Abstract

Spearman&rsquo;s hypothesis simply states that differences between groups on an IQ test are a function of the general intelligence (g). At the item level this would mean the magnitude of the differences between groups are smaller on low-g-loading items and larger on items with a high g loading. An empirical test by D&iacute;az, Sellami, Infanz&oacute;n, Lanz&oacute;n, & Lynn (2012) comparing Spanish and Moroccans taking the Raven&rsquo;s Progressive Matrices showed no support for Spearman&rsquo;s hypothesis, whereas other empirical studies showed modest to strong support. In the present study we tested whether another sample from the Arab world would replicate the outcomes of D&iacute;az et al. of no support for Spearman&rsquo;s hypothesis by computing a correlation between g loadings and group differences on items of an IQ tests and checking whether it showed a negative correlation. Children from Yemen (total N = 1916) were compared with a group of Russian children (N = 426) and a group of Kazakh children (N = 656) on the Standard Progressive Matrices Plus yielding outcomes of Spearman&rsquo;s hypothesis of, respectively, r = .86 (p < .0005), and r = .76 (p < .0005). The Nharmonic-weighted average correlation of .80 has a credibility interval ranging from .74 to .86. Based on the results we conclude that Spearman&rsquo;s hypothesis holds even at the item level and that the D&iacute;az et al. study appears to be an outlier