2024-election-modelling/README.md

# Nuño's deads imple election simulator

## First round: just consider the base rates.

- [x] Get past electoral college results since 2000
- [x] Get number of electors for each state with the new census
- [x] Combine the two to get an initial base rates analysis

This initial approach gives a 25% to republicans winning in the 2024 election. Why is this? Well, consider the number of electoral college votes: 

| Year | Republican electoral college votes | 
| ---- | --- | 
| 2000 | 271 | 
| 2004 | 286 |
| 2008 | 173 |
| 2012 | 206 |
| 2016 | 304 |
| 2020 | 232 |

| Year | Democrat electoral college votes
| ---- | --- | 
| 2000 | 266 | 
| 2004 | 251 | 
| 2008 | 365 | 
| 2012 | 332 | 
| 2016 | 227 |
| 2020 | 232 | 

When Democrats won with Obama, they won by a lot, whereas when Republicans won with Bush and Trump, they won by a smaller amount. Or, in other words, this initial approach *doesn't take into account that states are correlated*. 

Remedy: consider the conditional probabilities? But how? Or, relax assumptions using Laplace's law?

- [ ] Consider conditional probabilities
  - See how other models account for the correlation
- [ ] Add uncertainty using Laplace's law of succession?
  - Maybe only do this for contested states? Alabama is not going to turn Democratic?

## Second round: just consider polls

- [x] Download and format
- [x] Read
- [x] Add date of poll
- [x] Consider what the standards error should be 
- [x] Consider how to aggregate polls?
  - One extreme: Just look at the most recent one
  - [x] Another extreme: Aggregate very naïvely, add up all samples together?
- [x] Aggregate polls?
- [x] Exclude polls older than one month?
- [ ] Exclude partisan polls
- [ ] ...

https://stats.stackexchange.com/questions/274211/calculating-the-probability-of-someone-winning-from-a-poll

This time, the republicans win by a mile. 

What is happening?
- Trump polling ok in Minessota, 
- No polls in Iowa yet
- Trump polling very well in Pennsylvania, Wisconsin, Arizona, Michigan, Florida, Nevada, Georgia, North Carolina
  - In part because of the Kennedy/West/Stein vote.
- Biden is poling well in Colorado
- Not that many polls yet, in general
tweak polls in main and restructure README 2024-04-14 02:14:59 +00:00			`# Nuño's deads imple election simulator`

			`## First round: just consider the base rates.`

			`- [x] Get past electoral college results since 2000`
			`- [x] Get number of electors for each state with the new census`
			`- [x] Combine the two to get an initial base rates analysis`

			`This initial approach gives a 25% to republicans winning in the 2024 election. Why is this? Well, consider the number of electoral college votes:`

			`\| Year \| Republican electoral college votes \|`
			`\| ---- \| --- \|`
			`\| 2000 \| 271 \|`
			`\| 2004 \| 286 \|`
			`\| 2008 \| 173 \|`
			`\| 2012 \| 206 \|`
			`\| 2016 \| 304 \|`
			`\| 2020 \| 232 \|`

			`\| Year \| Democrat electoral college votes`
			`\| ---- \| --- \|`
			`\| 2000 \| 266 \|`
			`\| 2004 \| 251 \|`
			`\| 2008 \| 365 \|`
			`\| 2012 \| 332 \|`
			`\| 2016 \| 227 \|`
			`\| 2020 \| 232 \|`

			`When Democrats won with Obama, they won by a lot, whereas when Republicans won with Bush and Trump, they won by a smaller amount. Or, in other words, this initial approach doesn't take into account that states are correlated.`

			`Remedy: consider the conditional probabilities? But how? Or, relax assumptions using Laplace's law?`

continue processing polls, tweak README 2024-04-14 14:34:58 +00:00			`- [ ] Consider conditional probabilities`
			`- See how other models account for the correlation`
			`- [ ] Add uncertainty using Laplace's law of succession?`
			`- Maybe only do this for contested states? Alabama is not going to turn Democratic?`

tweak polls in main and restructure README 2024-04-14 02:14:59 +00:00			`## Second round: just consider polls`

add poll time 2024-04-14 13:58:18 +00:00			`- [x] Download and format`
			`- [x] Read`
continue processing polls, tweak README 2024-04-14 14:34:58 +00:00			`- [x] Add date of poll`
various tweaks 2024-04-14 16:03:27 +00:00			`- [x] Consider what the standards error should be`
			`- [x] Consider how to aggregate polls?`
continue processing polls, tweak README 2024-04-14 14:34:58 +00:00			`- One extreme: Just look at the most recent one`
			`- [x] Another extreme: Aggregate very naïvely, add up all samples together?`
various tweaks 2024-04-14 16:03:27 +00:00			`- [x] Aggregate polls?`
			`- [x] Exclude polls older than one month?`
continue processing polls, tweak README 2024-04-14 14:34:58 +00:00			`- [ ] Exclude partisan polls`
tweak polls in main and restructure README 2024-04-14 02:14:59 +00:00			`- [ ] ...`
add math to calculate chance of win per poll 2024-04-14 15:21:50 +00:00
			`https://stats.stackexchange.com/questions/274211/calculating-the-probability-of-someone-winning-from-a-poll`
various tweaks 2024-04-14 16:03:27 +00:00
			`This time, the republicans win by a mile.`

			`What is happening?`
			`- Trump polling ok in Minessota,`
			`- No polls in Iowa yet`
			`- Trump polling very well in Pennsylvania, Wisconsin, Arizona, Michigan, Florida, Nevada, Georgia, North Carolina`
			`- In part because of the Kennedy/West/Stein vote.`
			`- Biden is poling well in Colorado`
			`- Not that many polls yet, in general`