curve question Forum

[chrisiQ]

stupid question buy maybe someone can finally explain it to me . . . how is the curve determined and is it set so -(whaterever) still gives you what score?

[SchopenhauerFTW]

stupid question buy maybe someone can finally explain it to me . . . how is the curve determined and is it set so -(whaterever) still gives you what score?

-(whatever) means the maximum number of questions one can miss on the LSAT and still get a 170.

The curve is determined by...uh...well I've read so many conflicting explanations on TLS that it's hard to give a straight answer.
This might help: http://lsatblog.blogspot.com/2010/02/ls ... -lsac.html

[hutchesonian]

Why 170? Is it a major cutoff point? Or just a nicer number than 172/173 (99th percentile)?

[tourdeforcex]

Why 170? Is it a major cutoff point? Or just a nicer number than 172/173 (99th percentile)?

from what i've learned from TLS, 170 seems to be the make or break point for those w/ a suboptimal GPA (below median at desired school). and maybe 170 might be needed for most people trying to get into T14

[Jeffort]

Technically speaking, the LSAT is not a curved test in the way that term is used to describe the way many UG class grading systems work.

It is designed to accurately and reliably measure each persons ability/skill level with the mental abilities the LSAT is designed to test (logical reasoning, reading comprehension & analytical/deductive reasoning) and rate them on a standard scale (120-180) that is stable. That means that each possible score is comparable and represents the same performance/ability level no matter which test administration it is from. 170 from one test administration means the same thing as a 170 from a different test.

The differences in the raw to scaled score conversion charts with the tests is meant to adjust for the overall difficulty rating of the test form so that 170/etc. always means 170/etc. performance.

The curve of achieved scores, meaning the percentile ranking chart for reported scores is not a forced curve, it is the byproduct of the people that take the test. It's just the data of how people actually did in comparison to the entire 3 year test taker population. LSAC doesn't force anybody to have to get a 120's range score but people get them every administration.

[3ThrowAway99]

170 from one test administration means the same thing as a 170 from a different test.

Technically or ideally I agree this is the case, but individual tests may have emphasis that, although statistically perfectly equated (or near perfectly equated) with other tests in terms of scaled scores, may on the level of an individual person yield a different score due to what was emphasized in the test in relation to the individuals' skill set (apart from the obvious test-to-test differential due to nerves, sleep etc and not skill set). Although the tests all have the same format in term of LR LR LG RC in some combination, I think there is no question (for example) that in some tests the RC will yield a higher-rate of misses than RC for another test: this will no doubt be factored into the overall 'curve' of the test, but the fact that the overall score distribution remains static won't reflect the likely score differential for someone who struggles with a hard RC section disproportionately to how they may struggle with hard LG section (for example). In my experience different tests do emphasize different sections as being relatively harder or easier than average across tests, whether this is intentional by the makers or not.

I think this logic also applies to sections that are overall not more difficult statistically when compared to other similar sections on other tests (in fact this phenomenon may apply more to these situations): an example of this is where a person has a disproportionate weakness (disproportionate in regard to his or her general aptitude for that section) with a particular type of question stem in LR that happens to come up a ton on a particular test. The tests are equated so that technically speaking (at group level) 170= perfectly 170, but at individual level I think there is no question that a particular individual may find it harder to score that 170 on one particular LSAT than on another.