2020 Predictions

[JDG 66]

(03-10-2020, 04:34 PM)David Horn Wrote: ... I'm still refining the individual state scoring, so that may change...

The election over election state-by-state vote can vary quite a bit.

The variation in the 2004 vs. 2000 state-by-state vote for the "R" (GWB):

Change Bush ® 2000 vs. 2004 =

1.059 (rounded down)
1. RI 1.211
2. HW 1.208  
3. NJ 1.147
4. CT 1.143  Lieberman ‘00
5. NY 1.137
6. MA 1.131  Kerry ‘04
7. TN 1.110  Gore ‘00
8. AL 1.105 
9. DE 1.091
10. OK 1.087 
11. LA 1.079
12. WV 1.079
13. AZ 1.075
14. UT 1.070
15. KS 1.068
16. MD 1.068
17. FL 1.066
18. CA 1.065
19. GA 1.060
20. NE 1.058
21. AR 1.058
22. IN 1.058
23. MO 1.057
24. KY 1.053
25. MN 1.046
26. IL 1.044
27. DC 1.043
28. PA 1.042  
29. AK 1.041
30. NM 1.041
31. ND 1.036
32. OH 1.036
33. MI 1.035
34. WI 1.035
35. IA 1.034
36. MS 1.031
37. TX 1.030  Bush ’00 ‘04
38. VA 1.023
39. WA 1.023
40. SC 1.020
41. NV 1.019 
42. CO 1.018
43. ID 1.018
44. NH 1.016
45. WY 1.016 Cheney ’00 ‘04
46. OR 1.014
47. ME 1.013
48. MT 1.010
49. NC (1.001) Edwards ‘04
50. SD (1.007)
51. VT (1.049)

(Again, I apologize for the Word doc transfer).

For example, in 2004, GWB improved his performance in MN by a multiple of 1.046. His 2004 performance in TN and CT was much higher than his 2000 performance because Gore and Lieberman were no longer on the ticket (or so I assume). His performance in NC was much lower (one of only three states where GWB lost vote percentage), probably due to Edward's presence on the ticket. TX and WY were solid "R" states, and an incumbent is only likely to raise the vote so much. I can't explain MA, other than KERRY SUCKED! I honestly don't know what the deal was there.

Interestingly, the greatest improvement occurred in states that "didn't" matter i.e., states that were going to be "R" or "D" anyway, not in so-called battleground states.

How my system applies this to a 2020 prediction: Having compared 2000-2004 with 2016-2020, I figure that the state-by-state variation will be similar. Not that each state will literally see the exact same change in 2020 as it did in 2004, e.g., the incumbent's (Trump's) vote in MI (47.50%) must multiply by 1.035 (49.16%), but that to get to a certain percentage, say, 50.00%, that he would need a multiplier of 1.053 to get there (47.50% x 1.035 = >50.01%). What I use, is that a 1.053 multiplier occurred in 24 out of 51 elections in 2004. Now, I deduct the the elections where the guys on the ticket came from because it probably effects things a bit, so 24 out of 51 becomes 21 out of 45 (46.66%) chance of getting to 50.0%. I eventually deduced a way to calculate the non-"D" and non-"R" vote for 2020 by taking the 2016 miscellaneous vote and dividing by 3.75 (the miscellaneous vote in 2000 was 3.75%, in 2004, it was 1.00%). This gives me the percentage by which the Trump could win by a plurality, in this case 49.31%, which adds another 6/45 to his chance of winning, for a total of 27/45 (60.00%).