Mean number of children (continuous line and left scale) and percentage of mothers that had more than two children (interrupted line and right scale) determined for mothers with marital radiuses contained in different intervals arranged (in scale) along the horizontal axis.

Expected number of children as a function of marital radius (and the limits of a pointwise 95% C.I., interrupted lines), inferred by a truncated polynomial Poisson regression.

Graph representing the conditional correlation between fertility, marital radius, urbanicity, income, and education. Each variable is represented by a vertex (point), the pairs of variables for which the conditional correlation given the other variables is significantly different than zero (5% level tested using a bootstrap hypothesis test with 10,000 bootstrap samples and with permutation correction for multiple testing) are joined by an edge (line). The absence of an edge joining two variables indicates that the two variables are not significantly correlated given the other variables. The relevant conditional correlations are written beside the respective edge. If two edges are directly connected, then the corresponding two variables carry new information on each other that is not already contained in the other variables present in the graph. This theory allows both continuous and discrete variables in the same graph. The variable urbanicity separates the group of variables containing fertility and marital radius from the group containing education and income. This implies, according to the theory of graphical models, that income and education do not contain information on marital radius or on fertility that is not already contained in urbanicity.