There’s a fair amount of hand-wringing about how the race has “tightened” in Virginia over the last month or so, but I’m not buying it. The issue seems to be there there are a few firms (including Kellyanne’s own “The Polling company”) which did no polling until now, and have a big Republican lean and thus their first polls skew the average. But what happens when you look at how individual polls have trended since, say, September?*
Poll |
September
(Dem Lead)
|
Now |
Change |
Suffolk |
Tie |
+4 |
+4 |
Washington Post |
+13 |
+5 |
-8 |
Quinnipiac |
+10 |
+17 |
+7 |
Newport |
+6 |
+7 |
+1 |
Fox News
|
+4 |
+7 |
+3 |
Monmouth |
+5 |
-1 |
-6 |
Average |
+6.3% |
+6.5% |
+0.2% |
Virtually unchanged. And that’s including the Monmouth poll which a) only had a sample of 408, which means an error of 5%, and b) is almost 3 weeks old at this point, but which RCP considers “recent” in order to skew the results. By the way, from a statistical point of view, the uncertainty in the mean is about 2.2%, which means that provided there’s no systematic bias (and there may be! Don’t be complacent!), Northram is winning at the 3-sigma level.
I’m not saying that you shouldn’t get out and vote (do it! And phone bank!). And as with the Presidential polls in Wisconsin, it may be that there is a surprising systematic shift to all of these, but I will say that the press seems to want to push a narrative that doesn’t seem to be born out in the data.
*All polls taken from RealClearPolitics, and using their bizarre criterion for what is “recent.”