03-18-2020, 01:09 PM
(03-18-2020, 01:07 AM)Eric the Green Wrote: I can't follow what your numbers means or what your method is James. Can you explain a bit more?
I don't think we can put that much stock in how a state voted in 2000 and 2004 anymore, and certainly not in the 1980s and 1970s which was the last time states like CT and NJ voted Republican.
Yeah, maybe I'm not explaining this well. It's hard to show the whole thing because the 4TF plays havoc with Word docs.
The 2000 vs. 2004 election simply provides a pattern for the spread, of how states shift from a candidate's first election (in this case, 2016) to his reelection attempt (2020), and also provides a pattern for how much the 3rd party/miscellaneous vote will expand or contract from one election to another. I'm not implying that the state of CT will react in in 2020 as it did in 2004 (although it might). I am implying that the 50 states plus DC, combined, give a spread sample that would show a similar pattern between one election and the next WHEN ONE OF THE CANDIDATES is the same. I'm just using 2000/2004 as a pattern for 2016/2020 because I see them as likely to have similar spreads. For 2020/2024, with no incumbent running, I'd have to use a different pattern.
Here's how it works:
1) Take the 2016 state-by-state vote (in this case, the "R") vote:
https://en.wikipedia.org/wiki/2016_Unite...l_election
We'll use Florida as an example, where Trump got 49.02% of the vote in 2016.
2) Now, find the percentage by which Trump could win with plurality.
Take the third party/miscellaneous vote from 2016, then divide by 3.75 (based on the 2000 vs. 2004 shift of 3.75% to 1.00%).
In this case, 4.36% divided by 3.75 = 1.16% (I'm rounding down). Then divide by two (I'm assuming that the split is roughly even) and subtract from 50 = 49.42%. That implies that in 2020, Trump would need at least 49.42% to win FL with a plurality.
3) Find out how much Trump's vote needs to improve to win (or how much his vote could drop, and still win).
Take the 2016 vote and compare that to what he would need to win. In this case, that's a necessary multiplier of 1.012 (i.e., 49.02 x 1.012 = 50%+. I'm rounding up, to make sure it's more than the required number).
4) Find the likelihood that Trump can do this.
Compare the necessary multiplier (1.012) to the spread of state changes between 2000/2004 (I eliminated CT, MA, TN, TX, WY, and NC due to the possible impact of the presidential and vice presidential candidates on elections, although in most cases, using the sample of 45 vs. 51 doesn't make a big difference, but I do it anyway). We're not comparing what FL did in between 2000 and 2004 (which would be a multiplier of 1.066. NO!). We're comparing it to the sample of all the states (and DC). Not counting the discarded states, 42 out of 45 states (and DC) saw a 1.012 multiplier increase or better, implying that Trump would have a 42 out of 45 chance of winning FL.
5) Find the average number of EVs that Trump gets.
Multiply Florida's 29 electoral votes by 42/45 = 27.07 EVs for Trump from FL (I'm rounding to the nearest whole number).
Do that for every state and district (in reality, it's obvious when you're past the 1.211 multiplier on the high end or the 1.049 divisor on the other end).
The result I get is 292.64 EVs as is, or 294.37 EVs if you assume that Kaine (VA) will not run in 2016, but Pence does (Trump and Clinton were both from NY, which I assess Trump as having a 0/45 chance of winning, so they really don't change what happens in NY). When we discover who the "D" nominee and who his running mate is, I might have to modify the calculation. For example, Biden as nominee would certainly change Trump's chance of winning DE from 2/45 to 0/45, changing Trump's 294.37 EVs to 294.24 EVs. Calculating the effect that Biden's running mate might have on her (?) state is a lot more complicated and unpredictable, but would probably require a multiplier of somewhere between 1.020 and 1.120. I have a way to deduce it, sort of.
If anyone really cares, you can PM me and I'll send you the worddoc will all my work.