December 2010 Curve Prediction Poll Forum

[androstan]

i'm sure some ppl were just joking but how can 20%+ of people reasonably think this test was easier than October? And not just themselves, believe that LSAC will set the curve to below a -12 ?

my guess is that the people putting 8, 9, 15 or 16 are messing around, people putting 10 are messing around or are bad guessers, and th people putting 11 are giving a serious pessimistic prediction.

I don't think the -15 people are necessarily messing around. The fact that it is a 102 question test is significant and opens up the reasonable possibility (but not high probability) that the curve could be -15.

I think that anything less than -11 (91 correct) or more than -15 (87 correct) is impossible, though. This will not be -16, despite how badly many of us wish it. Even if they toss out a question to take it to 101, I think -10 is impossible. If they tossed out two questions... I suppose -10 is a remote possibility.

[kkklick]

-14 is the best we can hope for, PT 59 had a pretty hard LG but not too bad, RC was terrible, but the real curve setter was LR, I've never gotten so many LR wrong on a PT before. On this test LR was a bit easier but still tough as nails, RC was much easier, LG harder. Still -13 seems the most likely scenario all things considered.

[robotclubmember]

idk what yuo people are talking about. this was definitely a -15 curve. the october test was a cinch and that was -12. this test had 102 questions for god's sake. plus the logic games were out of order re: difficulty? are you kidding me? and the games themselves were hard. granted it was only 23 question, but i know someone with a 170 in october who had to random guess on 6. LR was average really, but there were still 3 or 4 serious questions. I guarantee -15.

just curious... what are your prep test scores by section on average would you say? i'm trying to figure out why you keep coming on here and insisting there will be a -17, -16, or -15 curve, when that is absolutely absurd and unrealistic.

[bartleby]

idk what yuo people are talking about. this was definitely a -15 curve. the october test was a cinch and that was -12. this test had 102 questions for god's sake. plus the logic games were out of order re: difficulty? are you kidding me? and the games themselves were hard. granted it was only 23 question, but i know someone with a 170 in october who had to random guess on 6. LR was average really, but there were still 3 or 4 serious questions. I guarantee -15.

just curious... what are your prep test scores by section on average would you say? i'm trying to figure out why you keep coming on here and insisting there will be a -17, -16, or -15 curve, when that is absolutely absurd and unrealistic.

+1 - Temper your expectations. Please. At least for the sake of others.

[robotclubmember]

well hello there, I'm right there with you thinking that the games on December were doable, I finished them, but you have to consider that many of the posters here, just like you have seen basically every games section available, and even if a section matched your skill set better, you might not be representative of the whole. Given the reaction to this games section, I think it has to be in the top 10 most difficult when considered across the whole population of test takers. If high scorers didn't finish it, which many if not most did not based on TLS posters, and medium to low scorers didn't even make it to the third game (anecdotal evidence) then I would say on the whole you are looking at a very difficult section even if you disagree.

Also, you have only named one section so far, PT 57, that is more difficult than this one. I think you would have to stretch to name 4 more, much less 19 more. While none of the games on PT 62 individually were as rough as Dinos or Snakes and Lizards, the section as a whole was a time drain for most people and people are talking about the section as a whole being hard.

For realz.

I am not even so sure that PT57 was more difficult than PT62. PT57 was a slam dunk -0 for me. PT62, had to blind guess on two because of time constraints (also got a splitting headache right before PT62 LG... thank god LG was section 5 for me).

PT62 as a whole is easily in the top five hardest games sections. It's right up there in the league of new and used CD's and mauve dinosaurs.

[Sandro]

lets be real, some people get off by saying this test was easy and they are so good and everyone who is saying its hard must be low scorers and there are 25 harder LG sections (meaning this is one of the easiest ~15 or so PTs lol.

[RebelRebel]

Many factors to take into account - since there were 102 questions, will one or two be deleted? The curve in this case should focus on how many questions one needs to get right and not how many one can get wrong... a minus 12 here would still be less generous than a minus 11 curve when there are 100 questions.

I'm asking myself what's up with having 102 questions? Other than a tighter curve are they trying to add to the difficulty of the exam by including more questions? What will it be in a few years? 5 passages, 5 games and 30 questions for each section of LR?

If anything, they should focus on maintaining the same amount of questions while making them harder. Speed =/= intelligence.

In any case, in light of the horrible conference game and a few ridiculous RC questions, it's GRE prep time!

