I intend to apply a hot spot analysi (Gi*) for which I want to provide the optimal conceptualization of spatial relationship. To get a hint on what might be the best suitable fixed distance band for my research purpose, I tried the tool Incremental Spatial Autocorrelatation. At that point the problem occurs: the results are just fine in case of proposed distance (z-score peak is around 50 meters which is OK) but the z-score is way too high! If I reduce the spatial extent of the analysis by running the script for a smaller amount of data or in better words, for a part of the data, the z-score decreases to appropriate value ranges (-1 til 3 or 4).
Any idea why this happens? Have I missed some detail here? I attached the graphical result to clarify my problem.