I've been trying to work on a project going back as far as possible to see how different candidates have performed when it comes to their percentage of the citizen voting age population (CVAP). However, in trying to find this data, I have found different official records with widely different numbers depending on the source. For example, take South Carolina in the 2000 election.
The Page
Historical Reported Voting Rates and the PDF file
Voting and Registration in the Election of November 2000 both put the CVAP turnout in SC at 60% and around 60% nationwide.
The page
Voting and Registration in the Election of November 2000, which gives actual estimated numbers, puts the CVAP turnout in SC at 48% and 57% nationwide.
The page
Voting-Age Population and Voting-Age Citizens: 2000, which seems unique to 2000 and gives actual estimated numbers, puts the CVAP turnout in SC at 47% and 55% nationwide.
So there are three different sources, all generated by the Census Bureau, all giving different numbers. I used 2000 and South Carolina as an example, but these differences exist across every state and every election because of the repeated sources of different numbers.
Does anyone know what I should be using as a source since all of these are coming from the Census but giving different numbers?