[androstan]

I'm asking myself what's up with having 102 questions? Other than a tighter curve are they trying to add to the difficulty of the exam by including more questions? What will it be in a few years? 5 passages, 5 games and 30 questions for each section of LR?

One hypothesis is that a lot of recent tests have "ungettable" scaled scores in the 170s. You know, the scaled score that has an asterisk in place of the raw score. By increasing the number of questions LSAC may be trying to get rid of this. It can't be a very good thing from a statistical standpoint. I believe that, in cases where there are more "ungettable" scaled scores, a person's scaled score is less tightly correlated with whatever LSAC is trying to measure.

Imagine, the best student gets hard question X correct. A student just a shade below the former in what-LSAC-is-trying-to-measure gets it wrong. The former student gets 175 and the latter gets 173. On the other hand, at a different place on the curve, a student gets moderate question Y correct and another student, a shade below him/her (just as much below as the 173er was below the 175er, gets it incorrect. The former gets 165 and the latter gets 164.

Theoretically, a 175er should be just as "superior" to a 174er as a 165er is to a 165er. But those "ungettable" scaled scores cause the finer distinctions between testers at the higher end of the curve to get washed out.

This is another reason I think the curve will be around -14. LSAC seems to like a -3 to be a 180. To maintain a good distinction between scorerers going down the curve they need another -1 at each scaled score. This would put -13 at 170, that would be good statistically and "psycometrically" I think. However, I believe this test was a bit on the hard side, especially because of the logic games. Also going -13 on a 102 question test is stronger than -13 on a 101 question test.

I think a -13 is more likely than a -15. -14 is my "official" guess.

[akikaze]

The LR on this exam was far more difficult than people are making it out to be, with two or three questions making very little sense in general; I expect a -14 or even -15 curve, nothing less.

[albusdumbledore]

Many factors to take into account - since there were 102 questions, will one or two be deleted? The curve in this case should focus on how many questions one needs to get right and not how many one can get wrong... a minus 12 here would still be less generous than a minus 11 curve when there are 100 questions.

I'm asking myself what's up with having 102 questions? Other than a tighter curve are they trying to add to the difficulty of the exam by including more questions? What will it be in a few years? 5 passages, 5 games and 30 questions for each section of LR?

If anything, they should focus on maintaining the same amount of questions while making them harder. Speed =/= intelligence.

In any case, in light of the horrible conference game and a few ridiculous RC questions, it's GRE prep time!

Where did this idea come from that LSAC just throws questions out for the sake of throwing questions out? The only time they ever do that is if there is some kind of legitimate controversy with one of the questions, and I don't see that happening this time. It was 102 questions, and it will stay that way.

[Sandro]

I'm sure the process of assembling a test and then equating it isnt too fun when after the test you have to go back and remove one question from scoring. I don't see a question being thrown out but who knows. -15 would be possible only with 102 Q IMO

[2014]

I don't think the increase from 100 or 101 to 102 questions is enough to justify a full extra point on the curve. A lot of it depends on what the extra question was though. If the test was already as hard as October for example after the equating sorcery happened and then an easy question was added on, the curve wouldn't change. If the question added happened to be one of the hardest questions though, perhaps it would matter.

I still feel like -13 is the eventual number with -3 for a 180.

-14 and -12 seem slightly less possible but still extremely plausible and outside of that 3 score variation, I personally don't see it going there.

[FlanSolo]

I'm sure the process of assembling a test and then equating it isnt too fun when after the test you have to go back and remove one question from scoring. I don't see a question being thrown out but who knows. -15 would be possible only with 102 Q IMO

Yeah, I'm not so sure why people think a question will be removed. LSAC definitely isn't the practice of releasing tests that they aren't sure are complete. Among other things (such as protecting their reputation), this practice would contradict the entire rationale for testing experimental sections in the first place.

Think about how this process works: LSAC runs the sections experimentally, at which point they rate the section against already "equalized" tests to see how scorers from different bands do and decide if the section accurately tests what they're looking for. LSAC would not intentionally release a question they had doubts about because they couldn't determine whether or not it "worked" by comparing to an already "equalized" test. The whole thing would be biased.

The fact is, it's an embarrassing thing for them to have to do -- each time they remove a question from scoring it calls into question their entire reason for being. There's just no way they go about this in a cavalier way.

