Statistical Analysis - Analyze (GIS4930)

For the analyze portion of this module, we took the data previously created from the prepare portion and used the Ordinary Least Squares tool. The OLS tool allows us to conduct a regression analysis which will determine what demographic variables affect the location of meth labs. Our dependent variable was the meth lab density layer previously created along with 29 census fields. After obtaining the initial results, six checks were conducting. The following outlines our process for the checks:

1. Are the independent variables helping or hurting?
2. Are the relationships what I expected?
3. Are there any redundant variables?
   1. Is there probability near zero?
   2. Is the VIF value less than 7.5?
   3. Is the coefficient strongly positive or negative?
   4. Did removing the variable help or hurt the model?
4. Is the model biased?
5. Are important independent variables missing?
6. How well does the model predict the dependent variable?

After the first 3 checks, bias in our model needed to be determined. I looked at the Jarque-Bera Statistics score, which the p-value was 0.000123. This value is far too low, with the value needing to be over 0.05. So after re-running the tool and adding in previously removed independent variables, I was able to achieve a score of 0.1. This shows that the model is relevant and working.