LSAT and IQ
03/27/2009
A+
|
a-
A continuing theme here is the distinction between acceptable public discourse about IQ (it's meaningless, it's biased, it's evil, it's useless) and private behavior (it's important). For example, consider law school admissions, which are a key part of the American power structure. Top law schools place a strong weight on one's test score on the Law School Admissions Test (LSAT). I never took that test, but a quick glance, it strikes me as a very good test of verbal IQ.

Via the interesting quant blog Le Carrefour de la Sagesse, I found the Law School Admissions Council report "LSAT Performance with Regional, Gender, and Racial/Ethnic Breakdowns:"

... Designed to measure analytical (or deductive) and verbal (or informal logic ) reasoning skills and reading comprehension, the specific item type makeup of the scored portion of the current test is as follows:

Item Type Number of Items Reading Comprehension 26—28 Logical Reasoning A 24—26 Logical Reasoning B 24—26 Analytical Reasoning 22—24

A 35-minute writing sample is also administered at the end of the test. Prior to the 2005—2006 testing year, the time given for this writing sample was 30 minutes. Writing samples are not scored, but copies of the writing assessment are sent to all law schools to which the test taker applies.

Here's a sample analytical reasoning question:
A law firm has exactly nine partners: Fox, Glassen, Hae, Inman, Jacoby, Kohn, Lopez, Malloy, and Nassar. Their salary structure must meet the following conditions: Kohn’s salary is greater than both Inman’s and Lopez’s.

Lopez’s salary is greater than Nassar’s. Inman’s salary is greater than Fox’s. Fox’s salary is greater than Malloy’s. Malloy’s salary is greater than Glassen’s. Glassen’s salary is greater than Jacoby’s. Jacoby’s salary is greater than Hae’s.

If Malloy and Nassar earn the same salary, what is the minimum number of partners that must have lower salaries than Lopez?

(A) 3 (B) 4 (C) 5 (D) 6 (E) 7

Answer on P. 21. "This is considered an item of ”middle difficulty.”"

And here's a Logical Reasoning question:

A study has shown that there are still millions of people who are unaware that they endanger their health by smoking cigarettes. This is so despite government campaigns to warn people of the dangers of smoking. Reluctantly, one has to draw the conclusion that the mandatory warnings that tobacco companies are required to print have had no effect. Which one of the following, if true, would refute the argument in the passage?

(A) Many people who continue to smoke are aware of the dangers of smoking. (B) Some people smoke cigarettes for legitimate reasons. (C) Government has had to force companies to warn potential customers of the dangers of their products. (D) Some people who are aware of the dangers of smoking were made aware of them by the mandatory warnings. (E) Smoking is clearly responsible for a substantial proportion of preventable illness in the country.

Answer on p. 23-24. "This question is classified as ”difficult”; only 44 percent of test takers answered it correctly." Forty percent of test-takers fall for a red herring answer.

How is it scored:

As the content of the LSAT has evolved over time, the scale used to report LSAT scores has also been changed on a couple of occasions. The original LSAT scale of 200—800 remained from 1948 until 1982. Due in part to a concern that this scale gave the impression of too much precision, a scale that ranged from 10—50 was established in June 1982. This scale was later reduced to 10—48. Major changes incorporated into the current version of the test, introduced in June 1991, resulted in another score-scale change, establishing the 120—180 scale...
The results form a near perfect bell curve with mean/median/mode just above 150. (See Figure 2) with a standard deviation of about 10. So, a 180 would be three standard deviations above the mean, which, since the mean test-taker is well above average, is off the charts.

Men outscore women on average by 1 to 2 points, with men and women split almost exactly equally among test takers.

Mean scores by race for 2005-2006:

African Americans: 142.3 Native Americans: 147.3 Asian Americans: 152.1 Caucasian: 152.7 Hispanic: 146.5 Mexican American: 147.7 Puerto Rican: 138.3 Other: 150.7 No Response: 155.2

The low scores of African Americans and Puerto Ricans relative to Mexican Americans are probably related to the large number who take the test (11,288 African Americans and 2,274 Puerto Ricans) relative to Mexican Americans (only 1,789).

In comparison, 72,700 Caucasians and 8,976 Asians took the test in 2005-2006. The tiny percentage of test-takers who were Mexican Americans reflects another theme common at iSteve: that America's Eastern elites vastly underestimate the impact of illegal immigration on America because it doesn't provide much competition for their own children. They just don't notice them around much. (On the other hand, since they don't pay any attention to them in their daily lives, they are suckers for bad ideas about them in their business and political lives: "Should we purchase a Mortgage-Backed Security consisting of 1,000 half-million dollar zero money down loans in California? Sure, why not? What's the worse that could happen?")

The really bad Puerto Rican scores probably also have something to do with the test being given in Puerto Rico, but the test is only given in English. Puerto Rico has its own little universe of colleges and law schools that nobody in America has ever heard of because they use Spanish as their language of instruction.

The LSAT is one test where African-American men outscore African-American women, but only about 60% as many black men take the test as black women.

Le Carrefour goes on to offer an LSAT to IQ conversion chart. I suspect, however, that it slightly underestimates IQs relative to LSAT scores. It assumes that somebody who scores 151 on the LSAT has an IQ of 105, which strikes me as five or ten points to low.