| New Polls (Oct. 3) | |||
| State | Poll | Margin | |
|---|---|---|---|
| Nevada | Rasmussen | +4 | |
| New York | Siena | +22 | |
| Rhode Island | Rhode Island College | +14 | |
| Washington | Rasmussen | +10 | |
Both the Silver state and the Tar Heel state have been consistently behind McCain at one point or another during this race. North Carolina has been some shade of red for the entire duration of this map series and Nevada has switched back and forth between the McCain and Obama toss up categories since the early summer months. And on the strength of the Rasmussen poll out in the Silver state today, flips again; this time to Obama.
| Changes (Oct. 3) | |||
| State | Before | After | |
|---|---|---|---|
| Nevada | Toss Up McCain | Toss Up Obama | |
But that brings us to broader discussion we've have had in the comments section here over the last week of so concerning FHQ's methodology. We have been using a weighted average that gives the most weight to the one most recent poll while discounting all the rest back to Super Tuesday. I like having those past polls in there -- that information is valuable, if inflated at the moment -- but they can serve as a drag on certain states that makes the average there sluggish in response to new data. So, while the average, on the whole does a good job of laying out the basic rank ordering of the states in this race, there are some problem areas that need to be addressed.
What are those states and what are the areas? Well, I think it is clear that something is up in both Minnesota and North Carolina that we just aren't capturing with our weighted average in its current configuation. Of the last seven polls in North Carolina, Obama has been either ahead or tied in six. The story is similar in Minnesota. McCain has only been ahead in the Survey USA poll out yesterday, but the Arizona senator has been within three points of Obama in six of the last eight polls there, dating back to September 11. While both states may not be as close as some other sites have them, they are both in my estimation, closer than what the map and the Electoral College Spectrum below indicate. I also think that Florida can be put in this group as well. In nine of the most recent ten polls, Obama has been ahead or tied with the Arizona senator. Now sure, I can be accused of cherrypicking the number of polls I'm looking at in each case, but in each state those polls represent a sizable chunk of the total number of polls in all three. The ten polls in Florida represent just under 20% of the data we have on the Sunshine state. In North Carolina it's a shade under 1 in every six polls. And in Minnesota, we're talking about a group of polls that makes up almost 30% of the total number of polls conducted there this cycle.
In other words, we're seeing a pattern. But the average isn't reacting as quickly as perhaps it should to those changes.
[Click Map to Enlarge]Well, despite that, we see that Obama adds Nevada's five electoral votes to bring his total from yesterday up to 278 electoral votes. But the fact remains that we just don't see many changes around here. That's fine. I don't mind being the among the conservative voices methodologically speaking in the great electoral college debate/discussion being had across the web. However, I do want the numbers to reflect as accurately as possible the actual state of the race. Fine, what are you going to do already, FHQ? As I said in the comments section today, I'll look into this over the weekend -- now that I have a little bit of time -- and will probably initially reexaine the weighting scheme. All polls are not created equally. May polls, for example, mean less now than September polls and our method needs to reflect that. I have done some trial runs of a couple of progressive weighting structures that discount those earlier polls more than the recent ones and the results are promising. This is on a trial basis in just a handful of states, but I want to try and work out the kinks before I use any of them across the board.
| The Electoral College Spectrum* | ||||
| HI-4 (7)** | ME-4 (157) | NH-4 (264/278) | ND-3 (160) | KS-6 (64) |
| VT-3 (10) | WA-11 (168) | CO-9*** (273/274) | WV-5 (157) | AR-6 (58) |
| RI-4 (14) | IA-7 (175) | NV-5 (278/265) | TX-34 (152) | NE-5 (52) |
| IL-21 (35) | OR-7 (182) | OH-20 (298/260) | GA-15 (118) | TN-11 (47) |
| MD-10 (45) | NJ-15 (197) | VA-13 (311/240) | AK-3 (115) | KY-8 (36) |
| DE-3 (48) | MN-10 (207) | FL-27 (338/227) | MS-6 (100) | AL-9 (28) |
| CT-7 (55) | WI-10 (217) | IN-11 (349/200) | SC-8 (94) | WY-3 (19) |
| NY-31 (86) | NM-5 (222) | MO-11 (360/189) | SD-3 (86) | ID-4 (16) |
| CA-55 (141) | MI-17 (239/316) | NC-15 (178) | AZ-10 (83) | OK-7 (12) |
| MA-12 (153) | PA-21 (260/299) | MT-3 (163) | LA-9 (73) | UT-5 (5) |
| *Follow the link for a detailed explanation on how to read the Electoral College Spectrum. **The numbers in the parentheses refer to the number of electoral votes a candidate would have if he won all the states ranked prior to that state. If, for example, McCain won all the states up to and including Pennsylvania (all Obama's toss up states, but Michigan), he would have 299 electoral votes. Both candidates numbers are only totaled through their rival's toss up states. In those cases, Obama's number is on the left and McCain's is on the right in italics. ***Colorado is the state where Obama crosses (or McCain would cross) the 270 electoral vote threshold to win the presidential election. That line is referred to as the victory line. It is currently favoring Obama, thus the blue text in that cell. | ||||
Alright, back to our regularly scheduled electoral map update. The only change on the Electoral College Spectrum today is the same change we saw above; Nevada turning blue...again. If the Obama gains there plateau or continue to rise, these fluctuations will work themselves out. And that goes for Virginia and Ohio as well. The basic idea remains the same though. McCain is on the defensive now that Michigan is off the board. However, if Minnesota is getting closer, unlike Michigan and New Hampshire and Pennsylvania (and to a lesser extent Colorado), then that might be a decent trade. Not ideal, but decent. If McCain holds the states in shades of red, adding Minnesota gets him to 270 exactly. That is a razor-thin victory, but a win is a win. Right Al Gore? Adding Colorado in place of Minnesota would trigger the tiebreaker in the House. But really those are the options now. Well, those two states and that second district in Maine. I can see now where that one could come into play.
