|
METHOD
This study presents data for 60 countries for national IQs, per capita incomes in 1998, and economic growth 1950-1998 and examines their relationships by the statistical techniques of correlation and regression analyses.
National IQs
National IQs have been calculated from normative data obtained in 60 countries for the Colored and Standard Progressive Matrices. The reasons for using these data are that the Progressive Matrices is the most widely used test in cross-cultural research, is non-verbal and hence is likely to yield more valid cross cultural data than verbal tests which require translation, is among the best measures of g, and the rate of secular increase is well established. The data have been obtained from the bibliographies of Progressive Matrices studies compiled by Court (1980) and Court and Raven (1995), from the data given by Raven in a series of manuals and research supplements for the Progressive Matrices, and from the Raven archive.
The Standard Progressive Matrices was constructed in Britain in the 1930s and was first published in 1938 with norms for 6-15 year olds and adults. This was followed by the publication in 1947 of the Colored Progressive Matrices, a simpler test suitable for 5-11 year olds. The Standard Progressive Matrices was renormed for 6 to 15 year olds in Britain 1979. A norm table is provided by Raven (1981) giving percentile equivalents of raw scores for half year age groups. The procedure for calculating the IQ of a country in which norms have been obtained for the Standard Progressive Matrices is to read off the raw scores of the 50th percentile from the norm table and obtain the British 1979 percentile. This is then converted to the British IQ equivalent using a conversion table. The raw score of the 50th percentile is the median IQ rather than the mean. Several studies have provided mean raw scores in addition to the medians and these show that means and medians are virtually identical. In most countries in which Progressive Matrices data have been collected norms have been given for a number of age groups. IQs are calculated for each of these and averaged to give a single national IQ. This IQ is then adjusted for the secular rise of the IQ which has been 2 IQ points per decade for the Standard Progressive Matrices in Britain over the period 1938-1979 (Lynn and Hampson, 1986). All national IQs are therefore expressed in relation to a British IQ of 100.
Norms for the Standard Progressive were collected for adults for Britain in 1992 and for the United States for 1993. The norm table for the United States provided by Raven, Court and Raven (1996) gives the most detailed information consisting of the percentile equivalents of raw scores. Less information is provided for the British standardization which gives only the raw score equivalents of the 5th, 10th, 25th, 50th, 75th, 90th and 95th percentiles. The British medians have been converted to American IQs by the use of the American norm table. The result of this calculation is that the British IQ is 102 on the American norms. Data for adults from other countries are converted to American IQs and then adjusted to British IQs by the subtraction of 2 IQ points.
There are no norms giving detailed percentiles for the Colored Progressive Matrices for Britain, the United States or elsewhere. To deal with data for the Colored Progressive Matrices, raw scores are converted to those of the Standard Progressive Matrices using the conversion table provided by Raven, Court and Raven (1995) and the IQs calculated in the way set out above.
In a few instances median raw scores fall below the 1st percentile of the British and American norm tables. The 1st percentile is equivalent to an IQ of 65. In these cases the countries are assigned an IQ of 64. For a number of countries Progressive Matrices data have been collected for two or more samples. These have been averaged to provide a single mean given to the nearest whole number.
The IQ for South Africa has been calculated as follows. The study by Owen (1992) gives the following IQs for the four racial groups. Whites: 94; blacks: 66; coloureds: 82; Indians: 83. The percentages of the four groups in the population are whites: 14%; blacks: 75%; coloreds: 9%; Indians: 2% (Ramsay, 1999, p. 158). Weighting the IQs of the four groups by their percentages in the population gives an IQ for South Africa of 72. The IQ of Singapore has been calculated in the same way by weighting the IQs of the ethnic groups (Malays, Chinese and Indians in Singapore) by their numbers in the population. The data on national IQs are shown in Appendix 1 which gives the IQ, the sample size, the test used (Colored or Standard Progressive Matrices) and the reference. For some countries there are two or more studies of the national IQ. These have been averaged to give mean IQs for these countries.
Because the concept of national IQ is new, it will be useful to examine its reliability and validity. To examine its reliability we have taken the sixteen countries for which there are two or more measures of IQ and calculated the correlation between the two measures. For the countries for which there are more than two measures (Brazil, Hong Kong, India and Mexico) we have used the two extreme values. The correlation between the two measures of national IQ is 0.937. This establishes that the measure of national IQ has high reliability.
To examine the validity of the national IQs we have examined their relation with national measures of educational attainment. This follows the long established methodology of the validation of intelligence tests among individuals by showing that they are positively correlated with test of educational attainment. The measures of education attainment are taken from the second and third international studies of educational achievement in mathematics and science. These data are shown in Table 1 for the countries for which we have IQ measures. The correlations between educational attainment and IQ are shown in the bottom two rows of the table. Five of the six correlations are statistically significant and establish the validity of the measures of national IQ.
See Table 1
|