General Mental Ability

Report
General Mental Ability
John Manning
Overview

History of General Mental Ability

Definition, Models
Measures of GMA
 Applications



Individual Differences


GMA and Job Performance
Race, Gender, Age
Alternatives?
What is GMA?





It goes by many names: g, general mental ability, IQ,
intelligence
A general measure of cognitive functioning that should
work across several different domains
First proposed by Francis Galton (1888), an English
geneticist and relative of Darwin
Spearman (1904) proposed “g”- a general underlying
intelligence factor that accounted for correlations between
different areas on measures of intelligence.
It remains one of the most studied of all human
characteristics
Spearman’s g Model and later models
Just how positively correlated are
these subdimensions?
Science
Arithmetic
Word knldg
Paragraph
Numeric
Coding
Auto+shop
Math knldg
Mech comp
Electronic
1
1.00
0.73
0.82
0.72
0.57
0.51
0.68
0.70
0.71
0.77
2
3
4
5
6
7
8
9
10
1.00
0.72
0.70
0.63
0.55
0.59
0.82
0.70
0.68
1.00
0.82
0.65
0.60
0.61
0.69
0.64
0.72
1.00
0.64
0.61
0.51
0.67
0.57
0.63
1.00
0.72
0.38
0.62
0.45
0.47
1.00
0.33
0.54
0.41
0.42
1.00
0.47
0.75
0.76
1.00
0.62
0.60
1.00
0.75
1.00
Source: National Longitudinal Study of Youth ASVAB Scores, sample size is 11878 test takers
Factor
1
2
3
4
Eigenvalue Difference Proportion Cumulative
6.40
5.64
0.91
0.91
0.76
0.54
0.11
1.02
0.22
0.08
0.03
1.05
0.14
0.16
0.02
1.07
Correlation between the supposedly distinct ACT and SAT?
R=0.91; R2=0.82
General mental ability and stability

The genetic
inheritance of GMA

Two statistics:



h2 (genetic effect)
c2 (shared effect)
The effects of heredity
only get stronger with
age
Correlates of GMA

Childhood outcomes




Average r=0.50 with school performance
However, children in Japan and China know more math than
American children even though IQ scores are very similar
Average r=-0.19 with juvenile crime
General psychomotor skills


Reaction speed r=.30
Judgment speed r=.55
GMA and Stability

Is it really a trait? Is it stable?

Test re-test




Age 6 to 18=0.77
Age 12 to age 18=0.89
Even higher correlations if tests are given on multiple occasions
What does this mean?


It means that rank order is very stable over time
It does not mean that an average (100 IQ) 6 year old can do what an
average 18 year old can do
Data clearly show that general mental ability as
measured by the ASVAB is correlated with
education levels, income, self-esteem, and
weeks of unemployment even with a 10-year
gap between measures
Measures of GMA




General Aptitude Test Battery (GATB)
Wonderlic Personnel Test
Armed Services Vocational Aptitude Battery (ASVAB)
IQ Tests (WAIS-IV, Stanford Binet, etc.)
Examples of items from the
Wonderlic Personnel Test
Examples of items from the GATB
Examples of items from the GATB
Overall it seems like GMA is a
perfect selection tool






It’s very stable over time
It’s related to measures of learning ability and
adaptability
It’s one of the best predictors of job performance
available
The tests are cheap (about $2.00 per applicant
tested)
It really doesn’t matter which test you use since they
all measure the same thing
But…
…there are potential problems with
GMA test scores

The subtests show differences for gender




Women score higher on verbal intelligence
Men score higher on visual-spatial intelligence and the math
section of the SAT
These differences are not very great and there are massive
overlaps
However, overall IQ scores are basically equivalent
for men and women
GMA Scores and Race/Ethnicity
 There
are clear differences in GMA
scores based on ethnicity
African-Americans scored about 1 SD below
Whites, although more recent data suggests
this difference is closer to 0.6 to 0.7 SD
 Latinos score between Whites and AfricanAmericans
 Asians score slightly higher than Whites
 Ashkenazi Jews score about ½ SD higher
than other White groups

Returning to our conception of
biased from earlier



GMA tests predict just as
well for both groups, but
one group gets lower scores
This passes the equity
standard, but not equality
There is no factual or
logically rigorous reason to
say this is okay or not
GMA for selection
Job complexity
General job families
High (synthesizing/coordination)
Medium (compiling/computing)
Low (comparing/copying)
Industrial job families
High (setup work)
Low (feedback/offbearing)
Nuclear weapons specialist
Air crew operations specialist
Weather specialist
Intelligence specialist
Fireman
Dental assistant
Security police
Vehicle maintenance
General maintenance
GATB Validity for:
Proficiency
Training
ratings
success
% of U.S. Workers
in these
occupations
0.59
0.51
0.40
0.50
0.57
0.54
15%
63%
18%
0.56
0.23
0.65
n/a
3%
2%
Percentage of training
success explained by:
Everything
GMA
else
77.3%
0.8%
69.7%
1.8%
68.7%
2.6%
66.7%
7.0%
59.7%
0.6%
55.2%
1.0%
53.6%
1.4%
49.3%
7.7%
28.4%
2.7%
The Relation Between General Mental Ability (GMA) and Performance in
Job Training and on the Job: Representative Findings From MetaAnalyses
Study
Occupation
Performance measures
On the job
In training
Hunter and Hunter (1984)
Medium complexity
.51
.57
Pearlman et al. (1980)
Clerical
.52
.71
Hirsh et al. (1986)
Law enforcement
.38
.76
McHenry et al. (1990)
Military—enlisted
.63b
NR
McHenry et al. (1990)
Military—enlisted
.65c
NR
Hunter (1986)
Military—enlisted
NR
.63
Ree et al. (1994)
Military—enlisted
.45
NR
Ree and Earles (1991)
Military—enlisted
NR
.60
Schmidt et al. (1979)
First-line supervisors .64
NR
Schmidt et al. (1979)
Administrative clerks .67
NR
Schmidt et al. (1980)
Computer programmers .73
NR
Callender and Osburn (1981)
Refinery workers
.50
.31
Applicant reactions to GMA tests




In general, results suggest applicants tend to find
GMA tests acceptable
They especially like tests that are called “personnel
tests”
They like tests more when they are paired with
individual interview information
They usually like tests of knowledge and abilities
more than they like personality tests or biodata.
Bobko, Roth, & Potosky:
Cognitive Ability and Alternative Predictors
Their meta-analytic matrix shows that structured interviews,
conscientiousness, and biodata all have much lower d-scores
than cognitive ability in terms of prediction, and many of these
alternatives have good correlations with job performance
Validity vs. Value


GMA is the single strongest predictor of performance
Value of interviews, collecting biometric data, etc.?
To wrap it up…






GMA is the totality of mental processes involved in
adapting to the environment.
Underlying component of all different aspects of
intelligence.
Stable over time, strong genetic/heredity component
Strong predictor of future job performance, especially
with complex and technical jobs.
People like GMA tests.
Value of alternatives?

similar documents