As for the December test, I just don't see it. I think people had more basis for howling about the Shakespeare question on the October test (it was odd, if not difficult) and that was scored. Just because a question was hard doesn't mean it flawed. You're not less valuable as a person for missing it, people make mistakes all the time, especially on the LSAT!

[Sandro]

I don't think the increase from 100 or 101 to 102 questions is enough to justify a full extra point on the curve. A lot of it depends on what the extra question was though. If the test was already as hard as October for example after the equating sorcery happened and then an easy question was added on, the curve wouldn't change. If the question added happened to be one of the hardest questions though, perhaps it would matter.

I still feel like -13 is the eventual number with -3 for a 180.

-14 and -12 seem slightly less possible but still extremely plausible and outside of that 3 score variation, I personally don't see it going there.

an extra question/two I think would definitely effect the curve. Remember that the questions would be #26/27/28 and thus missed by a lot of people who might not get to them in time.....

[kkklick]

Many factors to take into account - since there were 102 questions, will one or two be deleted? The curve in this case should focus on how many questions one needs to get right and not how many one can get wrong... a minus 12 here would still be less generous than a minus 11 curve when there are 100 questions.

I'm asking myself what's up with having 102 questions? Other than a tighter curve are they trying to add to the difficulty of the exam by including more questions? What will it be in a few years? 5 passages, 5 games and 30 questions for each section of LR?

If anything, they should focus on maintaining the same amount of questions while making them harder. Speed =/= intelligence.

In any case, in light of the horrible conference game and a few ridiculous RC questions, it's GRE prep time!

Slippery slope flaw at its finest. But does it matter? I'll offer my take. LSAC has made 102 question tests in the 90's, and adds to my theory that they are transitioning back towards the older tests. It doesn't mean they are going to try to increase the difficulty by increasing section material. LR has a max of 26 questions, there were 26 on both. LG has a max of 24, there were 23. RC has a max of 28, there were 27.

[fosterp]

I wonder if they base their curve on the raw score total, or on number incorrect. You would think it doesn't matter but for some reason in my head I seem to think there is a difference...seeing as how we had 102 questions.

[58932ugahoige]

Actually, speed is a part of intelligence.

[2Serious4Numbers]

Many factors to take into account - since there were 102 questions, will one or two be deleted? The curve in this case should focus on how many questions one needs to get right and not how many one can get wrong... a minus 12 here would still be less generous than a minus 11 curve when there are 100 questions.

I'm asking myself what's up with having 102 questions? Other than a tighter curve are they trying to add to the difficulty of the exam by including more questions? What will it be in a few years? 5 passages, 5 games and 30 questions for each section of LR?

If anything, they should focus on maintaining the same amount of questions while making them harder. Speed =/= intelligence.

In any case, in light of the horrible conference game and a few ridiculous RC questions, it's GRE prep time!

Slippery slope flaw at its finest. But does it matter? I'll offer my take. LSAC has made 102 question tests in the 90's, and adds to my theory that they are transitioning back towards the older tests. It doesn't mean they are going to try to increase the difficulty by increasing section material. LR has a max of 26 questions, there were 26 on both. LG has a max of 24, there were 23. RC has a max of 28, there were 27.

this.

[redsoxfan2495]

I haven't been on here in a couple days. I'm disappointed to see that the curve consensus seems to have shifted for the worse. I still think -14 is reasonable though. This was a much harder test than October.

[ksinghal]

Yes, but basing curve predictions on this forum is risky. I mean, someone pointed out in another thread that the "best" test takers generally take June/Oct tests, not Dec. And the people who aren't freaking out about the test aren't likely to be on here, giving us an unrepresentative sample.

I'm curious, so I'll bite. I've been trawling through this forum since walking out of the test the other day.

I scored 170 on 58, 175 on Feb '10, and was hoping for 180 on 62. I took 61 as a PT and scored 180 (RC -1/LR -1). I didn't regret not taking it in October because I hadn't had the time to prep RC by the test date.

I've been tutoring LG for 2 years, so don't hate me that I'm sure I got -0 on the LG. I had the exp LG for Sec. 2 and the real LG for Sec. 5, and when I saw real games 2 and 3, I kept saying to myself, "wow, this is rough." I had 13 students take this test, so while I was certain they'd ace games 1 and 4, I felt bad when I saw the screwy hypo-laden ones.

