More*** Election Takeaways
- Nikola Ranick
- Nov 20, 2020
- 7 min read
Part Two Features a Minor Adaption of My Initial Preview:
2. Polling Critiques Should (Have) Wait(ed)
Yes yes yes, the polls were off, but by how much? Recall that National Averages had narrowed to a reduced, but still significant, Biden lead by election day: About a 7 point advantage in contrast to what will likely look about a 4-5% win. As was the case in 2016, it was State Polling Averages which had the most deviation, although I would suggest even this inaccuracy was a bit overblown by media outlet (perhaps that is just because this writer finds the RealClearPolitics Averages to be so darn convenient in research). Indeed, most states ended up going the way polling predicted, albeit with much tighter margins, although I argue this had much to do with undecideds breaking in favor of the incumbent (a somewhat similar theme to Senate races, as discussed below).
Indeed, if we were looking for the states with the most skewed aggregate polling results, it would be Wisconsin, having over-estimated Biden by about six points (a 6.7% RCP Victory Average vs. an actual winning margin of only .7%), and Ohio, which underestimated Trump’s winning margin by about 7 (1% RCP Margin vs. 8.2% actual). As for which state’s final results led to a different victory, that was 1) Georgia with about a 1% shift in Biden’s favor leading to his narrow victory and 2) the ever-anomaly of Florida which saw a predicted Biden lead turn into a 3 point Trump victory (a landslide by Florida’s terms). Indeed, as McCatchy Editor-in-Chief, and ever-national-pulse-predictor Kristin Roberts, suggested, Floridians really do love Trump, a special political relationship rare in today's political scene. It is certainly worth noting that the biggest state swings resulted in an expanded victory for the presumed-leading candidate, a stark contrast to Rust Belt inaccuracies of 2016 that resulted in the election of Donald Trump in the first place.
Despite an overall lukewarm error, I would say that more reputable (and therefore loud-) individually-commissioned polls as well as media expectation/desire of victory, created a mantra of what I call ‘Landslide Inevitability’ on Biden and his party’s behalf. Despite control of ⅔ Positions of Federal Power for grabs, if not all post-Georgia-Runoffs, results have now clearly demonstrated a massive wave just was not the case. And especially as the uniquely-2020 counting style appeared to present on election night, when expectations are not met, there is an immediate assumption of inaccuracy and bias. Yet as panelists on the FiveThirtyEight Podcast suggested: ‘The polls showed that, even with a significant polling error, Biden would win, and that is what happened”
Granted, a lot of our lingering electoral reactions have to do with the narrative on election night vs. what has occurred since then. Indeed, the COVID era of voting tabulations resulted in either a wave favoring in-person (and pro-Trump) votes or a wave favoring mail-in (and pro-Biden) depending entirely on when and how each state ran its counting and processing. For some states like PA, this led to strong Trumpian numbers that eventually evaporated when eventual counting of absentee ballots commenced. In others like OH, the precise opposite happened due to a nearly polar opposite method of tabulation alongside specific results favoring urban areas coming in frist. Especially with recent history creating (the rather unusual) expectation of victory on election night, the electorate had vastly different perceptions not just based on what network they watched, but when they turned said network off at night. This has, of course, led to accusations of fraud and skepticism, a worrying but completely expected result in a country as polarized and partisan as today’s.
Aggravatingly, analysts had said for weeks that results would both be delayed and could change significantly as they were counted---but did anyone expect the American people to internalize that or just let it go in one ear and out the other in a desire to see what we wanted and not what would be? Alas, this author fell victim to the latter, as he ordered a 9 PM Mediterranean feast with a side of Turkish coffee to stay wired throughout the late hours of the night. Fortunately, caffeine provided its normal mellowing effect and I was knocked by 10:30 with a feeling of certainty that there would be uncertainty for quite a bit, and there sure was! Granted, any pragmatic analyst could tell you that the race was certainly starting to lean in one direction by early Wednesday morning. Yet as if to overcompensate for the American People’s expectation of immediate announcement, news outlets waited as long as possible to conclude mathematical certainty was on their side for a Biden call midday Saturday, something which likely will not assuage (and perhaps further inflame) a still politically-entrenched American public. Ultimately, this author wonders if the broken polling narrative still espoused presently will hold into the future, especially since the projected winner in the polls still ended up carrying the race (another stark difference to 2016).
