--
6 days until election day.
Showing posts with label electoral college projections. Show all posts
Showing posts with label electoral college projections. Show all posts
Wednesday, October 30, 2024
Tuesday, October 29, 2024
Monday, October 28, 2024
Sunday, October 27, 2024
Saturday, October 26, 2024
Monday, November 12, 2012
2012 Electoral College Wrap Up, Part 1
This is the first part in a two part look at how FHQ's weighted average stacked up in examining the 2012 electoral college. We'll first take a global look at FHQ in the context of the other models out there. Part two will take a micro view of the FHQ model in relation to the electoral college results.
Now that we are nearly a week removed from the re-election of President Obama, FHQ thought it would circle back around and take a look back at how we did in examining the state of play within the electoral college. The answer is not too bad. What was 49 out of 51 correct state-level projections based on our simple weighted average in 2008 morphed into a perfect 51 out of 51 score in 2012.
FHQ was not alone. Drew Linzer (Emory) at Votamatic and Simon Jackman (Stanford) blogging for the Huffington Post and Sam Wang (Princeton) at the Princeton Election Consortium all were either right on or in Wang's case cautiously calling a tie in Florida. [And truth be told, Florida was a tie, but one that consistently -- around FHQ anyway -- ever so slightly favored the president. Again, we're talking about a decreasing fraction of a point as election day approached.] Oh, and Nate Silver of FiveThirtyEight fame pegged it at 332-206, too. This was a great thing for the so-called "quants".
Despite that, there are a couple of notes that are floating around out there and are worth mentioning.
1) FHQ won't take any victory laps because of this.1 Don't get me wrong. It is nice to be a dart and not, say, the board itself, but this actually has very little to do with what FHQ was doing under the hood -- or what any of the above folks were doing, for that matter. If we were all making sausage, then FHQ and the others were merely turning the crank on our various sausage making apparati. The filling -- the polls -- was what really nailed the election projection on the state level.2 Drew first published his model in June. FHQ followed in July. The polls, even through our different lenses. told the story then. 332-206. Over the course of the summer and into the fall, that changed very little. For FHQ, Florida got as close as 0.04 points in favor of the president, but then took a turn back toward Obama. That was it. The Sunshine state was always the only state that ever truly threatened to jump what FHQ calls the partisan line into the Romney group of states. The polls were not only right on the money, but they were overall, pretty consistent. Jim Campbell's argument/observation that the September polls are a better predictor of November election outcomes came to pass. What we got in October was just noise before state-level polling reverted or began reverting to those post-convention, pre-October numbers.
As FHQ asked throughout October, were we witnessing a movement toward Romney in the polls or the typical sort of narrowing (Campbell 2008) that tends to mark the late campaign polls. The latter may not have been the true answer but it was closer than simply talking about Romney's momentum. Tom Holbrook's (1996) equilibrium theory of candidate support through the polls seems to have been the correct lens through which to view the dynamics of the race as election day drew nearer.
Score this one for the polls, then.
2) But where does that leave the models? After all, the sausage maker has some utility, too. Well, FHQ's natural inclination is to piggyback on the above point and state the obvious. The polls were right on and you didn't really need a statistical model -- complex or otherwise -- to accurately project the electoral college. In true self-deprecating fashion (Bear with me. I'll get there.) -- something FHQ is good at -- our little ol' weighted average was accurate enough to get all but two states right in 2008 and every last one in 2012. Again, it was the polls. In fact, if you removed the weighting and took the raw average of all the publicly available polls released on and before election day in all of 2012 you would come up with the same thing: 332-206. As I told Drew over the summer in a brief Twitter exchange, my hope was at that point just after the conventions that the race would tighten up so that we could, in fact, get a true measure of the utility of the more complex statistical models projecting the electoral college. As it stood then -- and how it ended up even with some narrowing -- there was a lot of overlap between the Bayesian models and the more pedestrian averages.
