Aerial Interpolation

[SWhatley_3]

I am trying to interpolate population per square mile from countywide census block groups to cadastral parcels. The model is Guassian and based on event counts. I'm looking at my cross-validation results and noticing that the Root-Mean-Square is 4107.80 and Average Standard Error is 10284.44. According to ESRI, the RMS and ASE should be roughly the same. My question is how do I get the RMS and ASE to be about the same? I've tried removing some of the larger outside block groups but haven't found a good covariance fit.

I've included the cross validation stats and model summary below. I welcome any advice. 

Thanks,

SWhatley_3

Below is the summary for model:

[SWhatley_3]

I've attached a zip file of a geodatabase containing the block groups feature class, the interpolation layer. The Value Field in the Geostatistical Wizard will be "Pop_SqMi". I've used "Event" for the data type.