As Nate Silver notes, the basic plot line is that Republicans are doing better than Democrats when the reference point is the 2008 election, and doing fairly well when the reference point is the 2006 midterm election, when early voting really took off. Republicans are holding their own in the early vote, but there's no evidence of a surge. Further, THIS WEEK is the key week for in-person absentee voting ... which is by far the most predictive metric.

There are four million ballots in the bank so far.

But figuring out trends is really difficult because there are some built-in biases to count, and every state has a different political and structural history.

1. Republicans tend to have an edge among older voters in no-fault absentee states where vote-by-mail is a more ingrained habit. 

2. The better metric is in comparing in-person early voting in one midterm election with in-person early voting in another -- 2010 v. 2006, for example.

3. The Republican lead in Pennsylvania, is, as Michael McDonald notes, deceptive. Or, rather, it's overdetermined. In Pennsylvania, you cannot simply walk into an early voting site and vote without providing a valid excuse. And historically, excuse-absentee voters are Republicans.

4. In a state like California, Democrats are doing better than they did in 2006, but Republicans are voting at a higher percentage because there are fewer registered GOPers in the state. So is the storyline that Republicans are doing much better than Democrats? Or is it that Democrats are doing better than they did in a year when they took control of Congress?

As Molly Ball notes, "In the 2006 early vote -- a great year for Democratic candidates -- each party drew 41 percent, a performance that was below Democratic registration and well above the Republican share."