For those of us who have been around DK for awhile, you will know that the now famous Polling Aggregator/Predictor of 538.com got his start here at DailyKos as simply a community member under the pen name of “Poblano” (or is it “Poblamo”). He apparently was a sports guy with a geeky background in mathematics, who used his skills in predicting sports outcomes. When he joined DK, he started to use his talents to predict political outcomes by using polling data. He quickly became very popular around here, not because he told us what we wanted to hear, but because he was extremely accurate in his predictions, outperforming Cook and other traditional polling aggregators.
He then took his show on the road, and formed his own organization known as 538.com. Using mathematical models he would rate polling firms for accuracy and weight them accordingly in his predictive programs. As folks who have followed him over the years know, his predictive models have been amazingly accurate, earning him the moniker “Trust in Nate” amongst the DK community. He has been our “go to” guy for a number of election cycles, that is up until this one.
So what went wrong? Why was he just as wrong as the others?
For one, whatever adjustments he has made to his poll rating mathematical model, if any, it has not made it better. Giving the notoriously GOP biased Trafalgar Group pollster an “A-” rating is just nuts. Using obviously right or left biased pollsters in 538’s predictive model doesn’t make for a more accurate prediction. I suppose one could argue that the more polls you use to predict, the better and that the left/right biased polls would cancel each other out. But that leads to the second problem Nate had this cycle.
The last minute flood of GOP biased polls, that were not cancelled out by polls on the left. Many even casual observers saw this was happening, but 538 did nothing to compensate for it in their predictions. Nate must have observed that the GOP biased pollsters were out of step from what the Independent Pollsters (e.g., NYT/Siena, etc.) were putting out, who in the end were proven to be far more accurate. So why did he leave the biased pollsters in his Predictive Model? Did he think the independents were some how off base? As it turned out, if Nate had only used the independent polls in the weeks leading up to the mid-terms, 538 would have been pretty much right on the mark, and he would have been once again been revered as a political predictive genius. But in the end, 538 was as wrong as the others.
To be fair, when it comes to polls on House races, they were few and far between. So a lot of races only had a few polls, with many being from GOP biased pollsters. So I will cut Nate some slack for some of t6he sparsely polled House races. But in Senate and other Statewide races in swing States, there were plenty independent polls, so no excuse there.
Not sure if Nate will read this post or if he even hangs around here anymore. But if he does, I would love to hear his side of the 538 story regarding the mid-terms, and what, if anything he plans to do with his models to improve their predictive capabilities next time. So please Nate, feel free to chime in.
My only suggestion as a lay person, would be to either eliminate the right and left biased pollsters from 538’s predictive model, or at least rate them very low based on their biased nature and poor predictive track records. That is if 538’s objective is still to strive for predictive accuracy, and not some other motive.
Now onto a little bit of This Week In Congress (TWIC):
In the short term, the House is in Session today, but only doing mundane crap like naming post offices, yawn!
However, the Senate will be continuing the debate on the House-passed marriage equality legislation protecting same-sex and interracial marriage under federal law, which was brought to the Senate floor by a successful Cloture Vote before the Thanksgiving Holiday. There is another procedural vote on the Bill scheduled for 5:30 pm today. This will likely be another Cloture Vote. This one to limit debate on the Bill so it can be voted on for final passage later in the week. Unless there are a lot of GOP defections from the previous Cloture Vote, which is unlikely, we should prevail today. Here is your C-Span Link to watch the Senate today.
Looking ahead, eyes are all on whether Congress will use its last Budget Reconciliation chip to pass Debt Ceiling legislation in this Lame Duck session by a majority vote in the Senate. This of course is essential since the GOP House is primed to use the default on our Nation’s debts as a bargaining chip in the next Congress. I am not as much afraid of us going over the debt cliff in 2023 as I am to what the Democrats will give away to avoid it.
Stay Tuned!