Why is there a big discrepancy between curves?

jblev1
Posts: 41
Joined: Mon Dec 27, 2010 9:55 am

Why is there a big discrepancy between curves?

Postby jblev1 » Mon Dec 27, 2010 10:01 am

From one month to the next? Does LSAC do this on purpose to give people who think differently better options? Like if you are a good test taker and can do the hard problems you will obviously score much higher on the higher curve / harder tests, but if you are a weaker test taker the easy question test is for you? And isn't the easy month October, the most common time to take? Kind of says that unusual students or takers are penalized to some degree, because a few question curve on a much harder exam is not equal.

LSAT Stats: http://www.youtube.com/watch?v=fburjZR6c4A

delusional
Posts: 1190
Joined: Thu Jul 15, 2010 7:57 pm

Re: Why is there a big discrepancy between curves?

Postby delusional » Mon Dec 27, 2010 10:44 am

jblev1 wrote:From one month to the next? Does LSAC do this on purpose to give people who think differently better options? Like if you are a good test taker and can do the hard problems you will obviously score much higher on the higher curve / harder tests, but if you are a weaker test taker the easy question test is for you? And isn't the easy month October, the most common time to take? Kind of says that unusual students or takers are penalized to some degree, because a few question curve on a much harder exam is not equal.

LSAT Stats: http://www.youtube.com/watch?v=fburjZR6c4A


The way I understand it, the curve, actually the equating, is also tailored to the particular level of test takers. If there are question that affect the spectrum of abilities differently, they affect the curve differently on different levels. When they experiment with the question, they see not only how many people got it right/wrong, but also how those testers fall on the percentile rating.

I strongly disbelieve that the curve is as good as they think it is - how accurately can it be possible to predict the effect of placement of a logic game? But on the other hand, there is almost no curve-question that I have been able to muster that LSAC does not have at least an official answer to.

User avatar
kkklick
Posts: 1021
Joined: Mon Jun 07, 2010 8:33 pm

Re: Why is there a big discrepancy between curves?

Postby kkklick » Mon Dec 27, 2010 11:27 am

delusional wrote:
jblev1 wrote:From one month to the next? Does LSAC do this on purpose to give people who think differently better options? Like if you are a good test taker and can do the hard problems you will obviously score much higher on the higher curve / harder tests, but if you are a weaker test taker the easy question test is for you? And isn't the easy month October, the most common time to take? Kind of says that unusual students or takers are penalized to some degree, because a few question curve on a much harder exam is not equal.

LSAT Stats: http://www.youtube.com/watch?v=fburjZR6c4A


The way I understand it, the curve, actually the equating, is also tailored to the particular level of test takers. If there are question that affect the spectrum of abilities differently, they affect the curve differently on different levels. When they experiment with the question, they see not only how many people got it right/wrong, but also how those testers fall on the percentile rating.

I strongly disbelieve that the curve is as good as they think it is - how accurately can it be possible to predict the effect of placement of a logic game? But on the other hand, there is almost no curve-question that I have been able to muster that LSAC does not have at least an official answer to.

It is actually my understanding that when the sections are given as experimentals, they are structured in the exact same way as we will see them on a future exam. So our LG section from the December test had the same 4 games, 2 of which had no space, and the last one being the easiest. Of course this hasn't been definitively confirmed but there is good reason to believe that this is the case.

delusional
Posts: 1190
Joined: Thu Jul 15, 2010 7:57 pm

Re: Why is there a big discrepancy between curves?

Postby delusional » Mon Dec 27, 2010 11:32 am

kkklick wrote:
delusional wrote:
jblev1 wrote:From one month to the next? Does LSAC do this on purpose to give people who think differently better options? Like if you are a good test taker and can do the hard problems you will obviously score much higher on the higher curve / harder tests, but if you are a weaker test taker the easy question test is for you? And isn't the easy month October, the most common time to take? Kind of says that unusual students or takers are penalized to some degree, because a few question curve on a much harder exam is not equal.

LSAT Stats: http://www.youtube.com/watch?v=fburjZR6c4A


The way I understand it, the curve, actually the equating, is also tailored to the particular level of test takers. If there are question that affect the spectrum of abilities differently, they affect the curve differently on different levels. When they experiment with the question, they see not only how many people got it right/wrong, but also how those testers fall on the percentile rating.

I strongly disbelieve that the curve is as good as they think it is - how accurately can it be possible to predict the effect of placement of a logic game? But on the other hand, there is almost no curve-question that I have been able to muster that LSAC does not have at least an official answer to.

