Military and IQ Redux.

If youve ever read Daniel Seligman's "A Question of Intelligence" Im sure the most fascinating info was when he quoted the relation between the AFQT and tank gunnery skills, or between AFQT and Patriot Missile hits.

The AFQT, for those not familiar with it, is a cognitive ability test given to all military inductees. It correlates well with conventional IQ tests, with its sole weakness being that it has a low ceiling (1 in a 1000 get a perfect score).

The AFQT breaks the freshmen into 6 broad categories.
AFQT Level I,II,IIIA,IIIB,IV and V

Lvl V rank in the 0-9 percentile, Lvl IV 10-30 percentile, Lvl IIIB 31-49 percentile, Lvl IIIA 50-64 percentile, Lvl II 65-92 percentile and Lvl I 93-99 percentile.

Converted to an IQ metric this is (on a 15 point Standard Deviation scale)
Lvl 1 = ~IQ123 and above
Lvl 2 = ~IQ106 to IQ122
Lvl 3A= ~IQ100 to IQ105
Lvl 3B= ~IQ94 to IQ99
Lvl 4= ~IQ81 to IQ93
Lvl 5 = ~IQ80 and below

Lvl V is, by law, not allowed to join the army. They tried once during Vietnam in something called Project 100,000. The results were disasterous.

Now lets compare the efficiency of AFQT scores with measures of Job Proficiency.
(Note AIT (advanced individual training) graduates score lower in every measure due to less experience vis-a-vis Unit members)
(Note 2: Original RAND article here. )
[Note 3: Yes, yes, I KNOW something is wrong with the spacing. If anyone knows what causes it, post a comment :) ]


1)When asked to make a communications system operational...
Table 1
Successful System Operation and AFQT
(predicted probability)

Sample Members	CAT I	CAT II	CAT IIIA	CAT IIIB	CAT IV
Unit Members	.89	.80	.63	.47	.29
AIT Graduates	.76	.60	.40	.25	.13

SOURCE: Winkler, Fernandez, and Polich (1992).
NOTE: The midpoint in each AFQT category is used in predicting the
probability of successful operation.


2)Number of faults detected in the Communications system...

Table 2

AFQT Level	1 or More	2 or More	3 or More	4 or More
CAT I	.97	.97	.66	.29
CAT II	.94	.78	.49	.17
CAT IIIA	.87	.60	.29	.08
CAT IIIB	.78	.43	.17	.04
CAT IV	.61	.25	.09	.02

SOURCE: Winkler, Fernandez, and Polich (1992).
NOTE: The midpoint in each AFQT category is used in predicting the
probability of successful fault detection. Cell entries are the
predicted probability that the group will successfully identify the
given number of faults.

3)The famous Patriot Defense System statistics!

Table 3
AFQT and Patriot Air Defense System Operator Performance, Specific
Measures

Measure	I	II	IIIA	IIIB	IV
Asset Hits (maximum 28)	10	11	12	13	14
Hostile Kills (maximum 78)	53	51	48	45	42
Number of Missiles used for 10 tactically correct kills	20	21	22	23	24

SOURCE: Orvis, Childress, and Polich (1992).

This slate article mentioned that Lvl IV graduates chewed up more expensive ($8 mil patriot missiles) hardware per successful shot. But its even worse than that! If asset hits means friendly fire of sorts - they were 40% more likely to blow up someone/something on their own side!

Of course the liberals will say if CAT IVs can get 4 or more faults detected (example from Table 2) 2% of the time. If we would stop being racist and prejudiced and train them harder - soon they will be outdoing the CAT Is' 29%...