October 24, 2017

Malnutrition and National IQ Differences

In this article, I will report the results of an analysis on the role of malnutrition in national and racial IQ differences. First, data on the prevalence of being underweight, wasting, and stunting, were factor analyzed to create a single variable measure of national malnutrition. Next, malnutrition was entered into a serious of regressions predicting national IQ alongside dummy variables for the region which nations were in. Malnutrition was found to have an effect on national IQ independently of region. East Asia and Africa were found to predict IQ independent of malnutrition. Next, malnutrition was entered into regressions predicting IQ along size dummy coded variables measuring whether a nation’s population was mostly Black, White, or East Asian. Malnutrition and each race were shown to predict national IQ independently of one another. Finally, malnutrition was entered into a regression alongside a dummy variable reflecting whether a nation was White or Black. Controlling for malnutrition was found to have no effect on the international Black-White IQ gap, and malnutrition was statistically insignificant in predicting national IQ among only Black and White nations after race was controlled for.

Samples and Measures

Data on the prevalence of stunting, wasting, severe wasting, and being underweight for 143 nations was taken from the World Health Organization. These variables were defined in the following way:

Severe Wasting: Percentage of children aged 0–59 months who are below minus three standard deviations from median weight-for-height of the WHO Child Growth Standards.

Wasting: Percentage of children aged 0–59 months who are below minus two standard deviations from median weight-for-height of the WHO Child Growth Standards.

Overweight: Percentage of children aged 0-59 months who are above two standard deviations from median weight-for-height of the WHO Child Growth Standards.

Stunting: Percentage of children aged 0–59 months who are below minus two standard deviations from median height-for-age of the WHO Child Growth Standards.

Underweight: Percentage of children aged 0–59 months who are below minus two standard deviations from median weight-for-age of the World Health Organization (WHO) Child Growth Standards.

National IQ data for the same nation was taken from Lynn and Vanhanen (2012).

Dummy variables were created to reflect the region a nation was located in (based on WHO designation) and whether a nation’s population was mostly White, Black, or East Asian. (Populations which were not White, Black, or East Asian were not assigned a value.) The distinction between these two variables is conceptually significant. For instance, WHO classifies Australia as a south Asian country and Turkey is grouped in with Europe. Racially, however, Australia is grouped in with Europe and Turkey is not.

Bivariate Analysis:

Stunting, wasting, and being underweight are all common health consequences of severe malnutrition. All three variables were found to significantly predict national IQ.

Correlation Matrix.JPG

All measures of malnutrition also correlated highly with one another. In order to avoid problems of collinearity, wasting, stunting, and underweight were factor analyzed.

Factor Analysis:

Because the latent variable (malnutrition) which causes covariance in severe wasting, wasting, stunting, and % underweight, was the target of measurement, exploratory factor analysis was utilized to create a single variable.

Scree plot analysis and the “eigenvalue of one” rule both clearly dictated that these variables measured a single common factor.

Scree.JPG

This single factor model was statistically significant (Chi square = 342.581, P<.0001) and produced the following factor loadings:

Factor Loadings.jpg

Using these loadings, each nation was assigned a factor score.

Regional Regressions

In bivariate analysis, several of the dummy coded regions defined by WHO were found to significantly predict national IQ.

Region Correlations 2.JPG

With the exception of Europe, these regions continued to significantly predict national IQ when malnutrition was included in the model. Moreover, malnutrition was found to significantly predict national IQ independent of regionality.

Region Regression.jpg

Due to excessive collinearity, all regions could not be entered into a model at once.

Racial Regressions

Dummy coded racial variables were found to be better predictors of national IQ than region was.

Racial Correlations 1.JPG

When entered into a model alongside malnutrition, all variables significantly and independently predicted national IQ. Malnutrition had the weakest independent effect of the four variables.

Racial Regression.jpg

The International Black-White IQ Gap

I also created a dummy variable which indicated whether a nation was black (1) or white (0) and excluded all other nations. The remaining sample size was 68 nations.

Black White Regression 1.jpg

Malnutrition was reduced to statistical and practical insignificance when entered into a model alongside Black vs White. The Black-White international IQ gap remained virtually unchanged after controlling for Malnutrition. However, analysis of the Q-Q plots, residual vs dependent chart, and residual histogram, suggest that some regression assumptions were moderately, though not severely, violated in the third model. Given this, the results should be interpreted with some caution.

Facebook Comments