So, I want to look at what would have happened if the election were held at different times. As we know, Clinton weakened significantly from where she was throughout most of the election, and as such, lost the entire thing. Here, I want to evaluate where Clinton would be assuming older polling data and applying some changes to it to compensate for how the polls were off. This is to give us a different picture of how the election would have played out.
Compensating for polling data
So let's look at the polling data I collected on November 7th and 8th, and compare it with the final election results. Here, I will compare the election data with the final results, and apply the data to previous election analyses that I gave in previous months. This will allow us to get an idea of how Clinton would have performed vs Trump in months leading up to the election and allow me to handicap previous polling data to give an idea of how things would have played out. This would be better than just slapping on, say, a Trump +3 model and calling it a day. On the downside, trends not present in the polls on election day may not have been present in previous months. Regardless, the handicaps are as follows:
Arizona: 0.1% Trump
Colorado: 0.1% Trump
Florida: 1.1% Trump
Georgia: 0.9% Trump
Iowa: 6.6% Trump
Maine: 1.8% Trump
Michigan: 3.7% Trump
Minnesota: 4.5% Trump
Nevada: 3.2% Clinton
New Hampshire: 0.2% Trump
New Mexico: 3.3% Clinton
North Carolina: 2.8% Trump
Ohio: 5.1% Trump
Pennsylvania: 3.1% Trump
South Carolina: 7.8% Trump
Virginia: 0.1% Trump
Wisconsin: 7.5% Trump
As we can see, overall the results were much more pro Trump than the polls indicated, but the margins varied widely. In some states we saw like 0.1-0.2% differences indicating the polls were more or less perfectly dead on, but in some states there was a fairly significant difference that could have turned the tide, even with Clinton being so far ahead.
If the election were held in October....
Let's say the election were held on October 7th instead of November 8th. In this election analysis, I concluded there was roughly a 90% chance Clinton would win and that the most accurate result was 322-216 in her favor. Let's apply the polling handicaps and see what happens.
Final election results: 287-251 Clinton
If the election were held a month before it was, Clinton would have won. She would have lost Wisconsin, Minnesota, and North Carolina, but she would have been able to maintain many swing states with enough of a margin to win.
If the election were held in September....
September was a weird month. On 9/11, Clinton collapsed due to pneumonia in public during a memorial service, and so did her poll numbers. As such, the election update I did on 9/3 and 9/18 have much different results and both require some analysis. Normally I would focus on the one closer to around September 8th, but due to Clinton's weaker position around the 18th it's might be good to analyze what would have happened there too happened there too. on September 3rd, Clinton was in a strong position with a commanding 340 electoral votes if polls are to be believed. This lead shrunk to a much narrower 293 if the election were held after her 9/11 debacle.
September 3: 297-231 Clinton
I didn't track many rust belt states at the time because Clinton was so far ahead and I didn't count states with more than a 6 point lead in either direction at this point. As such, I assume Michigan and Pennsylvania were safe, but I'm not sure about Wisconsin. I left it blank and it probably wouldn't have mattered because Clinton would have won without it anyway.
September 18: 270-268 Trump
Clinton's pneumonia incident would have cost her the election. She would have lost North Carolina and Wisconsin compensating for inaccurate polls and this would have narrowly handed the election to Trump. Michigan and Pennsylvania would have likely held though.
If the election were held in August....
For this, I'll use the data point of August 4th, because the one on the 9th had less data as I removed a bunch of states I deemed as uncompetitive (and we all know how that worked out in Wisconsin). Here, Clinton commanded a solid 365 electoral votes, and was virtually guaranteed a win. Let's see how she would have held up given Trump's effective handicap in the polls though.
Final election results: 308-230 Clinton
If the election were held in August, Clinton would have won, but her generous electoral lead would have been significantly shrunk because many states would have gone to Trump under this model. North Carolina, Iowa, Ohio, Wisconsin all would have flipped, and her electoral lead would not be anywhere as certain as it appeared. And of course she never really had a chance in Kentucky either. Still, she would have been able to carry the election without much of a challenge.
If the election were held in July....
In July, my election analyses were far more rudimentary and difficult to extract data from easily, but for this, I'm going to assume the election were held on July 19th. Here Clinton was expected to win 336-202.
Final election results: 279-249 Clinton
Clinton didn't do very great here and was in danger of losing the election, but she still managed to hold on. Obviously she never had Kentucky. She lost Iowa, North Carolina, and Ohio. Wisconsin is an unknown because I did not have data here. Pennsylvania is interesting. She was only ahead by 3.2% here, but the handicap was 3.1%. She BARELY held on to Pennsylvania here, and we could have literally seen a 2000 situation with it being so close there would be constant needs for recounts and stuff.
If the election were held in June....
Back in June, I asked a simple question: is Donald Trump really screwed? And at the time, I came to the conclusion that, yes, yes he was, and I did my original election analysis back then. Obviously, the actual election panned out much differently than expected, but now we get to look at how my original election analysis would have panned out for Trump. This one is particularly difficult to get numbers from since I did not record polling margins very well and did not have a scenario in which I gave all the states to who polled closer in the margins, as such, this one requires some guess work to figure out.
Interpolated results: 300-228 Clinton
This one was particularly tough to figure out, but Clinton was ahead in Arizona, Trump was ahead in Colorado, and for many of the swing states, the handicaps automatically pushed many states to Trump. Still, I'd argue Clinton would have won, and yes, Trump actually was screwed at the time.
What can we draw from this?
Well, Clinton's lead was deceptively large this whole election. All of those analyses predicting 330+ votes in Clinton's favor? Not even close. This election was far more competitive than we realized. Clinton consistently lost states like Ohio, North Carolina, and Wisconsin to Trump. However, on the whole, Trump still didn't have enough of a chance to win for most of the election. Clinton was so far ahead even WITH the poll data being corrected to account for discrepancies that Trump STILL would have lost. Regardless, Clinton was never as invulnerable as she seemed and she never actually broke 310 electoral votes.
Still, it seems like there was a trend at the end that cost Clinton the election. In September, when she faced concerns about her health, she was put in a position of weakness polling wise and Trump would have been able to win there. Going into October, when Trump's "pussygate" scandal broke, Clinton faced a rebound that would have won her the election. However, during the last month, as undecideds made up their minds, and as she faced fatigue due to wikileaks exposing what she really thought about things, and a last minute FBI investigation, she tanked again and lost in the last hour.
I still maintain that if Sanders ran, he would have won because he would not have faced Clinton's many weaknesses as a candidate, but more establishment excuses for the lost election have some sway too. Let it be said that while Clinton was a much weaker candidate than she appeared, she still held the upper hand most of the time.
No comments:
Post a Comment