It is actually my understanding that when the sections are given as experimentals, they are structured in the exact same way as we will see them on a future exam. So our LG section from the December test had the same 4 games, 2 of which had no space, and the last one being the easiest. Of course this hasn't been definitively confirmed but there is good reason to believe that this is the case.

It could be that that's the ideal, but what happens when the experimental proves that certain questions/games/passages need to be eliminated? What about LR questions?

Also, they do have lots of statistical stuff that is supposed to explain how they position the questions, which would be relatively unnecessary if they could just test entire sections.

User avatar
KevinP
Posts: 1324
Joined: Sat Sep 26, 2009 8:56 pm

Re: Why is there a big discrepancy between curves?

Postby KevinP » Mon Dec 27, 2010 11:33 am

jblev1 wrote:From one month to the next? Does LSAC do this on purpose to give people who think differently better options? Like if you are a good test taker and can do the hard problems you will obviously score much higher on the higher curve / harder tests, but if you are a weaker test taker the easy question test is for you? And isn't the easy month October, the most common time to take? Kind of says that unusual students or takers are penalized to some degree, because a few question curve on a much harder exam is not equal.

LSAT Stats: http://www.youtube.com/watch?v=fburjZR6c4A


I think a person would be much more at a disadvantage depending on how the section is arranged and what type of experimental section was given more so than the curve. LSAC does a really good job equating tests.*

*This is not counting Dec..... I think LSAC was just trying to be mean.

And I have no idea what the youtube vid has to do w/ anything.

User avatar
kkklick
Posts: 1021
Joined: Mon Jun 07, 2010 8:33 pm

Re: Why is there a big discrepancy between curves?

Postby kkklick » Mon Dec 27, 2010 11:44 am

LR is one of those sections that questions of various difficulty require roughly the same amount of time. A difficult 4 question may need a minute and a half, so when eliminating a Q that was set as a 4 star difficulty, they may replace it with another question that would take the same amount of time. LR doesn't go by the same rules as RC and LG.

User avatar
KevinP
Posts: 1324
Joined: Sat Sep 26, 2009 8:56 pm

Re: Why is there a big discrepancy between curves?

Postby KevinP » Mon Dec 27, 2010 11:48 am

kkklick wrote:LR is one of those sections that questions of various difficulty require roughly the same amount of time. A difficult 4 question may need a minute and a half, so when eliminating a Q that was set as a 4 star difficulty, they may replace it with another question that would take the same amount of time. LR doesn't go by the same rules as RC and LG.


Exactly. I read the Dec. 09 thread and most of the people said the LR was easy and underestimated the curve by a lot. I'm usually good at predicting LR difficulty and I can seriously say the first LR section was similar to Dec. 09 and had a lot of tricks in it. The second LR section was easy.

User avatar
kkklick
Posts: 1021
Joined: Mon Jun 07, 2010 8:33 pm

Re: Why is there a big discrepancy between curves?

Postby kkklick » Mon Dec 27, 2010 11:52 am

KevinP wrote:
kkklick wrote:LR is one of those sections that questions of various difficulty require roughly the same amount of time. A difficult 4 question may need a minute and a half, so when eliminating a Q that was set as a 4 star difficulty, they may replace it with another question that would take the same amount of time. LR doesn't go by the same rules as RC and LG.


Exactly. I read the Dec. 09 thread and most of the people said the LR was easy and underestimated the curve by a lot. I'm usually good at predicting LR difficulty and I can seriously say the first LR section was similar to Dec. 09 and had a lot of tricks in it. The second LR section was easy.

I felt the same as well, the LR with the babies/dictionary definitions had a LOT of similarities (in terms of difficulty) to the LR in December (babies/ladders). The 2nd LR was easy, I think we will see a ton of,

"I got -1 on the 2nd LR!!!"
"What about the first"
".........................."

User avatar
robotclubmember
Posts: 743
Joined: Wed Sep 15, 2010 7:53 am

Re: Why is there a big discrepancy between curves?

Postby robotclubmember » Mon Dec 27, 2010 12:03 pm



LSAT Stats? O RLY???? Cool story bro.

Thanks for the spam.

delusional
Posts: 1190
Joined: Thu Jul 15, 2010 7:57 pm

Re: Why is there a big discrepancy between curves?

Postby delusional » Mon Dec 27, 2010 1:01 pm

robotclubmember wrote:


LSAT Stats? O RLY???? Cool story bro.

Thanks for the spam.

Yeah, you're right. Well, who here likes a good conversation about how the curve is set up? /buzz killington.




Return to “LSAT Prep and Discussion Forum”

Who is online

Users browsing this forum: 34iplaw, Alexandros, earldasquirel, Greenteachurro, The_Pluviophile and 9 guests