Red states + Colorado + Maine's 2nd = 270
That may be easier said than done when Obama is making serious runs at Virginia/North Carolina and Ohio/Indiana.
| The Watch List* | |||
| State | Switch | ||
|---|---|---|---|
| Iowa | from Obama lean | to Strong Obama | |
| Michigan | from Toss Up Obama | to Obama lean | |
| Missouri | from Toss Up McCain | to McCain lean | |
| Nevada | from Toss Up Obama | to Toss Up McCain | |
| North Carolina | from McCain lean | to Toss Up McCain | |
| Ohio | from Toss Up McCain | to Toss Up Obama | |
| Oregon | from Obama lean | to Strong Obama | |
| Pennsylvania | from Toss Up Obama | to Obama lean | |
| Texas | from Strong McCain | to McCain lean | |
| Virginia | from Toss Up McCain | to Toss Up Obama | |
| Washington | from Strong Obama | to Obama lean | |
| *Weighted Average within a fraction of a point of changing categories. | |||
The Watch List's only change today is the potential change Nevada could make given new polling. If there is a shift there toward McCain, the average could push back into the red.
Recent Posts:
The Electoral College Map (10/3/08)
Live Blog and Open Thread: The Vice Presidential Debate
Where is McCain Playing Offense Now that Michigan is Off the Table?
As 538 pointed out, after initially publishing that Elon NC poll, that was really a poll of who voters trusted more on the economy, forcing them to retract it (though not before factoring it into their election simulations for the day).
ReplyDeleteI should have provided a link ti 538's retraction.
ReplyDeleteI should have spelled correctly, too.
ReplyDeleteGood catch, Jack. I got to look at the Elon release in the morning yesterday, but I was in the midst of updating the map for 10/3 then. It did look to have an economic tilt to it, but I never had a chance to go back and examine it more closely.
ReplyDeleteIt is strange that Elon would opt to go this route and not ask the regular presidential question. Now I'm inclined to go back and check their mid-September poll.
But yeah, I'll remove this poll from the data as it isn't applicable to what we're doing here.
Thanks again.
A follow up on the Elon poll:
ReplyDeleteTheir byline indicates that either they do have data on the presidential race directly or they are misleading us.
"The latest survey shows a tight race for the White House"
I'm willing to give them the benefit of the doubt for now. They have yet to post the full pdf of the data breakdown, but that should be coming in the near future. Until that time though, the poll has been removed from the average here at FHQ.
Wait, isn't it 39-39?
ReplyDelete"The poll, conducted Sept. 29 - Oct. 2 by the Elon University Institute for Politics and Public Affairs, found that 39 percent of respondents supported the Republican Party while another 39 percent supported the Democratic Party."
Can't believe I didn't notice that before.
Yeah, but it is rather ambiguous, isn't it? Why not throw an Obama or a McCain in there?
ReplyDeleteAlright, I just checked their last poll's data file. They don't specify candidate names on the presidential question. They don't on the senate or gubernatorial questions either, but they do follow up those questions with ones that do ask respondents whether they approve or disapprove of each of the candidates. This seems strange to me. It also might explain why their undecideds are so high.
That's bizarre, even if most people do know that the Democrat is Barack Obama and the Republican is John McCain. Not a great way to ask the questions at all.
ReplyDeleteIs this what they did in their September poll, too?
Yes, that previous poll is what I'm basing my argument on. Their write up for this one implies that they've done the same here.
ReplyDeleteHere's the data from the mid-September poll.
Maybe that poll should be dropped too. There's no reason to count that one and not include this one. Both asked about the presidential race giving only the party names.
ReplyDeleteI guess you'd have to treat both the same way - either include both (with this at 39-39) or drop them both.
While Pollster still has the poll up in the "blog" section on poll updates, they have pulled it from their list of all the North Carolina polls. I can't verify this, but I suspect they are doing the same thing I'm doing. Namely, waiting on Elon to verify those 39-39 figures before adding the poll back in.
ReplyDeleteJack,
ReplyDeleteHere's a follow up to your Minnesota comments last night:
The Star Tribune has a new poll out (sans presidential question as far as I can tell) that shows Franken up nine in that senate race. A complete switch from the Survey USA poll earlier in the week. I've got to say, and Scott has brought Minnesota up several times in our discussions, but Survey USA has been consistently out of line with where other polls have had Minnesota this whole cycle. Strangely, they were closest to all the other polls just after the GOP convention, showing absolutely no change from a month before.
Anyway, as we've discussed, this same day registration issue in Minnesota makes it a difficult state to poll and Survey USA is either the only firm that has gotten it right or gotten it wrong.
Minnesota is closer than it has been at other times during this race, but I don't think McCain is ahead, nor do I consider Obama to be up by more than 10. I'd put Minnesota in a range between the margin of error and five or six points.