Tuesday, December 12, 2017

Pre-Knee Jerk Reactions to the Alabama Senate Special Election

All day I have been reviewing the polls and the pundits on the Alabama Senate Special Election. I have looked at them with a primarily analytical lens, which kept back sliding towards strategy and wishful thinking.

The hard truths here are that Alabama is a place where the Republican Party can put up Putin or Ba'ath party type numbers on the Democrats. That’s not an indictment of Alabama on principle, the numbers in a place like San Francisco aren't exactly nail biters. For Reference the last time Nancy Pelosi (D - CA) received less than 80% of the vote in a general election it was 1990 and 77.2% would not have been thought to be squeaking by either. 

Doug Jones had a harder road to the election than the average Democrat. He was trying to reach a population that was not necessarily receptive to his message. There are many Democratic constituencies in the state, though. If those places turn out in fine form, and the Republicans stay home or vote for Nick Saban, then the race will be close. It is impossible to tell ahead of time how close this race will be. While some may dismiss polling altogether, the adept minds in the polling industry have been noting that, the polling is strongly showing there isn't a way to know. 

Polling, really all data analysis, is riddled with something called "noise." Variables that can create a problem in understanding what the numbers are actually trying to tell you. That would be referred to as the "signal." If the signal cannot be found in the numbers then the experts would agree with the casual observer that they are junk.

For a poll to be accurate, and make sense, the numbers need to be scaled properly. This prevents less populated areas or groups which are known to vote a certain way from over loading the averages. Ideally a poll consists of a demographic that matches the electoral demographic on a 1:1 ratio. This is almost never possible. To compensate, the responses of each person will be weighted to accurately reflect demographics, location, race, party ID, likely voter screen, ect. The polls have broken along methodology lines and along turnout model lines. 

Those who have had live interviews have had Jones ahead, triggering a shy voter effect conversation. Those that have polled cell phones have been kinder to the Democrat as well, signaling a possibly hidden youth vote. One poll was released with three different turn out models to signify just how little people currently understand the electorate in Alabama. 

And how could they? Alabama special elections are very rare, so there are not an abundance of guides on which to form a competent model.

Elections there are, also, usually not close. One of the closest was in 2012 when Roy Moore ran for Chief Justice and won by 4 points. His opponent, Robert S. Vance, got so close based on big turn out, and somehow out performed Barrack Obama by 10 points.

So tonight Democrats will probably watch a candidate that they have hyped lose. It will be a replay of Georgia 6th and Montana at large. People should remember that this wasn't supposed to be close. The main takeaway from this should be that the Republican Party decided to back a child molester, and threatened to change the dynamic of national politics. Each race has been an indicator of positive things for the Democratic Party, but they will not all be as much fun as the Virginia Statewide elections. The pendulum is swinging, even in Alabama. It just has further to travel. Of course I could say these glum words and then be proven wrong in a half hour. That’s been known to happen as well. 


No comments:

Post a Comment