I am carrying out different methods of assigning weight of Moran's I such as distance band, KNN, queen. Moran's I only tell us the p and z value. And I want to do sensitivity analysis to see how sensitive of these methods. But I need an objective to proceed to do the sensitivity analysis, such as i am taking the highest p value as my best model to do the sensitivity analysis. The issue is accuracy measure. I have done local spatial autocorrelation. And both results is highly signiificant.