Hi all:
I posted the following a couple years ago-- it may be pertinent to this thread.
-----------------
I think I have a proposal to come up with a legitimate distribution for
G:IQ scores. Tell me what you think:
1. We go to the store and buy an IQ test. It should have a scale at the end
that shows correct # of answers as compared to RL IQ.
2. We agree that G:IQ is a function of problem-solving ability utilizing the
3d6
distribution. That is, as outlined on Basic p.45,
Skill-10 50% success
11 62.5%
12 74.1%
13 83.8%
14 90.7%
etcetera
Now, I know skill and IQ are not the same, but from a distribution standpoint,
an increment of 1 IQ increases default or initial level skill by 1-level, as
well. So we arbitrarily assign IQ-10 to Skill-10 for purposes if IQ tests in
RL. Once we figure out the distribution in subsequent steps (outlined below),
we can shift the curve as necessary to better fit a normal distribution.
3. We assume that RL IQ will follow a Gaussian distribution. We try to find
references that confirms this-- in particular, we hope to find a percentage of
the population statistic for each IQ threshold...
4. We look at IQs relative to # of questions correctly answered. It can't be
linear, there are only a finite number of questions, so there will be a ceiling
effect at higher IQs, just as in the 3d6 distribution.
5. We superimpose the distribution of % of correct answers onto the 3d6
distribution, and see where the cutoffs are for each increment. For example,
if a RL IQ of 120 gets 65 answers out of 100, and a RL IQ of 130 gets 84% of
the
answers, we have almost a direct correlation with 3d6. Otherwise, we correlate
RL IQ with GURPS SKILL--IQ Test Taking, and get a best fit relationship between
RL IQ AND GURPS SKILL--IQ Test Taking.
6. We cross reference the distribution of RL IQs in a population with the
distribution of Skill-Taking, and initially assume this is reflective of G:IQ.
7. We shift the curve as necessary to accomodate the possible difference
between
G:IQ and Skill IQ Test Taking-- (which will also tell us whether IQ Test Taking
is a ME, MA, MH, or MVH Skill).
Methods:
Googled "intelligence quotient"
Trolled through and took several IQ tests.
Finally found a scale, as most online IQ tests mail you your IQ.
Results:
IQ is scored based on population distributions. (Gaussian.) Every fifteen
points represents a standard deviation above/below the mean. eg 85-115
represents one standard deviation. 60-130 is two, etc.
The score table is reprinted here:
Age 9 10 11 12 13 14 15 16 + I.Q
GURPS IQ
9 10 11 13 16 18 20 22 80
9
10 11 12 14 17 19 21 23 82
11 12 13 15 18 20 22 24 84
12 13 14 16 19 21 23 25 86
13 14 15 17 20 22 24 26 88
14 15 16 18 21 23 25 27 90
15 16 17 19 22 24 26 28 92
16 17 18 20 23 25 27 29 94
17 18 19 21 24 26 28 30 96
10
18 19 20 22 25 27 29 31 98
19 20 21 23 26 28 30 32 100
20 21 22 24 27 29 31 33 102
21 22 23 25 28 30 32 34 104
22 23 24 26 29 31 33 35 106
23 24 25 27 30 32 34 36 108
24 25 26 28 31 33 35 37 110
11
25 26 27 29 32 34 36 38 112
26 27 28 30 33 35 37 39 114
27 28 29 31 34 36 38 40 116
28 29 30 32 35 37 39 41 118
29 30 31 33 36 38 40 42 120
30 31 32 34 37 39 41 43 122
31 32 33 35 38 40 42 44 124
12
32 33 34 36 39 41 43 45 126
33 34 35 37 40 42 44 46 128
34 35 36 38 41 43 45 47 130
35 36 37 39 42 44 46 48 132
36 37 38 40 43 45 47 49 134
37 38 39 41 44 46 48 50 136
13
38 39 40 42 45 47 49 51 138
39 40 41 43 46 48 50 52 140
40 41 42 44 47 49 51 53 142
41 42 43 45 48 50 52 54 146
14
42 43 44 46 49 51 53 55 150
43 44 45 47 50 52 54 56 155
44 45 46 48 51 53 55 57 160
15
45 + 46+ 47+ 49 + 52 + 54 + 56 + 58+ 160+ 16
"GURPS IQ" column is my own, added based on the principle outlined in my
proposal for a solution-- that is, that IQ tests would follow a similar
"problem-solving probability of success" as outlined on the corresponding table
in G:Basic 3e/R p.45.
Here, I ASSUMED "Skill IQ test taking" as a MA skill-- I justified this since
most of the questions involved mental gyrations such as imagining geometric
shapes, logic puzzles, pattern recognition, and that this while not a learned
skill, represented problem-solving on a not quite easy basis.
Also, the numbers work better when you correlate IQ=Probability of Success via
3d6 than if you assumed IQ Test Taking is a ME skill...
Anyway, as you can see, the IQ test and the equivalent G:IQ work out pretty
nicely, but there are ceiling effects on both scales.
>From an IQ website (this is a Gaussian distribution)
1 std deviation = 68% of population
2 = 95%
3 = 97.5%
Since said distribution is symmetric, it implies that an
IQ115 84th percentile (smartest in a group of 6)
IQ130 97.5% (smartest in 40)
IQ145 98.75% (smartest in 80)
I don't have a statistical calculator handy to compute corresponding G:IQ
population distributions, but you get the picture. I'd be interested in the
actual numbers. Interestingly, based on this one scale, 12 RL IQpoints seems
to
correlate to 1 G:IQ point from 10+.
For G:IQ 16+ I'm inclined to stick with the scale as outlined in G:CI with
respect to "smartest in"... I use the "average" TL7 city for determining
population per center in terms of approximate numbers.
_______________________________________________
GurpsNet-L mailing list <[email protected]>
http://mail.sjgames.com/mailman/listinfo/gurpsnet-l