163 isn't 98th percentile. However, if you adjusted the figures, I'd be interested in playing with an ACT/LSAT predictor.

i think the implication is that the 90-something%ile of lsat takers (whatever 163 is) corresponds to the 98th percentile of human beings, b/c the population taking lsats is not representative of the population at large.

Right. 163 is in the 90th percentile for LSAT takers, but overall, it represents a significantly higher percentile for the general population. Mensa's cutoff is 163, and they take people in the 98th percentile and above. I assume this is accurate -- they tend to be pretty obsessive about this stuff.

The real number to be concerned about is what percentile of the general population a 1020 SAT scores correponds with. 75th percentile is an extremely rough guess, based off the commonly accepted notion by psychometricians that the average IQ of a high school grad is 110. Do roughly 25% of people ages 17 and 18 take the SAT? This seems plausible. Any significant deviation from this though and the numbers will be off.