Woozy wrote:You need to stop trying to compare skills the LSAT tests for with skills required by a practicing lawyer. The only function of the LSAT is to predict law school grades - in particular 1L grades - and it does this better than any other known metric. Whether its predictive capabilities would be higher if the AR section were removed is an empirical question, and one the LSAC almost certainly has the answer to in its data sets. Since their main goal is to maximize the LSAT's correlation with 1L grades, and they continue to include an AR section, unless you come up with some empirical data to support your position, your arguments will remain unconvincing.
+1 F'Sho. This was succinct and addresses a fundamental misconception in the OP.
kill lsat wrote:In examining the posts thus far, it looks like the only people who have weighed in are those who committed to the LSAT long ago -- that appears to be why they come to this blog. People just aren't evaluating my arguments directly. So, "your arguments will [b]remain [/b]unconvincing" (emphasis added) is hardly a statement based on a realistic sample of LSAT takers or the prospective law school students who never get past the LSAT threshold.
Since no one is critically evaluating my arguments, let's look at yours.
As for your data, you might want to check into this because you're mortally wounding your position. The standard number LSAC reports is a correlation of LSAT performance to 1L grades of .4. (See LSAT apologist Frank Homer at --LinkRemoved--.) That's an abomination because it's just the "r" value. r-squared is only .16, which shows that only 16% of the change in one variable (LSAT score) is explained by a change in the other (1L grades). One of my sociology professors warned us about this trick in the SAT results decades ago. It looks like the LSAT people are conmen too. Essentially, according to LSAC, the LSAT is a terrible indicator. We could probably do a take on the Random Walk Down Wallstreet and find that throwing darts at law school applications is a better predictor of A, B, C or F.
So, by your argument, we ought to abolish the LSAT altogether, not just AR.
I would hope, however, that we can demonstrate statistically that the AR section is not helpful. I will take your suggestion and see if LSAC publishes information on correlations by section. However, since the overall LSAT correlations are laughable and the test results have already been contaminated by the preparation process students are forced to go through, the results will be tainted as well. They can't be any worse than .16, I presume. I'll report back with what I find or if someone else has this data, please present it.
This is terrible reasoning. The original statement was correct, in that if the goal of the LSAT is only to correlate as best as possible with 1L GPA, then the inclusion of AR and its effect on that correlation can only be empirically determined. Not only is the LSAC in the best position to have that data, they also have the greatest incentive to maximize that correlation.
You want the LSAT (and the AR in particular) to mean something that it doesn't, and your arguments assume that illusory distinction. Whether or not the LSAT should be abolished or not is a completely separate question and has no bearing on the efficacy of the AR section in achieving LSAC's testing goals.
I think you will greatly appreciate this video. It has been posted before, but it addresses many of your concerns: http://www.youtube.com/watch?v=_7_xHsce57c
Personally, I think it should be included in every thread that bashes on the LSAT.
PDaddy wrote:Therefore, until a new test is devised, taking the best aspects of the LSAT into account, schools should be given scores in a manner that prevents misuse. Give all scores within ranges. If you score between 151 and 160, that's the core you get: "150-159", ditto "120-129", "130-139", "140-149", "160-169", and "170-179", and "180" all alone.
You would effectively have one of only seven possible scores!
That will force admissions officers to take more account of other parts of EVERY applicant's profile.
In fact, if you read the first few pages of any PrepTest, they explain that there is an implied range associated with the scores (I believe they state a "Standard Error of Measurement" of 2.6).