The LSAT Curve Forum

[mewalke1]

Hi Guys Im new to the forum. I tried to avoid them until after I took the LSAt because they were stressing me out.

Anyway can someone explain how the lsat curve works? I thought that it wasn't curved but "equated"

I see a lot of numbers like -9,-10 etc

Can someone fill me in on what this means?

Thanks

[TheRainMan]

Hi Guys Im new to the forum. I tried to avoid them until after I took the LSAt because they were stressing me out.

Anyway can someone explain how the lsat curve works? I thought that it wasn't curved but "equated"

I see a lot of numbers like -9,-10 etc

Can someone fill me in on what this means?

Thanks

The LSAT curve to which people often refer is the number of non-credited (incorrect) responses one can have on any given LSAT to receive a score of 170. I would say tomato-tomato here wrt to equate vs curve, but I think it loses some of its meaning when not said aloud or written phonetically.

[mewalke1]

Awesome thanks

[Ded Precedent]

If I understand it right, the curve is based off the experimental sections of previous tests and then compiled in such a way that each test should have a similar range of takers falling into each of the score categories. So, for the most part the curve has already been determined before you sit down for the test, which is different from what we would usually refer to as a curve, which would be based of the number of correct answers and how that falls into percentiles on the same exact test taken.

I just realized I wrote that in a really inarticulate and confusing way, does that make any sense to you?

[LSAT Blog]

+1 to both previous responses.

I actually wrote a series of posts to explain the "curve" a while back. Here's the first:

http://lsatblog.blogspot.com/2010/02/ls ... -lsac.html

[mewalke1]

+1 to both previous responses.

I actually wrote a series of posts to explain the "curve" a while back. Here's the first:

http://lsatblog.blogspot.com/2010/02/ls ... -lsac.html

I read that post good stuff

[sjwest]

It is a good read. I have a quick supplemental question. Since score distribution is set beforehand, do you think there is anything they can do to take into account changes that were not previously tested? I'm referring specifically to the fact that the games section was spread over 2 pages this past exam.

I'm expecting the answer to be no, they're not going to do anything about it. However, I've heard a few people mention that they were going to write in about it. Figured it couldn't hurt to ask.

[TheRainMan]

It is a good read. I have a quick supplemental question. Since score distribution is set beforehand, do you think there is anything they can do to take into account changes that were not previously tested? I'm referring specifically to the fact that the games section was spread over 2 pages this past exam.

I'm expecting the answer to be no, they're not going to do anything about it. However, I've heard a few people mention that they were going to write in about it. Figured it couldn't hurt to ask.

If they were expecting an average of 16/22 but they instead got 13/22 they will likely adjust the curve accordingly. The score distribution is not set in stone beforehand, but instead they merely predict what they think it will be. LSAC spends these 3 weeks between testing and score release going over shit like this. The fact that the format was different will not change anything unless there was also a significant deviation in scoring.

[Jeffort]

If the previous two posts where the phrase 'score distribution' is used was meant to refer to the percentile ranking of each achieved scaled score, you two misunderstood the major points in Steves articles and the information about how the test form raw points to scaled score conversion charts are created, and how it is an equated, not a curved test.

The percentile rank of each scaled score is a natural byproduct of the performance of test takers, it is not forced. If it was a forced score distribution to insure that only a certain percentage of people score above each given scaled score, that would mean it is graded on a curve, which it is not. The percentile rank of each achieved scaled score fluctuates slightly year after year depending on how the population of test takers performed over the previous three years of administered tests.

170 used to be a 99th percentile score and was so for roughly at least 15+ years. Currently it is 97.xxth percentile. 150 held strong at being roughly the 50th percentile for a huge amount of time, currently it is ~44.xxth percentile.

Theoretically, if only a bunch of LSAT superstars take a particular administration and they all rock it, they would all get 99th percentile scores in the 173+ range and nobody would get a reported score below that. Of course that would never happen in reality, but it is possible with how the test is designed and scored.

LSAC doesn't force some people to end up with scores in the 120s range every administration, but there are always at least few 'oh that person' that score a perfect 120 per administration and many others that get reported scores in the 120s range for whatever various personal reasons on test day. Same thing with all other possible scaled scores.

Each achieved scaled score is meant to represent a particular ability level the test taker performed at on test day that is comparable to everyone else that took the test in the previous three years.

[mewalke1]

Interesting.


Does LSAC release all the scores for public viewing? We will be able to tell how many 163s or 128s were scored format he june test?