Two failures of Spearman's hypothesis: The GATB in Holland and the JAT in South Africa

C.V. Dolan, W. Roorda, J.M. Wicherts

Research output: Contribution to journalArticleScientificpeer-review

Abstract

Spearman's hypothesis states that the differences between Blacks and Whites in psychometric IQ are attributable to a fundamental difference in general intelligence (g). To investigate this hypothesis, Jensen devised the method of correlated vectors. This method involves calculating the correlation between the factor loadings of the subtest and the observed differences in means. Although the hypothesis concerns U.S. populations, Jensen's test has also been used to investigate other groups. The aim of the present paper is to test Spearman's hypothesis in a published Dutch and a published South African data set. Both data sets were previously analyzed by Jensen's method, and the results were interpreted in support of Spearman's hypothesis. In this paper, we reanalyzed both data sets by Multigroup Confirmatory Factor Analysis (MGCFA). We find that the hypothesis of factorial invariance, which implies that the same construct is measured in the groups, must be rejected. This greatly complicates any comparison of the group with respect to the test scores and makes it impossible to determine the role, if any, of g in explaining the observed differences in psychometric IQ. This conclusion runs counter to the conclusion that Spearman's hypothesis is supported in these data sets.
Original languageEnglish
Pages (from-to)155-173
Number of pages19
JournalIntelligence
Volume32
Issue number2
DOIs
Publication statusPublished - 2004
Externally publishedYes

Fingerprint

Netherlands
Statistical Factor Analysis
South Africa
Psychometrics
Africa
Fundamental
Confirmatory Factor Analysis
Test Scores
Invariance
Hypothesis Test

Cite this

@article{30d814542228423d97f8c8cc40e1f410,
title = "Two failures of Spearman's hypothesis: The GATB in Holland and the JAT in South Africa",
abstract = "Spearman's hypothesis states that the differences between Blacks and Whites in psychometric IQ are attributable to a fundamental difference in general intelligence (g). To investigate this hypothesis, Jensen devised the method of correlated vectors. This method involves calculating the correlation between the factor loadings of the subtest and the observed differences in means. Although the hypothesis concerns U.S. populations, Jensen's test has also been used to investigate other groups. The aim of the present paper is to test Spearman's hypothesis in a published Dutch and a published South African data set. Both data sets were previously analyzed by Jensen's method, and the results were interpreted in support of Spearman's hypothesis. In this paper, we reanalyzed both data sets by Multigroup Confirmatory Factor Analysis (MGCFA). We find that the hypothesis of factorial invariance, which implies that the same construct is measured in the groups, must be rejected. This greatly complicates any comparison of the group with respect to the test scores and makes it impossible to determine the role, if any, of g in explaining the observed differences in psychometric IQ. This conclusion runs counter to the conclusion that Spearman's hypothesis is supported in these data sets.",
author = "C.V. Dolan and W. Roorda and J.M. Wicherts",
year = "2004",
doi = "10.1016/j.intell.2003.09.001",
language = "English",
volume = "32",
pages = "155--173",
journal = "Intelligence",
issn = "0160-2896",
publisher = "Elsevier Limited",
number = "2",

}

Two failures of Spearman's hypothesis : The GATB in Holland and the JAT in South Africa. / Dolan, C.V.; Roorda, W.; Wicherts, J.M.

In: Intelligence, Vol. 32, No. 2, 2004, p. 155-173.

Research output: Contribution to journalArticleScientificpeer-review

TY - JOUR

T1 - Two failures of Spearman's hypothesis

T2 - The GATB in Holland and the JAT in South Africa

AU - Dolan, C.V.

AU - Roorda, W.

AU - Wicherts, J.M.

PY - 2004

Y1 - 2004

N2 - Spearman's hypothesis states that the differences between Blacks and Whites in psychometric IQ are attributable to a fundamental difference in general intelligence (g). To investigate this hypothesis, Jensen devised the method of correlated vectors. This method involves calculating the correlation between the factor loadings of the subtest and the observed differences in means. Although the hypothesis concerns U.S. populations, Jensen's test has also been used to investigate other groups. The aim of the present paper is to test Spearman's hypothesis in a published Dutch and a published South African data set. Both data sets were previously analyzed by Jensen's method, and the results were interpreted in support of Spearman's hypothesis. In this paper, we reanalyzed both data sets by Multigroup Confirmatory Factor Analysis (MGCFA). We find that the hypothesis of factorial invariance, which implies that the same construct is measured in the groups, must be rejected. This greatly complicates any comparison of the group with respect to the test scores and makes it impossible to determine the role, if any, of g in explaining the observed differences in psychometric IQ. This conclusion runs counter to the conclusion that Spearman's hypothesis is supported in these data sets.

AB - Spearman's hypothesis states that the differences between Blacks and Whites in psychometric IQ are attributable to a fundamental difference in general intelligence (g). To investigate this hypothesis, Jensen devised the method of correlated vectors. This method involves calculating the correlation between the factor loadings of the subtest and the observed differences in means. Although the hypothesis concerns U.S. populations, Jensen's test has also been used to investigate other groups. The aim of the present paper is to test Spearman's hypothesis in a published Dutch and a published South African data set. Both data sets were previously analyzed by Jensen's method, and the results were interpreted in support of Spearman's hypothesis. In this paper, we reanalyzed both data sets by Multigroup Confirmatory Factor Analysis (MGCFA). We find that the hypothesis of factorial invariance, which implies that the same construct is measured in the groups, must be rejected. This greatly complicates any comparison of the group with respect to the test scores and makes it impossible to determine the role, if any, of g in explaining the observed differences in psychometric IQ. This conclusion runs counter to the conclusion that Spearman's hypothesis is supported in these data sets.

U2 - 10.1016/j.intell.2003.09.001

DO - 10.1016/j.intell.2003.09.001

M3 - Article

VL - 32

SP - 155

EP - 173

JO - Intelligence

JF - Intelligence

SN - 0160-2896

IS - 2

ER -