The spatial weights graph (queen contiguity (1 island(s) attached to nearest neighbor), row-standardized, n=343) links 343 units with 5.52 neighbors each on average (ranging from 1 to 12); 1.61% of all possible unit pairs are connected, so each spatial lag averages over a small, local set of neighbors.
The graph forms a single connected component — every unit can reach every other through a chain of neighbors, which is what global spatial statistics (Moran’s I, spatial models) implicitly assume.
1 unit(s) initially had no neighbors at all (islands: BO-05-10-02). Islands receive an undefined spatial lag, so the weights builder attaches them to their nearest neighbor by default.
Cardinalities are quite uneven (from 1 to 12 neighbors): densely connected units get smoother, more averaged lags than sparsely connected ones — worth keeping in mind when comparing local statistics.
These are associations, not causal effects. A causal reading needs a research design.