Extended Data Table 2 Principal component analysis of demographic and health survey (DHS) variables versus block complexity

From: Infrastructure deficits and informal settlements in sub-Saharan Africa

  1. The statistical significance and expected directional effect of block complexity holds across all three model specifications. Principal Component 1 (PC1) is the dependent variable and is taken from the PCA on 67 DHS indicators. PC1 explains 47.6% of the variance in the PCA, followed by PC2 at 8.6%. Regressions were based on 219 observations at the subnational administrative level. The Spearman-rank correlation between block complexity and PC1 is -0.709. The model is an ordinary least squares regression. Standard errors to estimated parameters are shown in parentheses. Note that the adjusted R2 increases with country fixed effects and controls for other spatial variables. Of these controls only the region’s level of urbanization (population share in urban areas) is significant.
Back to article page