I know Doug did a prediction poll a couple of days ago, but I wanted to throw out a few in-depth thoughts and see if anyone wants to jump in. Ladoga and I discussed below a quick spreadsheet I threw together collecting all the Likely Voter polls since Oct. 1 (at Ladoga's suggestion, I've narrowed it down to polls since Oct. 20). By messing with coefficients and so forth (methodology in a footnote below), I've found that, overall, there was a slight move toward the GOP this weekend, but not in every state. I have some thoughts about the most interesting states. If anyone wants to talk about any of them, let's nerd out:
1. Alaska - Almost impossible to predict because there isn't a lot of data. The polls themselves give Sullivan only a 1 point lead, but Alaska polls notoriously overestimate Democratic turnout. Prediction: Sullivan 50-46.
2. Arkansas - One of a small handful of states to swing Democratic over the weekend. When I added the last polls, Cotton's lead went from 7.3 to 4.6. Still, too little, too late for Pryor. Prediction: Cotton 50-46.
3. Colorado - TONS of polling data for this state. Gardner has a small lead of slightly over 2 points, but Colorado polls tend to underestimate Democratic performance. The early mail-in votes could mess this up, too. Prediction: Gardner 49-48, but I am not at all confident about this. If you are looking for states where the Dems might get an upset and give themselves at least a chance to hold on to the Senate, this is one state to watch.
4. Georgia - This got away from Nunn at the end, but Perdue is still falling just a bit short of 50%. My final count has Perdue winning 49.7-47.2. So, Prediction: Runoff.
5. Iowa - Another state to keep a very close eye on, since, like Arkansas, it has actually shifted back to Braley a bit at the end, and Iowa polls tend to overestimate GOP votes. I have it as essentially a dead heat, with Ernst holding a 0.4 point lead. Prediction: Ernst 49-48.
6. Kansas - Orman is going into today the favorite. He swears he'll caucus with the winners. Several of you here say he's full of it, and he's a Democrat tried and true. We'll see. Prediction: Orman 48-47.
7. Kentucky - No way McConnell loses this. Only bright spot for Dems is that there hasn't been nearly enough polling done, so there's always the chance for a shock. Prediction: McConnell 52-45.
8. Louisiana - This is interesting because the polls have tons of undecideds. Ironically, Landrieu has a chance to top 50% and avoid a runoff, while Cassidy does not. However, once a run off happens, Cassidy is the strong favorite. So the Dems will "win" this state tonight, but it will be a bittersweet victory if it's not enough to avoid the runoff. I have Landrieu winning 47-40, so Prediction: Runoff.
9. New Hampshire - The only safe state for the Dems in the bunch, but I use the word "safe" loosely. Brown made an impressive run, but it just won't be enough. Prediction: Shaheen 50-48.
10. North Carolina - This is another virtual dead heat. Raw polling data has Hagan up by less than a point. Another state to keep a close eye on, because the Dems need it, and if Tillis wins, it's a sign of a big night for the GOP. Prediction: Hagan 49-47.
So, add this all up, and after tonight, the GOP will have 50 seats, the Democrats will have 47, and there will be Orman to deal with. That leaves the two runoffs that are likely to result in GOP victories. End result, either 52-48 or 53-47 for the GOP, depending on what Orman ends up doing.
Anyone want to jump in the sandbox with me?
goat
Methodology: I took all the LV polls from Oct 1 to today, and weighted them by date. Then I adjusted them for various possibilities based on how undecideds break. I also did the same for all polls starting with Oct 20, which slightly favors the GOP overall. My predictions use the Oct 20 numbers, with a slight national break for Dems (based on apparently national poll bias), but then re-adjusted for individual state biases, which favored the GOP in most cases (e.g., that's the reason I have Alaska as a fairly safe GOP seat, despite the polls saying it's a tossup).
This post was edited on 11/4 4:16 PM by TheOriginalHappyGoat
1. Alaska - Almost impossible to predict because there isn't a lot of data. The polls themselves give Sullivan only a 1 point lead, but Alaska polls notoriously overestimate Democratic turnout. Prediction: Sullivan 50-46.
2. Arkansas - One of a small handful of states to swing Democratic over the weekend. When I added the last polls, Cotton's lead went from 7.3 to 4.6. Still, too little, too late for Pryor. Prediction: Cotton 50-46.
3. Colorado - TONS of polling data for this state. Gardner has a small lead of slightly over 2 points, but Colorado polls tend to underestimate Democratic performance. The early mail-in votes could mess this up, too. Prediction: Gardner 49-48, but I am not at all confident about this. If you are looking for states where the Dems might get an upset and give themselves at least a chance to hold on to the Senate, this is one state to watch.
4. Georgia - This got away from Nunn at the end, but Perdue is still falling just a bit short of 50%. My final count has Perdue winning 49.7-47.2. So, Prediction: Runoff.
5. Iowa - Another state to keep a very close eye on, since, like Arkansas, it has actually shifted back to Braley a bit at the end, and Iowa polls tend to overestimate GOP votes. I have it as essentially a dead heat, with Ernst holding a 0.4 point lead. Prediction: Ernst 49-48.
6. Kansas - Orman is going into today the favorite. He swears he'll caucus with the winners. Several of you here say he's full of it, and he's a Democrat tried and true. We'll see. Prediction: Orman 48-47.
7. Kentucky - No way McConnell loses this. Only bright spot for Dems is that there hasn't been nearly enough polling done, so there's always the chance for a shock. Prediction: McConnell 52-45.
8. Louisiana - This is interesting because the polls have tons of undecideds. Ironically, Landrieu has a chance to top 50% and avoid a runoff, while Cassidy does not. However, once a run off happens, Cassidy is the strong favorite. So the Dems will "win" this state tonight, but it will be a bittersweet victory if it's not enough to avoid the runoff. I have Landrieu winning 47-40, so Prediction: Runoff.
9. New Hampshire - The only safe state for the Dems in the bunch, but I use the word "safe" loosely. Brown made an impressive run, but it just won't be enough. Prediction: Shaheen 50-48.
10. North Carolina - This is another virtual dead heat. Raw polling data has Hagan up by less than a point. Another state to keep a close eye on, because the Dems need it, and if Tillis wins, it's a sign of a big night for the GOP. Prediction: Hagan 49-47.
So, add this all up, and after tonight, the GOP will have 50 seats, the Democrats will have 47, and there will be Orman to deal with. That leaves the two runoffs that are likely to result in GOP victories. End result, either 52-48 or 53-47 for the GOP, depending on what Orman ends up doing.
Anyone want to jump in the sandbox with me?
goat
Methodology: I took all the LV polls from Oct 1 to today, and weighted them by date. Then I adjusted them for various possibilities based on how undecideds break. I also did the same for all polls starting with Oct 20, which slightly favors the GOP overall. My predictions use the Oct 20 numbers, with a slight national break for Dems (based on apparently national poll bias), but then re-adjusted for individual state biases, which favored the GOP in most cases (e.g., that's the reason I have Alaska as a fairly safe GOP seat, despite the polls saying it's a tossup).
This post was edited on 11/4 4:16 PM by TheOriginalHappyGoat