Despite this rather abruptly concluded narrative, larger polling errors are evident down the ballot- such of which I would argue are neither intuitively surprising nor mathematically unexplainable. In regards to specifically the Senate, I point back to my record of skepticism on Senate-Presidential Split Tickets. Although Susan Collins managed to break this trend in blue-enough Maine, every other Senate race coincided with my theory, meaning holding GOP seats in Iowa and North Carolina should not be shockers (and even Mark Kelly’s win in AZ can be explained via the Biden victory there). As for the specific mathematical results, one should never dismiss the often larger margin of errors coming from these races and more importantly, the effect of a state’s prior partisan lean in accounting for the eventual decisions of undecideds-particularly if neither candidate reaches an average of 50%. Case-in-point was North Carolina: On RealClearPolitics' Final Average, Democrat Cal Cunningham maintained a leading plurality of 47.6% to Tillis’ 45%. Note both the distance from a majority as well as the high level of lingering undecideds (or 3rd party votes)-8.6%. When it came to the eventual results, Cunningham barely under-served his polls, with a final vote of 47%. The deciding factor was instead the break towards Tillis who ultimately clocked a 48.7% plurality of victory. What does this case study mean? In a red state (or in North Carolina’s state, one that is supporting a Red Presidential Candidate), it can be assumed most, if not all, undecided voters, will break for the dominant party, just as it is true in blue states.
This was a seemingly obvious first alerted trend first alerted to me in the 2015 Kentucky Gubernatorial Election. Dem nominee Jack Conway almost exactly hit his polling estimates, but ended up losing badly to Matt Bevin because the near entirety of undecided voters swung to the right (although Kentucky Dems long had a separate state brand, recent politics have proved everything is nationalized). A similar trend occurred in the Missouri and Indiana Senate contests in 2018 with Incumbent Dem Senators McCaskill and Donnelly matching or even slightly outperforming their polling, but to no avail as their Red opponents benefited from red state lean that swung all undecided voters. I would then like to apply this scenario to both North Carolina and Iowa, as well as Maine, if one could consider longtime personal-loyalty to outweigh partisan lean (Collin’s opponent Gideon never did cross the 50% average and too had polling averages close to the eventual results, while Collins mopped up the undecided vote share).
Out of all, it is House Polling which may need the most work. My previous piece spoke of inevitable partisan correction in strong Trump districts, but I do also believe there was an unaccounted for split-ticket effect of many suburbanites. I wonder if this is from shy Republican Voters or just the iffy-ness that comes from polling as local as Congressional Races. I expect such hypotheses will be studied more in the near future, especially with future House Majorities likely to be incredibly narrow and therefore more spotlighted these next couple cycles
Of course, as much as context can create error, the error itself is still substantial, and as our election anxiety itself enunciates, even an elaboration of why or how something is the way it is can still be rejected if it is either not wanted nor adequately understood. Nonetheless, this author sees chief error in a lack of discussion even amongst political know-hows on assumptions behind undecided voters in polling. My theory suggests that a more contextualized, if not assumed, incumbent (meaning dominant party) swing for independents should be communicated in contextualizing polls. True, this would create doubt behind close polls, in say, Kansas, unless a third party candidate is polling highly, or more likely, a Democrat is hovering near 49-50% (and therefore is not enough undecided electorate left to change the outcome). Yet I believe this is a degree of doubt necessary to helping Americans further understand the political process and ideally tether their assumption of victory toward one candidate over the other
Admittedly, this is a concept unpopular both with the losing campaign and news media, both of which always like to hype up seemingly-competitiveness races for both funds and attention, but it is a narrative that should be communicated to the people in hopes that it can create more trust and understanding in a time where there is anything but. Indeed, for my neighboring state of Kansas, this was NOT the case, and many assumed a narrow race when they saw polls only putting Dem Dr. Barb Bollier about 3 points behind GOP Dr. Marshall. Alas, as my theory suggests, such a narrative of competitiveness was fueled on the assumption that polled undecideds were truly that and not likely to overwhelmingly favor the GOP candidate in a historically red state. Undoubtedly, such voters swung towards the GOP, resulting in a double digit victory that has most people scratching their head as to why they were so caught up in this contest in the first place. I am no pollster, but I do know what common sense Americans (would probably say): If it is broke, at least add some context.
Comments