Law School Discussion

Why scaling/bell curving the LSAT is a flawed concept

Julie Fern

  • *****
  • 25797
  • hillary clinton say "boo!"
    • View Profile
Re: Why scaling/bell curving the LSAT is a flawed concept
« Reply #20 on: July 05, 2007, 03:48:16 PM »
note also lsat has rather hefty margin of error, which relate to confidence level.


????


I guess in the sense that if you're prepared and/or tend to do well on the exam, you feel more confident.

The only real "margin of error", in my mind, would tend to apply to guessing correctly/incorrectly.

confidence level statistical measure, not emotional, dipshit. guess they not teach you that at testmaster.


Maybe you should speak more clearly, fucktard. Guess they don't teach you that in special ed. :)

pearls before swine, numbnuts.

and funny how you only one not understand julie.

Julie Fern

  • *****
  • 25797
  • hillary clinton say "boo!"
    • View Profile
Re: Why scaling/bell curving the LSAT is a flawed concept
« Reply #21 on: July 05, 2007, 03:51:03 PM »

I've never seen julie string togeher a coherent thought on here, so I think she was probably referring to emotional confidence. However, it's nice of you to recharacterize it in a thoughtful manner.

all dickwads have their excuses for failure.

hey, maybe you can invent new excuse for dropping your fictitious lsat score on us again!

Julie Fern

  • *****
  • 25797
  • hillary clinton say "boo!"
    • View Profile
Re: Why scaling/bell curving the LSAT is a flawed concept
« Reply #22 on: July 05, 2007, 03:53:28 PM »
With Julie we at least know for sure that she thinks GW is a bad bad silly stupid man.
(not implying I disagree with that thought. I'll reserve political rants for the OTB but it sure seems like Scooter is getting a free pass to continue scooting along :D)


"gw"? at least call him by proper name: gump.


Forrest Gump was a retarded southerner of limited intellect who was extremely fortunate in life, and fell into all sorts of wonderful situations through no effort of his own. As such, this would seem to apply more to our previous president than our current one, who has been relatively unfortunate in having to clean up the messes left by the previous one.

However, as noted, such discussions belong on the OTB, not here.

ah, now we see your real problem with life, you dumb bastard diehard republican wingnut.

by way, people still like bill clinton.  a lot.

ě

  • ****
  • 4497
  • non sequitur
    • View Profile
Re: Why scaling/bell curving the LSAT is a flawed concept
« Reply #23 on: July 05, 2007, 03:55:34 PM »
Bush cleaning up after Clinton? Good one.

Julie Fern

  • *****
  • 25797
  • hillary clinton say "boo!"
    • View Profile
Re: Why scaling/bell curving the LSAT is a flawed concept
« Reply #24 on: July 05, 2007, 03:58:13 PM »
With Julie we at least know for sure that she thinks GW is a bad bad silly stupid man.
(not implying I disagree with that thought. I'll reserve political rants for the OTB but it sure seems like Scooter is getting a free pass to continue scooting along :D)


"gw"? at least call him by proper name: gump.


Forrest Gump was a retarded southerner of limited intellect who was extremely fortunate in life, and fell into all sorts of wonderful situations through no effort of his own. As such, this would seem to apply more to our previous president than our current one, who has been relatively unfortunate in having to clean up the messes left by the previous one.

However, as noted, such discussions belong on the OTB, not here.

LOL

wow, einstein, you start alternate account for this?

julie always happy to see all you proud republicans out there...those few of you left, that is.  have fun going rest of way over cliff with gump!

Julie Fern

  • *****
  • 25797
  • hillary clinton say "boo!"
    • View Profile
Re: Why scaling/bell curving the LSAT is a flawed concept
« Reply #25 on: July 05, 2007, 04:00:25 PM »
Bush cleaning up after Clinton? Good one.

right-wing assholes must cling to their myths.

Julie Fern

  • *****
  • 25797
  • hillary clinton say "boo!"
    • View Profile
Re: Why scaling/bell curving the LSAT is a flawed concept
« Reply #26 on: July 05, 2007, 04:01:30 PM »
you welcome.

