Statistics of Raven's Advanced Progressive Matrices (reported raw scores)

© Paul Cooijmans

Introduction

Scores from this test are sometimes reported as "I.Q." with a standard deviation of 24, and sometimes as raw scores out of 36. This report deals with the raw scores.

Notilce that the candidates reporting raw scores are not the same individuals as those reporting "I.Q."'s (although a few report both so there is a small overlap). So the scores in this report are from a different group than those in the report dealing with R.A.P.M. I.Q.'s.

Scores on Raven's Advanced Progressive Matrices (raw) as of 16 February 2023

27 *
28 *
29 ****
30 *
31 *
32 ****
33 ***
34 *********
35 **********
36 ***************

It is obvious this test too easy for those taking high-range tests and there is a strong ceiling effect.

Correlation of Raven's Advanced Progressive Matrices (raw) with other mental ability tests

Test name n r
Verbal section of Test For Genius - Revision 201640.99
Test For Genius - Revision 201640.94
Logima Strictica 24 (Robert Lato)40.89
Cooijmans Intelligence Test - Form 440.87
Mega Test (Ronald K. Hoeflin)50.83
Reason - Revision 200840.75
Test of Inductive Reasoning / J.C.T.I. (Xavier Jouve)40.71
Numerical section of Test For Genius - Revision 201070.63
Strict Logic Sequences Exam I (Jonathan Wai)100.63
Titan Test (Ronald K. Hoeflin)50.63
Cooijmans Intelligence Test - Form 370.61
Combined Numerical and Spatial sections of Test For Genius - Revision 201040.56
The Final Test50.47
Genius Association Test60.42
Raven's Advanced Progressive Matrices (I.Q.)70.41
Qoymans Multiple-Choice #140.34
Non-Verbal Cognitive Performance Examination (Xavier Jouve)80.32
Associative LIMIT40.30
Cattell Culture Fair90.28
The Sargasso Test60.26
Unknown and miscellaneous tests180.24
Wechsler Adult Intelligence Scales80.21
Logima Strictica 36 (Robert Lato)90.21
Lieshout International Mesospheric Intelligence Test70.19
Numbers110.08
Cooijmans Intelligence Test - Form 270.07
Spatial section of Test For Genius - Revision 2004110.06
Test For Genius - Revision 200470.03
Verbal section of Test For Genius - Revision 200410-0.01
Qoymans Multiple-Choice #36-0.12
Qoymans Multiple-Choice #49-0.13
Combined Numerical and Spatial sections of Test For Genius - Revision 20164-0.22
Space, Time, and Hyperspace - Revision 20164-0.29
Space, Time, and Hyperspace11-0.42
The Test To End All Tests6-0.49
Association subtest of Long Test For Genius5-0.62
Reason Behind Multiple-Choice - Revision 20084-0.66
Culture Fair Numerical Spatial Examination - Final version (Etienne Forsström)7-0.84
Qoymans Multiple-Choice #54-0.88

Weighted average of correlations: 0.181 (N = 259)

Estimated g factor loading: 0.43

Ranking in above table is based on the unrounded correlations. All available data is present in this table, no tests are left out except for those with less than 4 score pairs. All known pairs are used, including possible floor/ceiling scores or outliers.

The correlation with high-range tests is on the whole very low. This may be caused by factors like a lack of hard enough problems, a few bad items, a few errors in the scoring key (not at all uncommon with psychological tests), inconsistent testing conditions, or fraud. It is chilling to consider that this test has been used for admission to I.Q. societies for decades while obviously failing woefully to measure intelligence in the high range.

Estimated loadings of Raven's Advanced Progressive Matrices (raw) on particular item types

These are estimated g factor loadings, but against homogeneous tests (containing only particular item types) as opposed to non-compound heterogeneous tests. Although tending to surprise the lay person, it is not uncommon for tests to have high loadings on item types they do not actually contain themselves. Such loadings reflect the empirical fact that most tests for mental abilities measure primarily g, regardless of their contents; that the major part of test score variance is caused by g, and only a minor part by factors germane to particular item types. It is of key importance to understand that this is a fact of nature, a natural phenomenon, and not something that was built into the tests by the test constructors.

Typeng loading of Raven's Advanced Progressive Matrices (raw) on that type
Verbal59-0.15
Numerical280.64
Spatial37-0.08
Logical80.85
Heterogeneous420.68

N = 174

Compound tests have been left out of this table to avoid overlap.

Balanced g loading = 0.39

National means for Raven's Advanced Progressive Matrices (raw)

Country n mean score
Denmark335.33
Netherlands335.00
United_States733.71
Belgium533.60
Germany633.50
Sweden633.00

For reasons of privacy, only countries with 3 or more candidates are included in this table. Ranking is based on the unrounded means, and then alphabetic.

Correlation with national I.Q.'s of Raven's Advanced Progressive Matrices (raw)

Correlation of this test with national average I.Q.'s published by Lynn and Vanhanen:

Correlation of Raven's Advanced Progressive Matrices (raw) with personal details

Personalia n r
PSIA Neurotic - Revision 200740.90
Observed behaviour50.88
PSIA Aspergoid - Revision 200740.35
Father's educational level250.25
Sex490.19
Educational level270.18
PSIA Cruel - Revision 200740.15
Mother's educational level250.11
Disorders (own)250.02
Gifted Adult's Inventory of Aspergerisms5-0.01
Year of birth44-0.22
Disorders (parents and siblings)25-0.25
PSIA Rare - Revision 20074-0.38
PSIA Rational - Revision 20074-0.42
PSIA Antisocial - Revision 20074-0.49
PSIA Deviance factor - Revision 20074-0.50
PSIA Extreme - Revision 20074-0.55
Observed associative horizon7-0.68
PSIA Ethics factor - Revision 20074-0.69
PSIA True - Revision 20074-0.71
PSIA Cold - Revision 20074-0.74
PSIA Orderly - Revision 20074-0.80
PSIA Introverted - Revision 20074-0.82
PSIA Just - Revision 20074-0.97
PSIA System factor - Revision 20074-0.98

Estimated g factor loadings for restricted ranges

In parentheses the number of score pairs on which that estimated g factor loading is based. The goal of this is to verify the hypothesis that g becomes less important, accounts for a smaller proportion of the variance, at higher I.Q. levels. The mere fact of restricting the range like this also depresses the g loading compared to computing it over the test's full range, so it would be normal for these values to be lower than the test's full-range g loading.

Below 1st quartile0.31 (72)
Below median0.33 (208)
Above median0.08 (64)
Above 3rd quartileNaN (0)