Results cannot be computed when there is either severe global or severe local multicollinearity (redundancy among model explanatory variables).
To check for global multicollinearity, create an OLS model using the same dependent and explanatory variables. Variables with large VIF values (above 7.5, for example) are redundant. Remove redundant variables from the model. Finding local multicollinearity is more difficult. Create a thematic map for each of the explanatory variables and look for areas with little or no variation in values. Avoid using dummy/binary variables, variables reflecting categorical/nominal data, or variables with only a few possible values. Consider combining variables to create more variation in values and their spatial distribution.