County-level results

[dwillis]

Recent Kentucky county-level election results are contained in electronic PDFs that can be parsed using Tabula or other utilities that can extract formatted text from PDFs. Convert presidential, statewide and state legislative race results into CSV files. An example of what a finished CSV file looks like is here.

• 2018 general
• 2018 primary
• 2016 general
• 2016 primary
• 2015 general
• 2015 primary
• 2014 general
• 2014 primary
• 2012 general
• 2012 primary

Earlier election results are in image PDFs that will require OCR:

• 2011 general
• 2011 primary
• 2010 general
• 2010 primary

[edwinorange]

Here are 2015, 2011, 2007 and 2003 county-level gubernatorial results.

https://docs.google.com/spreadsheets/d/119-c0hrHfniB6IO7CM1vF0edoRkucv27HFBgIYuKirw

Here are 2015, 2011 and 2007 county-level GOP gubernatorial primary results.

https://docs.google.com/spreadsheets/d/1vv5uRyKJcu_h-L1RLjcqQBfPEDzqAht5wXJYbed741Y

Here are 2014, 2010, 2008 and 2004 county-level U.S. Senate results.

https://docs.google.com/spreadsheets/d/1boIv7OwO5mBv1aEAqgtRnnqZhvGLRDA8YLgBkmMGom4

[chrisdick14]

I will start working on the 2016 general election results.

[dwillis]

Thanks, @chrisdick14!

[jcuriel-unc]

So it looks like a decent number of counties are off for KY in 2018. Some of the ones that I noticed include Fayette, Franklin, Jessamine, Nicholas, and Robertson, though there are more. Looking at the raw data, it appears to be due to the initial scans of the pdfs being off, with vote totals for Fayette being off by 5% for the sixth US House district. It will take some time to isolate the rest of the counties that are off.

[dwillis]

@jcuriel-unc thanks for this. just to clarify, when you say off, off compared to what - the state totals?

[jcuriel-unc]

I made use of the county level precinct returns to identify discrepancies in the elections data. I first initially identified races that were off by comparing them against aggregated elections returns data on the statewebsite, then summed the results by county to identify counties that were off from official totals. So far, I've been mainly focused on Congressional races. Looking at the raw data, elections results from handwritten and shaded in returns seem to be problematic being read in. The results for Amy McGrath for the sixth Congressional district within Franklin County, for example, is off by 5332 votes of the 11702 she officially received. Additionally, it seems that the raw precinct data on the pdfs do not match up to values within the csv data.

[dwillis]

@jcuriel-unc Thanks - that's helpful. It sounds like this should be a separate issue of its own. Kentucky is, unfortunately, one of the worst states in terms of election results, but that just means we need to dig in harder.

[jcuriel-unc]

@dwillis , no problem. And yeah, KY is fairly awful with its data. I just found one county, Nicholas, where apparently the county's elections official believes that 310+51 = 87. There is only so much that can be done when the data is reported so poorly. Therefore, thank you for all of the hard work put in to deal with these data and for getting us what we have. Things would be a lot worse without you.

[jcuriel-unc]

After a check of against all of the Congressional raw precinct data, it appears that there are a total of 17 counties that are off due to scanning errors. They are as follows:

Breckenridge , Calloway, Edmonson, Fayette, Fleming, Franklin, Hopkins, Jefferson, Kenton, Letcher, Lincoln, Mason, Mercer, Pulaski, Rockcastle, Simpson and Whitley.

[dwillis]

@jcuriel-unc Ok, thanks!