They can't know the actual vote, because those ballots aren't opened and counted until election day.
However, there are counts of a.) how many absentee (or early vote) ballots have been sent out to voters registered with a given party and b.) how many of these ballots have been returned.
A claim that the "early vote" is 66/32 Obama/Romney would probably be based on the number of Democrat vs Republican ballots returned.
” A claim that the “early vote” is 66/32 Obama/Romney would probably be based on the number of Democrat vs Republican ballots returned”
No, those numbers, which happened in 2 polls now, one in Ohoi and one in Iowa, is based on the subsample of the poll, they polls and a fraction of those polled in that voted early ... and because it is a SMALL subsample the margin of error is huge. example: if 5% of the 500 voter sample voted early, then that is 25 people... so if 15 people said they voted O, 10 people voted R, then its a ‘huge Obama lead in early voting ... but switch just 5 votes and its the opposite!
The margin of error in a sample size of 25 is 15-20%!
In short the polls that were reporting that were reporting garbage. You cant conclude anything from such small samples.
If you want REAL DATA ON EARLY VOTING ... It would look like this:
“Colorado Republicans have so far cast more than 2,000 more ballots than Democrats since voting by mail began in the state.
Voting by mail began Monday and the Secretary of State’s office said Friday that so far 25,377 people have cast ballots. Of those, 10,884 are Republicans, and 8,516 are Democrats. Another 5,727 are unaffiliated voters.
Most of the mailed ballots so far come from conservative Douglas County, where 4,495 people have already voted. In competitive Larimer County, 3,899 voters have mailed their ballots.”
In the real world, the early voting is looking good for the GOP in Virginia, Colorada, North Carolina, and Ohio.