Description
A spatial weights matrix is intended to be sparse, containing mostly zero values. Connectivity refers to the proportion of existing neighbor connections relative to all possible neighbor connections. At 100 percent connectivity, each feature is a neighbor of every other feature, while 0 percent connectivity indicates each feature has no neighbors. When connectivity is too high, estimates can become biased.
Solution
Specify a neighborhood configuration that results in fewer total neighbors. If using a distance band, consider reducing its size. If using polygons, consider using contiguity instead of number of neighbors or a distance band. Alternatively, if using point data, consider using Delaunay triangulation or number of neighbors.