Re: Why scaling/bell curving the LSAT is a flawed concept
« Reply #27 on: July 05, 2007, 04:02:35 PM »

I won't bother asking for proof that all people practicing for the LSATs have consistent raw scores - but I'm baffled by your last sentence there. A test has hard questions and a test has easy questions (both true), therefore the average difficulty of the questions is the same across all LSATs?

Do this - go to the LR section on the June 07 test. Find three questions that are harder than average and three questions that are easier than average. Now - envision an exam where the three easier questions are replaced with the three harder questions - tada, you have a logically harder exam. In the aggregate, people will perform differently (worse) on this new exam and you need a scale to equate performances from the two exams.



I don't buy any of this. First, just like your hard and easy questions above, these things balance out. If someone makes a random guess, they have a 1/5 chance of being correct. Now, some individual people may get 35% of their random guesses correct while others may get only 5%, but I'd bet that across the entire population, when people randomly guess, they are correct 1/5 times, and this does not change across administrations.

The difference is very obvious. The LSAT writers(people who make the LSAT) do their best to try and make the LSAT as fair and alike to previous LSATS as possible. This is a variable that drives my "averaging" theory.

While it is true that over the course of all the LSATS taken ever the raw scores of the randoms who come unprepared(who I believe drive the LSAT scale just to refresh memories) will average out, there exists a possibility for each smaller sample(each LSAT test) to have fluctuations in accuracy from this group. Again, the difference is that the LSAT writers go for an average result, alike to previous ones. LSAT test takers are going for an extreme, not average, result(extreme meaning a perfect score....since it is everyones goal to get every question they answer correct.)

ě

  • ****
  • 4497
  • non sequitur
    • View Profile
Re: Why scaling/bell curving the LSAT is a flawed concept
« Reply #28 on: July 05, 2007, 04:14:43 PM »
I was going to say "quit while you're ahead". Now, you're obviously far from ahead, but I'd still suggest giving up.

Re: Why scaling/bell curving the LSAT is a flawed concept
« Reply #29 on: July 05, 2007, 04:41:53 PM »

I won't bother asking for proof that all people practicing for the LSATs have consistent raw scores - but I'm baffled by your last sentence there. A test has hard questions and a test has easy questions (both true), therefore the average difficulty of the questions is the same across all LSATs?

Do this - go to the LR section on the June 07 test. Find three questions that are harder than average and three questions that are easier than average. Now - envision an exam where the three easier questions are replaced with the three harder questions - tada, you have a logically harder exam. In the aggregate, people will perform differently (worse) on this new exam and you need a scale to equate performances from the two exams.



I don't buy any of this. First, just like your hard and easy questions above, these things balance out. If someone makes a random guess, they have a 1/5 chance of being correct. Now, some individual people may get 35% of their random guesses correct while others may get only 5%, but I'd bet that across the entire population, when people randomly guess, they are correct 1/5 times, and this does not change across administrations.

The difference is very obvious. The LSAT writers(people who make the LSAT) do their best to try and make the LSAT as fair and alike to previous LSATS as possible. This is a variable that drives my "averaging" theory.

While it is true that over the course of all the LSATS taken ever the raw scores of the randoms who come unprepared(who I believe drive the LSAT scale just to refresh memories) will average out, there exists a possibility for each smaller sample(each LSAT test) to have fluctuations in accuracy from this group. Again, the difference is that the LSAT writers go for an average result, alike to previous ones. LSAT test takers are going for an extreme, not average, result(extreme meaning a perfect score....since it is everyones goal to get every question they answer correct.)

In order for any of this to be valid (and I really doubt any of it is, you'd actually have to produce some evidence that, after administering a test, LSAC bureaucrats gather around  a spreadsheet somewhere and decide how to scale the test so that the ratios of test takers receiving a given score is consistent with the corresponding percentile rank of that score historically.

Yes, there is a greater possibility of statistical fluctuation on one test than there is on, say 100 tests. This is akin to flipping a coin - the more times one flips the coin, the closer the distribution should get to 50/50.

However, don't you think the LSAC people already know this, and would therefore not tweak the scale to adjust for the test-to-test fluctuation in "guesser" performance?