Poll Analysis: Obama Romney gains a bit

Comments

1. 1

   eponymous coward spews:

   Sunday, 9/30/12 at 8:11 pm

   Obama wins North Dakota? Riiiiiiiiight…

2. 2

   Darryl spews:

   Sunday, 9/30/12 at 9:05 pm

   Coward @ 1,

   Oops! There was a copy of the Ohio poll (with Obama leading 49 to 42) pasted into N.D.!

   Thanks for pointing it out.

3. 3

   Richard Pope spews:

   Sunday, 9/30/12 at 9:32 pm

   So with North Dakota (3 EV) back into the Romney column, it should come out around 196 to 197 EV for Romney (up from 195 EV in your last analysis).

   Are you going to be changing the headline to: “Poll analysis: Romney gains a bit”?

4. 4

   Meme1 spews:

   Sunday, 9/30/12 at 9:33 pm

   Cue all the teabagging BS about “oversampling” and evil conspiracies designed to turn the tide of the election.

5. 5

   proud leftist spews:

   Sunday, 9/30/12 at 10:17 pm

   As the Nitty Gritty Dirt Band sang years ago, “Piss on the fire, call in the dogs, and head it on back to Bowlegs.” That’s what the Mittster needs to be planning for.

6. 6

   Darryl spews:

   Sunday, 9/30/12 at 10:21 pm

   Richard @ 3,
   Yep…that’s what I did.

7. 7

   Rael spews:

   Sunday, 9/30/12 at 10:36 pm

   Conservative trolls:

   Romney is doomed.

   You know it.

   Give up.

8. 8

   Politically Incorrect - free minds, free markets, free people spews:

   Sunday, 9/30/12 at 11:07 pm

   It won’t be so bad should the Milk Chocolate Messiah win in November. The House will be held by the Republicans, and they will gain in the Senate, too. Not much will happen between January 2013 and January 2017, but the Republicans will win in 2017 just because the electorate will be tired of the Democrats by then. The re-election of Obama also means the end of Hillary’s political career, and that’s a good thing, too.

9. 9

   proud leftist spews:

   Sunday, 9/30/12 at 11:35 pm

   PI,
   Do you really think divided/crippled government is a good thing at this time? Really? A government that cannot make decisions, at all, because of partisan divide? And, why in 2016 would the electorate be sick of Democrats when, according to your scenario, the Republicans would have been holding Congress? Actually, I think the Rs are headed for internecine warfare after this election. Romney will lose, and the righties will proclaim that the party needs to move ever farther to the right to win a national election. That party cannot move farther to the right. Implosion will occur. The world is moving to the left. This country needs to move that way, too.

10. 10

    proud leftist spews:

    Sunday, 9/30/12 at 11:37 pm

    I forgot to say that the movement of history of Western Civilization has always been to the left. It is undeniable.

11. 11

    Richard Pope spews:

    Monday, 10/1/12 at 1:32 am

    Darryl @ 6

    You must have rerun all 50 states (and DC) with 100,000 elections, instead of just running North Dakota 100,000 times. Flipping North Dakota would, on average, change the EV totals by almost 6 for each candidate. But Romney’s total moved from only 191 EV with ND being blue to only 193 EV with ND being red.

    I take it that running several 100,000 election simulations with the same data will result in the average EV totals varying by several EV’s (like maybe 2 to 5 EV’s) in each simulation?

    Does it require a lot of computer time to run these things nowadays? A lot of your own time?

12. 12

    rhp6033 spews:

    Monday, 10/1/12 at 8:00 am

    And… still no current S. Carolina polls!

    Are pollsters so convinced that Romney would win S. Carolina that it’s not even worth polling there?

13. 13

    rhp6033 spews:

    Monday, 10/1/12 at 8:06 am

    I talked to my sister in Pennsylvania over the weekend. She said they were bombarded by ads over the summer, but now the only political ads are local ones. Looks pretty clear that Romney isn’t going to be investing any more there, for him it’s a lost cause.

    But if he doesn’t have a chance in Pennsylvania, then he’s really got to run the board of the other “battleground” states. And it’s not looking so good for him right now.

    News today says that he has been practicing reciting “zingers” for his debate. He hopes to repeat Ronald Reagan’s peformance against Mondale, or Benson’s performance against Quale. I can’t imagine what he has in mind.

14. 14

    Darryl spews:

    Monday, 10/1/12 at 10:06 am

    Richard,

    “You must have rerun all 50 states (and DC) with 100,000 elections, instead of just running North Dakota 100,000 times.”

    I did run the whole thing again. That is much easier than trying to hand-update each piece (rebuild the maps, all the tables, etc.).

    “Flipping North Dakota would, on average, change the EV totals by almost 6 for each candidate.”

    Only if the state flipped from 100% for one candidate to 100% for the other candidate.

    “But Romney’s total moved from only 191 EV with ND being blue to only 193 EV with ND being red.”

    If you have a cached copy of the pre-revised web page in your browser, you can see where the differences are in the long state table.

    “I take it that running several 100,000 election simulations with the same data will result in the average EV totals varying by several EV’s (like maybe 2 to 5 EV’s) in each simulation?”

    Yes. There is “sampling error” from only running 100,000 simulations. The EVs rarely change by more than 2 from the tests I’ve run. Essentially, the more 50-50 states there are, the more differences between simulations. This follows from the fact that the variance of a binomial process is greatest, for a fixed sample size, at p=0.5. And I am incorporating two sources of binomial variability: (1) from each original poll (that is, each simulated election, I draw a new p from the sampling distribution based on the sample size of the state’s polls) and (2) variability from the simulated voting of a sample.

    BTW: That there is sampling error from only doing 100,000 simulations is why I run a million simulations for single races, like the Inslee—McKenna analyses.

    “Does it require a lot of computer time to run these things nowadays? A lot of your own time?”

    A single run of 100,000 elections requires roughly 5 minutes. When I do the time graph for the first time (say, after changing the “current poll” window) multiply that by 53 for the first seven days. But once I’ve done each day of the week once, the results are saved and I only subsequently re-run the past couple of weeks. That gets messed up whenever I find an older poll to include or change the current poll window size.

    Each of these election analyses posts requires about an hour of my own time. A lot of that is writing about the polls and generating and uploading the extra state poll graphs.

    If I skipped the “new polls” table and discussion, a post would only take about 10 minutes to tweak, once the analysis is done. Of course, hunting down and entering polls takes some time on top of the analyses.

15. 15

    greg spews:

    Monday, 10/1/12 at 11:39 am

    Darrell, here is a US Senate Poll from Hawaii http://www.civilbeat.com/artic.....on-lingle/

16. 16

    Darryl spews:

    Monday, 10/1/12 at 12:14 pm

    greg,

    Thanks! I’ve stuffed it into the database…and not under ND.