I'm responding because I am one who is not freaking out about the test and would otherwise have refrained from posting. Several of my students, who are standard 160-ish 3.5-3.8ers, felt fine about it. They may have choked on stained glass, but at least half aren't worried overall because they hadn't prepped to such an extreme that the 2 screwy games caught them totally off-guard. They applied the skill sets they'd learned and did the best they can, and I think their scores will be where they expect them to be given the relative friendliness of the RC and what I considered average LR.

Personally, I thought it was overall a more difficult test than 61, although the RC was certainly easier than average. I stumbled on one RC and two LR, so whether or not I made my goal likely depends on whether it's a -2 or a -3, hence my interest. What concerns me is that I'm confident I got sunburns, aliens, and stars right, so I'm thinking I just overthought 2 of the "easier" LRs and wrote off the right answers as too extreme or what have you.

Anyhow, best of luck to everyone.

[fosterp]

moststronglysupported.com is calling this a -12, though the person that wrote it didn't actually take it, just based off feedback from "some group of people." Basically it was nothing special except a harder than average games section, which has been the trend the last 5 tests or so.

The possibility that the representation on this forum might be "off" than normal but the predictions for dec09 were actually under by 2 points (avg prediction was -12 and it turned out -14). So who knows.

I am betting for those that are 175+ constantly probably maintained their top scores despite the harder LG - they just finished with one minute left instead of the usual ten lol. Those who PT'd the high 160s low 170s probably felt the worst about the LG section, as usually the low 170s means you did pretty well on every section and didn't really "blow" any of them, however on this test the 3rd game was such a headache, unless you were an absolute pro at LG you probably scored below average on that section. And its common to rate the test as a whole based on how you fared in the worst section. Those in the mid to low 160s probably didn't feel the burn as well since they are probably accustomed to losing a lot of points on one section already.

Just my guess.

[r2b2ct]

It seems like most people on this site tend to underestimate the curves a little. In October I saw a lot of predictions of -10 and -11, yet it turned out -12. And as fosterp said, Dec09 predictions were similarly off.

I felt like this test was much more difficult than Oct (I took both as real tests), so based on this I can't believe anything tighter than -12. My guess is -14.

[RebelRebel]

Actually, speed is a part of intelligence.

Yes and no. Speed is another factor that they can use to standardize test scores. Being able to process information fast does not necessarily equate with being able to process more complex information.

Anyhow, no point in arguing since this is a timed standardized test where all questions independently of the level of difficulty are accorded the same value so I will accept that ON THE LSAT speed can be somewhat correlated with intelligence.

[Sandro]

It seems like most people on this site tend to underestimate the curves a little. In October I saw a lot of predictions of -10 and -11, yet it turned out -12. And as fosterp said, Dec09 predictions were similarly off.

I felt like this test was much more difficult than Oct (I took both as real tests), so based on this I can't believe anything tighter than -12. My guess is -14.

bingo. go look at the Dec09 prediction thread, -14 was only predicted by a small portion of TLS. I think some people need to realize that if you are going to estimate the curve you cant estimate it off of how YOU felt after the test - theres thousands of other people taking the test that effect the curve. IMO LSAC curves it off of absolute difficulty, not curving it to what 170 scorers fared. but thats just a guess.

[AshtonB]

I'm sure the process of assembling a test and then equating it isnt too fun when after the test you have to go back and remove one question from scoring. I don't see a question being thrown out but who knows. -15 would be possible only with 102 Q IMO

Yeah, I'm not so sure why people think a question will be removed. LSAC definitely isn't the practice of releasing tests that they aren't sure are complete. Among other things (such as protecting their reputation), this practice would contradict the entire rationale for testing experimental sections in the first place.

Think about how this process works: LSAC runs the sections experimentally, at which point they rate the section against already "equalized" tests to see how scorers from different bands do and decide if the section accurately tests what they're looking for. LSAC would not intentionally release a question they had doubts about because they couldn't determine whether or not it "worked" by comparing to an already "equalized" test. The whole thing would be biased.

The fact is, it's an embarrassing thing for them to have to do -- each time they remove a question from scoring it calls into question their entire reason for being. There's just no way they go about this in a cavalier way.

As for the December test, I just don't see it. I think people had more basis for howling about the Shakespeare question on the October test (it was odd, if not difficult) and that was scored. Just because a question was hard doesn't mean it flawed. You're not less valuable as a person for missing it, people make mistakes all the time, especially on the LSAT!

That question was very odd (as was the chef/virus one), but not particularly challenging. If you use basic logic, it's fairly simple to deduce the correct response. Ditto for the chef/virus question.