Week Three Part Two:

Hello there, and welcome back to my blog. 

It looks like this week has been one of re-examining the things that I thought I knew. I began that theme with my first blog post of the week, and I am continuing it here. 

Earlier this week, I decided to check whether the Isserman classification system (the method I used to determine the urban-ness or rurality of counties) was impacting my results in a significant way. In order to do this I decided to compare the results that I was getting using Isserman classifications with ones using urban/rural classifications produced by the USDA’s (yes, that is the U.S. Department of Agriculture — no I don’t know why they have a rurality classification system either) rural-urban continuum codes.  

Isserman classified counties as urban, mixed-urban, mixed-rural, or rural based on factors I will not list out here. The USDA ranked counties on a scale of how urban they were, with one being the most urban and nine being the most rural. Some of my sources treat counties receiving a 1-3 on the scale being urban, and counties receiving a 4-9 as rural. I have done the same in this project. 

Forty-three of the counties in Virginia which are being examined in this project were rural according to the USDA, as opposed to the 56 identified as rural by Isserman. I have not done that thorough of an analysis of the USDA data yet, but preliminary analyses suggest a great similarity between these two datasets overall. 

One interesting finding I have uncovered, however, is that the most rural districts according to the USDA (7-9) are statistically significant and likely to do better than their peers when it comes to English Pass Rates (EPR). This significance disappears when the full spectrum of 4-9 is examined. 

That’s not all I discovered this week, however. I also managed to find and integrate measures of the percentage of children living in poverty (CLIP), as well as funding for schools coming from specific sources–i.e. percentage locally financed (PLF), percentage state financed (PSF), and percentage federally financed (PFF). 

While PLF, PSF, and PFF were less useful than I might have hoped, CLIP has already proven a tremendous resource. CLIP allowed me, for the first time in the course of this project, construct a model which is significant across all tests and measurements and explains much of the variation in educational outcomes. The components of this measure are as follows:

1. The Percentage Of Children Living In Poverty
2. Percentage Of The Population Which Is African American
3. If The County Was One Of The Most Rural According To The USDA

This model is important, and, barring my adding any more variables or discovering something else which is important, this will likely end up being one of my final models if not my final model itself.

I will probably talk about this more next week, but I believe that this post is already plenty long enough already.