Statistical Analysis:Ordinary Least Square – Meth Lab Indicators

This week we continue to work on our project of Statistical Analysis. I used ArcMap and the Spatial Statistics tools available to me: ArcToolbox, > Spatial Statistics Tools> Modeling Spatial Relationships> Ordinary Least Squares. Regression analysis is the most commonly used statistic in the social sciences, hence I used the OLS tool for this study. In this lab exercise I am examining the relationship between the existence of Meth Labs (Meth Lab Density), my dependent variable, and certain socio-economic factors. I ran the Ordinary Least Square tool in ArcMap more than 22 times to remove extraneous variables from the 2010 US Census data so that I could finally arrive at the six explanatory (independent variables) that I will use in my study report.

My explanatory variables are listed below:

The objective of running the OLS tool so many times was to create a valid model that would aid in the prediction of the presence of Meth Labs. This involved examining a variety of statistics, listed above, in a methodical way to evaluate six separate Checks:

     1. Are your independent variables helping or hurting?

     2. Are the relationships what I expected?

     3. Are there redundant explanatory variables?

     4. Is your model biased?

     5. Are any important independent variables missing?

     6. How well is the model predicting the dependent variable?

After reviewing each "Check" above and after sifting through approximately 70 US Census attributes, I arrived at my destination with data that still needs to be evaluated. But, I can say at this point that I have six explanatory variables that show some relationship to the prediction of Meth Lab Density.