Hypergeometric simulation in Sect. 3.3.5 implies a symmetric null distribution with standard deviation 0.23 for the weighted sample correlation of homogamic pairs. One suggested alternative to random matching is to generate the null distribution by randomly permuting the vector \(\left(\pi_{2, i}, m_{2, i}ight)\) of double-mating homogamic fractions, keeping the sample-size attached to each fraction. Check that random permutation of generations also gives a symmetric null distribution with standard deviation at least 10\% larger than the hypergeometric null. Which of these null distributions is the relevant one to use as a reference in this setting? Explain your reasoning.