|
RESULTS
We examine first the correlations between national IQs and the two measures of national per capita income. These are presented in Table 2. It shows that the two measures of per capita national income are highly intercorrelated (.945). It also shows that the correlations between national IQs and the two measures of per capita national income are strongly positive as hypothesized. The national IQs are correlated .706 with per capita GNP and .757 with per capita real GDP. Both correlations are statistically significant at p<.001. We examine next the relation between national IQs and rates of economic growth. The correlation between national IQs and economic growth rates of GDP per capita over the period 1950-1990 is .605 (N=54, p<.001). The correlation between national IQs and economic growth rates of per capita GNP over the period 1976-1998 is .643 (N=56, p<.001).
It has been suggested by a referee that the mean IQs of sub-Saharan African countries are so low that they cannot be valid and that they spuriously inflate the correlations between the national IQs and the measures of per capita income and economic growth. We believe that we have to some degree met this point by showing in Table 1 that attainment in mathematics in Nigeria and South Africa is well below that in the rest of the world and that this goes some way to establishing the validity of the IQs for the countries of sub-Saharan Africa. Nevertheless to meet this point more fully we have excluded the 15 African countries and rerun the calculations. The results are that the correlation of IQ and per capita GNP 1998 falls from .706 to .625; the correlation of IQ and real GDP per capita falls from .757 to .586; the correlation of IQ and economic growth per capita GDP 1950-90 falls from .605 to .600; and the correlation of IQ and economic growth per capita GNP 1976-98 falls from .643 to .513. Thus the exclusion of the 15 African countries reduces the correlations to some degree, as would be expected with the reduction of variance in the reduced sample, but all four correlations remain substantial and statistically significant at p<.001. We are forced to conclude that the exclusion of the 15 countries of sub-Saharan Africa makes no significant difference to the associations between national IQs and economic growth.
It has been pointed out that correlation analysis does not establish causality because of the fact that correlations merely measure covariation. Let us conseder what causality presupposes. Manheim and Rich (1986: 21-22) say that it is justified to postulate causal relationships only when four conditions are simultaneously met: First, the postulated cause and effect must change together, or covary. Second, the cause must precede the effect. Third, we must be able to identify a causal linkage between the supposed cause and effect. Fourth, the covariance of the cause and effect phenomena must not be due to their simultaneous relationship to some other third factor. We think that the relationship between national IQ and the measures of per capita income and economic growth meets these requirements quite well. First, correlations indicate that the postulated cause and effect change together. Second, because differences in national IQs are partly genetic, they have certainly preceded contemporary differences in economic conditions. Third, the causal linkage between the hypothesized cause and effect will be discussed and explained in the next section. Fourth, it is highly improbable that the observed covariance between cause and effect could be due to any third factor. This last requirement will be discussed in greater detail in the next section. Consequently, we are quite confident that the relationship is causal.
Although the correlations between national IQs and the measures of per capita income are high, there are some countries which have much higher per capita incomes than would be expected from their national IQs and other countries whose national per capita incomes are much lower than expected. To examine these anomalies a regression analysis has been carried out to disclose which countries deviate most from the regression line. This analysis is limited to the regression of real GDP per capita 1998 on IQ. Real GDP per capita 1998 was selected for this analysis because real GDP per capita (purchasing power parity) can be regarded as a more valid measure of living standards than per capita GNP and because the correlation between national IQs and real GDP per capita is stronger than the correlation between national IQs and per capita GNP (see Table 2). The results of regression analysis are given in Table 3.
See
Table 2
Table 3
Table 3 shows how much individual countries deviate from the regression line, which represents the average relationship between national IQs and real GDP per capita in 1998. "Fitted GDP" indicates the predicted value of real GDP per capita in 1998. If the correlation between IQs and Real GDP per capita were perfect, all countries would be at the regression line and all residuals would be zero. Because the correlation (0.757) is not perfect, all countries deviate to some extent from the regression line. The residuals indicate the size and direction of the deviations. Positive residuals indicate that nations have higher real GDP per capita than is predicted on the basis of the average relationship between IQs and real GDP per capita, while negative residuals indicate that their per capita incomes are lower than expected. The sum of "Residual GDP" and "Fitted GDP" is always the same as the actual value of real GDP per capita given in Table 3. There is no natural distinction between countries with large and small deviations. Because one standard error of estimate is 5,583 real GDP per capita dollars in this regression analysis, it is reasonable to regard as highly deviating cases all countries for which positive or negative residuals are larger than 6,000. Positive residuals are large for eight countries: Belgium, Canada, Denmark, Ireland, Qatar, South Africa, Switzerland and the United States. Negative residuals are large for nine countries: China, Iraq, South Korea, the Philippines, Romania, Russia, Slovakia, Thailand and Uruguay. We consider the explanations for these anomalies in the discussion.
|