These results take into account a wide range of factors that influence spoiled-ballot rates, including education, gender, income, age, number of absentee votes, voting-machine type, ballot type and whether votes were counted at the precinct or centrally.
What I want to know is how this information is available when looking at the ballots and making decisions. I thought we had "secret" ballots in this country.
Race could be inferred by the ethnic composition of the precinct.
Democrat vs Republican could be determined with accuracy by the voting pattern in other races.
In other words, there would be sufficient data available to support the assertions made by Lott and Glassman.