Monday 15 June 2015

Intelligence and social class in Peruvian children


Peru did not come off well after being visited by Spaniards. That painful confrontation is the stuff of legend.  W.H. Prescott’s A History of the Conquest of Peru (1847) and more recently John Hemming’s The Conquest of the Incas (1970) are the books to read, the latter the best.

Peru is 45% Amerindian, 37% mestizo (mixed Amerindian and white), 15% white (European background) and 3% other (e.g., black, Japanese, Chinese). Geographically, the country is divided in three regions: the Coast (53% of inhabitants), the Andean mountains (38%); and the jungle (9%). Although domes and arches, iron smelting and wheeled vehicles were unknown during the Inca Empire, the Incas were brilliant stonemasons and goldsmiths, and used Quipu, a base 10 coding system of knots on strings indicating that some sophisticated mental abilities were present in that ancient population. Subsequent history has not been happy and in a 1979 visit my conversations with farmers in the countryside were about hard times, political stalemate and their shame at national backwardness in economic development. Things are much better now: life expectancy of 75 years; 74% of population live with improved sanitation facilities; 78% are living in urban areas; mortality rate under-5 (per 1000 live births) is 18.2; and 77% are enrolled in a secondary school. However, Peru is bottom of the list on PISA exams, and has no universities rated among the top 500. The discrepancy between Peruvian low school performance and reasonable cognitive performance (mean IQ in Lima 96, Andean samples 78) requires explanation, which the authors seek to supply.

Denisse Manrique Millones, Carmen Flores-Mendoza, Rosa Millones Rivalles Intelligence in Peru: Students' results in Raven and its relationship to SES.  Intelligence 51 (2015) 71–78

Their sample was 1097 children (46.5% male), with mean age of 11.6 years (SD = 0.4; ) from 18 randomly selected schools in Lima (58.9% public, 41.1% private). They gave Raven's Standard Progressive Matrices and a local measure of socio-economic status and parental education based on a national housing quality assessment.

The first finding was that the results were skewed towards higher scores. SPM is tilted towards upper average performers, and the somewhat under-average are less well represented. (Good to see some data plotted: just the sort of statistics I can understand). Overall, the results are IQ 97 but the Flynn adjustment brings it down to IQ 91. From a philosophical point of view it is strange to apply the same adjustment to different cultures. I suppose one can argue that Peru has had 3 decades of improvement, but it is still far short of European living standards. Perhaps lifespan is the best measure, and the correction valid after all. The authors guestimate the Andean IQ at 66-78 and the Amazonian IQ in the same range. They say:

Weighing the distributions of inhabitants who live on the Coast (53%), in the Andean mountains (38%) and in the Amazonia region (9%), the mean IQ for the entire country could be 84 [(91 ∗ 53 + 78 ∗ 38 + 66 ∗ 9) / 100], almost the same IQ previously estimated by Lynn and Vanhanen (2012) for Peru.

The authors then look at the implied IQs of the 573 recent arrivals out of 1097 mothers, according to what region (coast, Andes, Amazon) they were born in before emigrating to Lima. This is a little difficult, because we do not know if these internal emigrants are brighter than the people they leave behind, which is usually the case. Also, although there was data on father’s education this is not utilized in this particular comparison, as far as I can see. (There is a supplementary file which may have it). The authors claim that there is an interaction effect which vitiates a biological interpretation of the basic differences between the three regional groups, but I am not sure about their argument. Here is what they say:

As expected, children whose mothers were born in the Coastal region had a higher mean IQ. The lowest mean IQ was obtained by children whose mothers were born in the Amazonian region. These results would favor the biological hypothesis (e.g. positive correlation between IQ mother and IQ offspring). However, the results indicated an interaction between genetics and environment. For instance, the mean IQ of children of Andean origin studying in Lima was higher than the estimated IQ of children who lived in the Andean zone (IQ between 78 and 66; Majluf, 1993; Raven et al., 2000). Effectively, our results indicated the following: a mean IQ of 84 (adjusted by the Flynn effect) for children whose mothers were born in the Amazonian region (N = 28; mean SPM score = 36); a mean IQ of 85 for children whose mothers were born in the Andean mountains (N = 147; mean SPM score = 37); and a mean IQ of 94 for children whose mothers were born in the Coastal region (N = 398; mean SPM score = 41). When weighing child distribution (in percentage) according to the place of birth of their mothers, the mean IQ for the partial sample (N = 573) was 91 [((5 ∗ 84) + (26 ∗ 85) + (69 ∗ 94)) / 100], the same IQ obtained when the total sample was used (N = 1121; IQ = 91).

If emigrating mothers marry coastal Lima fathers, a likely possibility, this would sufficient to explain the partial uplift in the intellects of the resultant children. Selective migration, as described above, might also be involved.

Fathers are included in the analysis of the education/SES effects with the apparent finding that educational differences have effect in lower classes but not in the highest class. The authors say: It should be noted that the size of relationship between parents' educational level and SES was not high (r = .387), due to the imperfect meritocratic social structure that exists in developing countries, such as Peru.

However, the very well developed United Kingdom shows exactly the same pattern, as described by Daniel Nettle (2003). Note that correlations are higher for lower social classes, consistent with higher intelligence being a way out of those occupations.




The authors say: countries like Peru need to improve and expand [ ] all the means relevant for educational and cognitive development. This begins [with] aggressive gains in health care and nutrition and concludes with university education that includes average students all the way to top ability levels. Finally, improvement in education will lead to significant advances in the cognitive condition for the next generation.

I do not argue with improving education anywhere, but one notable omission in the paper is any analysis of results by racial composition. On this purely biological hypothesis the authors are silent. Nonetheless, this is a very useful study, carefully done, and makes a good contribution to the literature on intelligence in Peru. A re-analysis looking at paternal and maternal educational and genetic backgrounds would strengthen the interpretation of the results.

Final disclosure: I am not a descendent of Inca Garcilaso de la Vega, the first high-born mestizo whose bones, sent back by Spain, I saw interred in Cuzco cathedral in 1979 (preceded by an excellent declamation by the Town Mayor that being mestizo was no cause for shame) but when I saw it his portrait it struck me as a stirring image of a chronicler, to which any scribbler could feel resemblance.


  1. I read Prescott when I was a lad, having found his books on my father's bookshelves. It's pretty good stuff. Only decades later did I discover that the poor chap was almost blind.

    "Almost blind" might also refer to people who suppress investigation of interesting possibilities because they suspect that what they might discover will offend the High Priests of the day.

  2. Yes. Prescott wrote very well, and must have seen the story very vividly in his mind. Will maintain the battle against the High Priests.