Characteristic indicators for the connection among cities in cross-regional v. regional clusters.

Two sets of 10,000 random samples were drawn from the whole 344-city dataset without replacement, one with 67 cities each for comparison with cross-regional clusters, and one with 25 cities each for comparison with regional clusters. The portions of cities lying in the 1st or 3rd quadrant of 1) the joint distribution of the net influx of migrants and per capita GDP and 2) the joint distribution of the percentage of minorities and natural growth rate were used as proxy characteristics for the joint distributions. Numbers are the observed portion v. the mean of the bootstrap samples and 95% CI in the square brackets; cells in bold indicates that the observed characteristic for either the cross-regional or the regional clusters is significantly different from samples at random at the 0.1 level.