into a law school below both medians are special cases ( URM, crazy life story, etc.). Just look at the acceptance graph for this year.

[Br3v]

Also have to take I to account gpa/LSAT floors

[Tiago Splitter]

http://ucla.lawschoolnumbers.com/applic ... ,8&type=jd

These are people below both medians. Odds aren't too good.

To your question, yes, LSP isn't build to understand how important being over one median is.

[bernaldiaz]

The most interesting part of this thread is the whole 1 LSAT point = .09 GPA. I've been wondering what it would be.

[Scuppers]

I am going to cut this short, look at law school numbers. Most people that get into a law school below both medians are special cases ( URM, crazy life story, etc.). Just look at the acceptance graph for this year.

There is a field for URM. And LSN data is used in the predictor, so my questioning of LSP and your suggestion to look at LSN seems a little...

[Br3v]

I am going to cut this short, look at law school numbers. Most people that get into a law school below both medians are special cases ( URM, crazy life story, etc.). Just look at the acceptance graph for this year.

There is a field for URM. And LSN data is used in the predictor, so my questioning of LSP and your suggestion to look at LSN seems a little...

Am I the only one who can't figure out what comes after the ... ?

[Scuppers]

To your question, yes, LSP isn't build to understand how important being over one median is.

#1 3.73 167 = 48% Below both medians.
#2 3.00 170 = 18% Below one median.

So you pick #2 over #1, correct?

[Scuppers]

I am going to cut this short, look at law school numbers. Most people that get into a law school below both medians are special cases ( URM, crazy life story, etc.). Just look at the acceptance graph for this year.

There is a field for URM. And LSN data is used in the predictor, so my questioning of LSP and your suggestion to look at LSN seems a little...

Am I the only one who can't figure out what comes after the ... ?

Got you thinking. Interactivity through ambiguity.

[t14fanboy]

To your question, yes, LSP isn't build to understand how important being over one median is.

#1 3.73 167 = 48% Below both medians.
#2 3.00 170 = 18% Below one median.

So you pick #2 over #1, correct?

Except I swear that #1 would have the better shot at Cornell.

[Tiago Splitter]

To your question, yes, LSP isn't build to understand how important being over one median is.

#1 3.73 167 = 48% Below both medians.
#2 3.00 170 = 18% Below one median.

So you pick #2 over #1, correct?

Yes. I'd put the odds for both at below 18% though.

[Samara]

To your question, yes, LSP isn't build to understand how important being over one median is.

#1 3.73 167 = 48% Below both medians.
#2 3.00 170 = 18% Below one median.

So you pick #2 over #1, correct?

Yes. I'd put the odds for both at below 18% though.

FWIW, the LSP percentages aren't odds of admissions. As to which scenario is more likely to receive admission, it depends on the splitter-friendliness of the school. 3.0/170 is going to have a much better chance at UVa than Berkeley.

[jkpolk]

The most interesting part of this thread is the whole 1 LSAT point = .09 GPA. I've been wondering what it would be.

When someone figures out a useful conversion based on a regression analysis of LSN, controlling for all the confounding variables and supported by an eyeball test of all admission information, color me surprised if 1 LSAT point = .09 GPA

[Tiago Splitter]

The most interesting part of this thread is the whole 1 LSAT point = .09 GPA. I've been wondering what it would be.

When someone figures out a useful conversion based on a regression analysis of LSN, controlling for all the confounding variables and supported by an eyeball test of all admission information, color me surprised if 1 LSAT point = .09 GPA

That works with UCLA's index, but for all applicants there really isn't ever going to be a neat formula like that.

[Samara]

The most interesting part of this thread is the whole 1 LSAT point = .09 GPA. I've been wondering what it would be.

When someone figures out a useful conversion based on a regression analysis of LSN, controlling for all the confounding variables and supported by an eyeball test of all admission information, color me surprised if 1 LSAT point = .09 GPA

That works with UCLA's index, but for all applicants there really isn't ever going to be a neat formula like that.

Yeah, you can't really do a "conversion" with much applicability. Even the schools that are pretty balanced between GPA and LSAT have GPA floors and possibly LSAT floors. Then, there's also the problem of diminishing returns. Is there really an appreciable difference between an applicant's chances at UCLA with a 3.6/171 and a 3.6/174? Or even 3.6/169 and 3.6/170? It would affect money, but probably not admission chances.

[Br3v]

Problem you can't do simple conversion. I would imagine the ratio is on a inversely sliding scale basis. So to take the ratio you'd have to use calculus.

[Tiago Splitter]

Problem you can't do simple conversion. I would imagine the ratio is on a inversely sliding scale basis. So to take the ratio you'd have to use calculus.

[bernaldiaz]

The most interesting part of this thread is the whole 1 LSAT point = .09 GPA. I've been wondering what it would be.

When someone figures out a useful conversion based on a regression analysis of LSN, controlling for all the confounding variables and supported by an eyeball test of all admission information, color me surprised if 1 LSAT point = .09 GPA

What do you think it would be, roughly? That seems about right to me.