Converting GRE Scores to D&D Intelligence

[Wolf72]

I actually tutor the GRE and other tests for a living, and I have to say that they don't measure anything beyond your ability to take that test well.

In fact, no test measures intelligence, because intelligence has never been properly defined.

Try to define it in a way that can be tested, I dare you.

The school I did my masters at doesn't require GRE's ... they have found no correlation between high GRE scores and succes in a masters program.

like you said, it measures how well you can take a test ... ala SAT

 

[shilsen]

Another agreement, as someone who murdered the GREs, has taught other people to take them, and have (as an MA and Ph.D. student) seen a lot of people who took the GREs functioning in grad school. There really is nothing those tests tell you other than how well you did on that test. They don't tell you how well you'll do in grad school. And they definitely don't say anything about intelligence.

 

[Whizbang Dustyboots]

if they are the only score available, could be used for the purposes I desire despite potential accuracy problems.

Surely many more people take the SAT than the GRE. In any case, it's hardly the "only score available."

 

[Roman]

Sigh, I appreciate the insights about GREs and I have my own beefs with standardized testing of this nature in RL, but surely if GRE scores are good enough for admissions departments of most major U.S. universities, they should be good enough for the much less serious matter of being converted into the Intelligence score in a fantasy game.

May be it would be better if I phrased this as a math/statistical theory question, rather than anything to do with the GREs:

Let's say I have three normal distributions: A, B and C

Mean of distribution A: G
Mean of distribution B: H
Mean of distribution C: I

Standard deviation of distribution A: M
Standard deviation of distribution A: N
Standard deviation of distribution A: O

Correlation between distribution A and distribution B: X
Correlation between distribution B and distribution C: Y
Correlation between distribution C and distribution A: Z


How can I combine these three distributions into a single distribution?

 

[IcyCool]

Well, your maximum human intelligence for a non-epic, non-magically altered 20th level human in D&D is 26 (18 base + 5 from stat boosts +3 from venerable age).

That at least gives you an upper bound.

 

[Roman]

Well, your maximum human intelligence for a non-epic, non-magically altered 20th level human in D&D is 26 (18 base + 5 from stat boosts +3 from venerable age).

That at least gives you an upper bound.

I think if we compare the means and the standard deviations, Int. 26 will be unreachable or at the upper bound naturally. The problem is how to combine the three distributions into one in order to be able to do this.

 

[Nightfall]

GRE scoring is over-rated. I'd rather use Tetrix scores!

 

[interwyrm]

I actually tutor the GRE and other tests for a living, and I have to say that they don't measure anything beyond your ability to take that test well.

In fact, no test measures intelligence, because intelligence has never been properly defined.

Try to define it in a way that can be tested, I dare you.

Well... merriam webster gives me..
Intelligence: the ability to learn or understand or to deal with new or trying situations

I think the best way to test this is pattern completion problems. Any verbal or mathematical test is inherently flawed.

 

[Thanee]

How can I combine these three distributions into a single distribution?

You can't.

Unless you have a formula, which details, how they are related to each other, so you can actually combine them based on that knowledge.

What you probably could do, just to get some kind of result, is to match them on the D&D attribute range (i.e. 3-18, disregarding higher levels and such, or 3-25 maybe, but values above 18 should be incredibly rare) individually, making sure that the mean is at 10.5 and the full range is somehow covered (this probably means you have to raise the lower bound for the input, i.e. 200->300 or somesuch by 'compressing' the distribution from the mean downwards, since the mean is otherwise too high to get a proper match), and then give them weights that are based on the correlations (highest influence for quantitive, since it has the lowest average correlation, then analytical, then verbal with the lowest).

Bye
Thanee

 

[CRGreathouse]

You can't.

Unless you have a formula, which details, how they are related to each other, so you can actually combine them based on that knowledge.

Roman already gave that implicitly -- he mentioned that the distributions are normal.

Let's say I have three normal distributions: A, B and C

Mean of distribution A: G
Mean of distribution B: H
Mean of distribution C: I

Standard deviation of distribution A: M
Standard deviation of distribution A: N
Standard deviation of distribution A: O

Correlation between distribution A and distribution B: X
Correlation between distribution B and distribution C: Y
Correlation between distribution C and distribution A: Z


How can I combine these three distributions into a single distribution?

First, standardize the scores (setting mean=0 and stdev = 1):
(A-G)/M
(B-H)/N
(C-I)/O

If you have only one score, you're done -- just use it, or convert backward into the other forms. If you have two or three you can combine them into a weighted average. Say you have standardized score 2.1 on the GRE, -0.2 on the SAT, and 1.3 on an IQ test. If you think the SAT is about as meaningful as the GRE, and they're worth as much combined as the IQ test, calculate weighted average = 2.1 * 1/4 + -0.2 * 1/4 + 1.3 * 1/2. This can be converted back into any form desired, for example D&D Int scores (with mean 10.5 and stdev around 2).

This process is only difficult when you have different distributions.