Mind you, I'm not saying that there is no place for these models. Boy, is there. I'm with John Sides on this one: The more models we have, the better off we all are on this sort of thing. Rather, my point is to suggest that the simple averages are a decent baseline. As November 6 approached and the FHQ numbers did not budge in the face of changing information following the Denver debate, I began to think of the FHQ weighted average like the Gary Jacobson measure of congressional candidate quality. Now sure, there have been herculean efforts littering the political science literature to construct multi-point indices of candidate quality, but they don't often perform all that better than Jacobson's simple test. "Has challenger/candidate X held elective office?" That simple, binary variable explains most of the variation in the levels of success that various candidates -- whether challenging an incumbent or vying for an open seat -- have enjoyed across a great number of elections. The multi-point indices only slightly improve the explanatory power.
Now, lord knows, I'm not trying to draw definitive comparisons between the work here at FHQ and Jacobson's oft-cited body of work. Are there parallels? Yes, and I'll leave it at that. Sometimes the best models are the simplest ones. Parsimony counts and to some extent that is what FHQ provides with these electoral college analyses. And again, the reason I was hoping that the polls would tighten as we got closer to election day was to demonstrate just exactly how much better the more complex models were. My expectation was that there would be a noticeable difference between the two. But there wasn't; not in terms of projecting which states would go to which candidates. By other measures, the more complex models wiped the floor with FHQ (as, admittedly, they should have).3
--
The tie that binds all of these models -- if you really want to call the pre-algebra that FHQ does a model -- is a reliance on polling. And that raises a different question as we shift from reviewing 2012 to looking at 2016 and beyond. The quants "won" this one. But it was not without a wide-ranging -- and fruitful, I think -- discussion about the accuracy of polling. The one question that will continue to be worth asking is whether the seemingly perpetually dropping rates of response to public opinion polls continue to drop and what impact that will have. If that continues, then there would almost certainly have to be a tipping point where phone-based polls begin to more consistently miss the mark. The good news moving forward is that the online polls -- whether YouGov, or Google Consumer Surveys or Angus-Reid -- performed quite well in 2012; offering a ray of hope for something beyond phone polls in a time when cell phones are hard to reach and landlines are disappearing.
Still, we are now at a point where pollsters are talking about the "art of polling" as a means of differentiating from other pollsters instead of the overarching science of polling. That has implications. If all pollsters guess wrong about the underlying demographics of the electorate, all the polls are wrong. Of course, the incentive structure is such that pollsters want to find something of a niche that not only separates them from the competition to some extent but helps them crack the code of the true demographic breakdown of the electorate. [Then they can all herd at the end.]
The bottom line remains: these projections are only as good as the polling that serves as the sausage filling. If garbage goes in, then garbage is more likely to come out. On the other hand, if the polling is accurate, then so too are the projections.
--
1 I won't take any victory laps, but I will extend to all of those who have been both loyal and happenstance readers alike a very sincere thank you for spending some or all of election season with us. And yeah, that stretches back to late 2010. Thank you.
2 This is something Harry Enten of the Guardian mentioned via Twitter on Saturday and AAOPR more or less confirmed today.
3 One factor that should be noted here that may separate FHQ from the more involved models is polling variability. 2008 was witness to a great deal of polling variability. The margins in that open seat presidential election jumped around quite a bit more than in 2012 when an incumbent was involved. 2016, in some respects is shaping up as a repeat of 2008. That is even more true if both Hillary Clinton and Joe Biden pass on runs for the Democratic nomination. Both races would be -- at least from our vantage point here three years out -- wide open and influence the polling that is conducted across firms and across states. Yet, even with that unique situation, FHQ lagged just one correctly predicted state -- North Carolina -- behind FiveThirtyEight.
Now that we are nearly a week removed from the re-election of President Obama, FHQ thought it would circle back around and take a look back at how we did in examining the state of play within the electoral college. The answer is not too bad. What was 49 out of 51 correct state-level projections based on our simple weighted average in 2008 morphed into a perfect 51 out of 51 score in 2012.
