There's still perhaps a little to sort out on a few house races, but I wanted to get the first update in... From the earlier thread, I'd offered a 750 of Makers Mark to whomever could predict the election results with the lowest percentage error, on four categories:

1. Obama Electoral Votes
2. GOP seats in the house of representatives
3. Democrat seats in the Senate
4. Romney electoral votes...

The results, sorted from lowest to greatest error (actuals still not quite final, but pretty darn close)...



Uh, oh. Looks like I win my own prize! How can this be? But let's take a bit deeper look at the predictions, and Error/Bias that is computed...

First, the predictions -- see the colors? Red predictions that ended up being overly optimistic towards the GOP. Blue predictions ended up being overly optimistic towards the Democrats. White predictions were CORRECT... Can you believe how Republican leaning our predictions were? Even the poster child "libtards" had lots of them...

So what is the "Bias" column? So in this case, I assigned positive values for "error" if the estimate was biased towards the RIGHT (positive on a number line) politically. I assigned negative values for error for things that were biased towards the LEFT politically. So a person could in theory have a 20% error, but a 0% bias if their errors were split evenly -- some too GOP friendly, some too DEM friendly... For most folks their wrong estimates were all GOP-friendly, so you see the same value as their absolute error... Let's look at a graph of these numbers...



Everyone ends up leaning towards the GOP, some quite significantly more than the others. Fascinating stuff.