Associatie - Statistics

Introduction

This is the association subtest of the Netherlandic version of the Long Test For Genius, no longer in use. Currently only the English version of the Test For Genius (Revision 2016) is in use, to prevent candidates from comparing the different language versions to thus gain an advantage. Also, there was very little interest in the non-English versions.

The lack of interest is regrettable, since the problems of the Netherlandic verbal subtests, just as those of the non-verbal subtests, are of an exceptional quality. Some of them live on in the English Test For Genius, and those that have no English versions are now revived in De Golfstroomtest - Revision 2019.

This test was created based on a few dozen of early experimental association tests; the final compendium of what is possible with this type of problems, having about the highest ceiling of any test ever, and later succeeded by the Genius Association Test. An important lesson from this test is that a test with just one narrow type of items does not suffice to measure general intelligence, and this is also reflected in the relatively low g factor loading.

Scores on Association subtest of Long Test For Genius (Netherlandic) as of 9 February 2023

Contents type: Verbal.   Period: 1996-2005

• n: 20
• Median: 20.5
• Quartile deviation: 5.5
• Range: 27
• Maximum possible: 60
• Male median: 23.0 (n = 14)
• Female median: 17.0 (n = 6)
• Proportion of males among candidates: 0.700
• Hardness: 0.63
• Resolution: 0.48

Correlation of Association subtest of Long Test For Genius (Netherlandic) with other mental ability tests

Weighted average of correlations: 0.413 (N = 114, weighted sum = 47)

Estimated g factor loading: 0.64

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 3 score pairs. All known pairs are used, including possible floor/ceiling scores or outliers.

Correlation of Association subtest of Long Test For Genius (Netherlandic) with other tests - for females

Weighted average of correlations: 0.609 (N = 20, weighted sum = 12.17)

Estimated g factor loading among females: 0.78

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 3 score pairs. All known pairs are used, including possible floor/ceiling scores or outliers.

Estimated loadings of Association subtest of Long Test For Genius (Netherlandic) 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.

N = 77

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

Balanced g loading = 0.62

National medians for Association subtest of Long Test For Genius (Netherlandic)

For reasons of privacy, only countries with 2 or more candidates are included in this table. Ranking is based on the medians, and then alphabetic.

Correlation with national I.Q.'s of Association subtest of Long Test For Genius (Netherlandic)

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

• r = 0.14 (n = 20)

Correlation of Association subtest of Long Test For Genius (Netherlandic) with personal details

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.

Reliability

• Split-half (odd-even) = 0.88
• Split-half (1, 2… 5, 6… vs. 3, 4… 7, 8…) = 0.91
• Cronbach's alpha = 0.87

Due to its length, this test reaches almost the reliability of a standalone test, despite containing only one very specific item type.

Error

• Standard error = 2.8 raw score points

Scores by age

N = 20

Scores by year taken

r_{year taken × median score} = 0.46 (N = 20)

Robustness and overall test quality

• Robustness by month = 0.64 (r_{raw scores × months} = 0.18)
• Quality = 0.617

Item analysis

Item statistics are not published as that would help candidates. To detect bad items, answers and comments from candidates are studied, as well as, for each problem, the correlation with total score on the remaining problems (item-rest correlation) and the proportion of candidates getting it wrong (hardness of the item). Possible bad items are revised, replaced, or removed, possibly resulting in a revised version of the test.

[More statistical reports]