FiveThirtyEight.com: Electoral Projections Done Right: Uncounted Votes May Push Begich Past Stevens Although Ted Stevens currently holds a lead of approximately 3,200 votes in ballots counted to date in Alaska's senate contest, there is good reason to believe that the ballots yet to be counted -- the vast majority of which are early and absentee ballots -- will allow Mark Begich to mitigate his disadvantage with Stevens and quite possibly pull ahead of him. The reasoning behind this is simple: some early ballots have been processed, and among those ballots Begich substantially leads Stevens. A tally of Alaska's 40 house districts as taken from Alaska's Division of Elections webpage suggests that Begich has won about 61% of the early ballots counted so far, as compared with 48% of ballots cast on Election Day itself. (Notes: Totals and percentages exclude ballots cast for minor-party candidates. Data for District 3 was incomplete on the Divisions of Elections website and is extrapolated from returns in the Young-Berkowitz in that district. Percentages are not calculated in districts with fewer than 100 early votes have been counted). As you can see, there is an essentially linear relationship between the percentage of regular votes received by Mark Begich in a particular district and his percentage of early votes, with his share of the early vote generally running 10-15 points higher: There are currently at least 9,500 early votes remaining to be counted in Alaska. In addition, there are more than 50,000 absentee votes, which are essentially early votes conducted by mail. Lastly, there are at least 18,000 "question" or "questioned" ballots, which consist principally of voters who may have cast ballots away from their home precincts. Let's go about allocating these votes in the following way: Early Votes. Uncounted early ballots will be allocated between Begich and Stevens in the same proportion as already-counted early ballots in a given district. In districts where fewer than 100 early votes have been counted so far, the early vote allocation is extrapolated based on election day votes, as determined from the regression line in the chart above. In each district, 5 percent of early votes are reserved for third-party candidates befoer the Begich-Stevens allocation is made. The early vote allocation alone is worth a net of about 1,750 votes to Begich, cutting his deficit with Stevens roughly in half. Absentee Votes. It is unclear whether there is a meaningful distinction between absentee votes, which are conducted by mail, and early votes, which are conucted in person. In some districts, there were almost no in-person early votes cast, suggesting that the only way to vote early in that district may have been by mail. What I have decided to do in allocating the absentee vote is to split the difference between early and regular ballots. For example, in District 32, 60 percent of early votes went to Mark Begich as did 46 percent of regular votes. Therefore, 53 percent of absentee votes (half-way between 60 and 46) were assigned to Begich in this district, and the other 47 percent to Stevens. As in the case of early votes, 5 percent of absentee ballots were reserved for third-party candidates before the Begich-Stevens allocation was made. This allocation produces a net of about 4,700 votes for Begich. Questioned Ballots. Questioned ballots are allocated according to the proportion of regular (Election Day) ballots received by Begich and Stevens in each district. However, we assume that one-third of questioned ballots will be deemed illegitimate and will be thrown out. In addition, 5 percent of questioned ballots are assigned to third-party candidates. This allocation produces a net of about 250 votes for Stevens. *-* Combining the already-counted votes with our allocation of early, absentee and questioned ballots produces a projected total of 142,174 votes for Mark Begich and 139,258 for Ted Stevens -- a win for the Democrat by approximately 3,000 ballots: Obviously, there is a lot of uncertainty in this estimate, particularly regarding the nature of absentee ballots. If absentee ballots behave like in-person early ballots, which gave a substantial advantage to Mark Begich, then Begich will defeat Stevens, perhaps by some decent margin. If they behave more like regular, election-day ballots, then Stevens will hold on to a narrow victory. If they are somewhere in between, as we have assumed, then Begich is the favorite to win, although the outcome will be close. With so few early and absentee ballots counted to date, Ted Stevens' lead is not nearly so robust as it appears. Until we get better information about the nature of the absentee vote in Alaska, the race should probably be regarded as a toss-up.