About upset primaries -- A post I regret not making sooner.

I had hinted at it before: The political pundits had it backwards and, rather than having a strong chance of winning back the Senate, they might actually lose seats.

To recap: Mitch McConnell, the minority senate leader, is now stuck in a runoff and may lose, despite having boat loads of cash, to a Tea Party favorite.  Eric Cantor, house majority leader lost his primary, by double digits.  The Oregon senate race, at once thought to be competitive, is clearly not -- heck, Monica Wehby hasn't aired a single advertisement since winning her primary, while Jeff Merkley has been airing three different spots.  Suddenly, a look of the races suggest that -- gasp -- GOP may lose seats!

The following was a post that I left 3/4 done. I wish that I had completed it at the time and published it, considering the shift in the odds of late.  Many people had thought that Obama's poor numbers mattered quite a bit.  I thought otherwise.  Here's my unpublished post:

Something doesn't look right

Something's been bugging me for a while.  You see, many pollsters and pundits have come out claiming that the GOP are poised to take over both chambers of Congress.  I think they're full of it, as we haven't yet completed the primary season and they're proclaiming some divine knowledge of where the races are headed.  Whatever.

You see, every time I look at the favorable / unfavorable polling, no matter who conducts the tracking, it clearly and continually shows Democrats ahead of Republicans in Congress.  That is to say, generally the public looks less favorably upon Republicans than Democrats.

On a generic vote, how is it that both sides are roughly even, despite long-standing polling showing that the general public has frowned upon Republicans more than Democrats?

What's wrong

Here's the answer: The people who do political polling of candidates and these generic ballots, often do a terrible job of accurately designing their demographic set up.  They generally tend to get the party registration (D) - (R) - (I) split correct, but when it comes to registered independents (I), many pollsters make the error of splitting them nearly down the middle between leaning-Democrats and leaning-Republicans.

One group -- Gallup -- prods independent voters to identify the party that they most lean towards, and this is extremely important.  When you look at their combined party affiliation tracking, an extraordinary trend emerges.

Gallup's data matters quite a bit, because this annualized data is based on nearly 19,000 interviews.  By comparison, any given political poll averages about 1,000 interviews.  This gives Gallup's data a 1% Margin of Error at the 95% confidence interval -- something you don't see in political polls.

So do you see it?  Every time there's been a swing election, the year prior and up through the election year, there's been a big swing towards the winning party.

You can read it like a stock chart.  Because more Americans generally lean Democratic, when this number dips, it means that Independents are aligning themselves with Republicans and Republicans are set up for a major win.  When the number jumps, it means Independents have shifted towards Democrats and Democrats are headed for big wins.  So what do you see entering 2013?

I'm not saying that Democrats will take back the House, but it certainly does not bode well for the early expectations that the GOP will take over the Senate, now does it?