Fresh off another humbling loss in GA-6, Democrats are left wondering what happened that their star pupil, Jon Ossoff, was defeated by a bumbling, Planned Parenthood hating Republican (I’m only being a little fececious here).
They point to the fact the district was historically Republican, that it took an all out effort by the GOP to win, and that national factors doomed Ossoff in the end. But, worse, despite a majority of not just public but internal polls showing Ossoff ahead until the end the party now has to question the validity of its own data.
Public pollsters have widely acknowledge they have struggled to address their woes. These struggles were laid bare in 2012 when national polls underestimate Obama’s victory by almost four points. In 2014, the polls were off by so much in Democrats favor they might have swung a key Senate race or two to the left. But after missing the mark in Kentucky’s 2015 gubernatorial contest the worst blow came in 2016.
Pollsters utterly blew the 2016 election. Though the average of national polls were off only by about two percent, in the swing states of Florida, Pennsylvania, Wisconsin, Michigan and the not so swing state of Texas they were massively off they were off by margins of 5+ percent on average. To put this in contest, not a single poll in Pennsylvania or Wisconsin showed Trump ahead after June.
But the worse news was Democrats own data analytics team the Clinton camp spent millions on being significantly off. The coalition they expected to show up was supplanted by the white working class collectively showing a middle finger to their ancestral political roots.
Supposedly, Democrats, just as public pollsters have, revamped their ways. They tailored new techniques to target harder to reach downscale, white or immigrant voters. These new techniques supposedly have heralded a new Democratic resurgence and the Congressional GOP’s weakness in the era of Trump.
Some of this is simple math in reality. Republicans have a 24 seat edge in Congress but Clinton won 23 districts held by a Republican. By default, some of these districts like VA-10 in Northern Virginia would be vulnerable.
The problem is GA-6 was supposed to be the ultimate test case. The DCCC and Ossoff campaign spent almost $2 million on focus groups, testing messages. They also spent millions on polling. Their internals showed them ahead. Public polls until the last day of the race showed Ossoff ahead or tied. They even showed him with a massive lead in the early vote (which never panned out).
Worse, Democrats and public pollsters were visibly shocked that so many Republicans and Independents came out for Handel. Their surveys failed to capture a dynamic GOP pollsters and focus groups did, Pelosi is toxic for Democrats. Instead, the Ossoff campaign’s internals showed them ahead up to Election Day. Handel and the GOP’s lack of return surveys showing her ahead supposedly was proof she would lose.
Yet, just as they did in 2016, in the closing days of the race, the Republican leaning Trafalgar Group came out with a survey closest to the actual result in the race (Handel up two points and wins by four).
The question has to be asked if public and Democratic pollsters are this flummoxed about recent results might their analysis and expectations about GOP weakness be wrong? The obvious answer is you bet.
Much of the political analysis whether it be from data aggregate FiveThirtyEight, the Cook Report, or Realclearpolitics is based on past results in midterms. However, it gives little credence to the fact Republicans ran ahead of Trump in many purple states and districts.
Certainly, midterms have not been historically kind to the party in power, but we have never had a President like Trump before. Last year, Republicans showed they could tailor their brand to the unique needs/dynamic of their districts. They did and it worked! Midterms might be a different animal Presidential elections but let’s keep in mind two things. First, Trump lost many suburban, educated Clinton districts by big margins (VA-10 and CO-6 being obvious examples). Second, Republicans have not suffered a series of retirements in these swing districts as would be expected if the party thought it was going to lose big.
The lack of solid results from data raises questions about what to expect next year, especially as it pertains to the suburban, educated districts in the Sunbelt and Southwest/east that Democrats will target. If the data cannot accurately capture even close to the results in GA-6, missing political patterns obvious to even ad designers, how can they accurately capture what moderate voters are thinking?
Add all this together and you have a perfect storm for the generic ballot to expect Democrats to be currently leading by six points and Trump’s approval mired in the low 40’s. But, if the data is suspect the actual numbers we are seeing is thus likely wrong. Until pollsters can get their act together it is very likely we will see a surprise in 2018 that benefits Republicans significantly.