FHQ was not alone. Drew Linzer (Emory) at Votamatic and Simon Jackman (Stanford) blogging for the Huffington Post and Sam Wang (Princeton) at the Princeton Election Consortium all were either right on or in Wang's case cautiously calling a tie in Florida. [And truth be told, Florida was a tie, but one that consistently -- around FHQ anyway -- ever so slightly favored the president. Again, we're talking about a decreasing fraction of a point as election day approached.] Oh, and Nate Silver of FiveThirtyEight fame pegged it at 332-206, too. This was a great thing for the so-called "quants".
Despite that, there are a couple of notes that are floating around out there and are worth mentioning.
1) FHQ won't take any victory laps because of this.1 Don't get me wrong. It is nice to be a dart and not, say, the board itself, but this actually has very little to do with what FHQ was doing under the hood -- or what any of the above folks were doing, for that matter. If we were all making sausage, then FHQ and the others were merely turning the crank on our various sausage making apparati. The filling -- the polls -- was what really nailed the election projection on the state level.2 Drew first published his model in June. FHQ followed in July. The polls, even through our different lenses. told the story then. 332-206. Over the course of the summer and into the fall, that changed very little. For FHQ, Florida got as close as 0.04 points in favor of the president, but then took a turn back toward Obama. That was it. The Sunshine state was always the only state that ever truly threatened to jump what FHQ calls the partisan line into the Romney group of states. The polls were not only right on the money, but they were overall, pretty consistent. Jim Campbell's argument/observation that the September polls are a better predictor of November election outcomes came to pass. What we got in October was just noise before state-level polling reverted or began reverting to those post-convention, pre-October numbers.
As FHQ asked throughout October, were we witnessing a movement toward Romney in the polls or the typical sort of narrowing (Campbell 2008) that tends to mark the late campaign polls. The latter may not have been the true answer but it was closer than simply talking about Romney's momentum. Tom Holbrook's (1996) equilibrium theory of candidate support through the polls seems to have been the correct lens through which to view the dynamics of the race as election day drew nearer.
Score this one for the polls, then.
2) But where does that leave the models? After all, the sausage maker has some utility, too. Well, FHQ's natural inclination is to piggyback on the above point and state the obvious. The polls were right on and you didn't really need a statistical model -- complex or otherwise -- to accurately project the electoral college. In true self-deprecating fashion (Bear with me. I'll get there.) -- something FHQ is good at -- our little ol' weighted average was accurate enough to get all but two states right in 2008 and every last one in 2012. Again, it was the polls. In fact, if you removed the weighting and took the raw average of all the publicly available polls released on and before election day in all of 2012 you would come up with the same thing: 332-206. As I told Drew over the summer in a brief Twitter exchange, my hope was at that point just after the conventions that the race would tighten up so that we could, in fact, get a true measure of the utility of the more complex statistical models projecting the electoral college. As it stood then -- and how it ended up even with some narrowing -- there was a lot of overlap between the Bayesian models and the more pedestrian averages.
Mind you, I'm not saying that there is no place for these models. Boy, is there. I'm with John Sides on this one: The more models we have, the better off we all are on this sort of thing. Rather, my point is to suggest that the simple averages are a decent baseline. As November 6 approached and the FHQ numbers did not budge in the face of changing information following the Denver debate, I began to think of the FHQ weighted average like the Gary Jacobson measure of congressional candidate quality. Now sure, there have been herculean efforts littering the political science literature to construct multi-point indices of candidate quality, but they don't often perform all that better than Jacobson's simple test. "Has challenger/candidate X held elective office?" That simple, binary variable explains most of the variation in the levels of success that various candidates -- whether challenging an incumbent or vying for an open seat -- have enjoyed across a great number of elections. The multi-point indices only slightly improve the explanatory power.
Now, lord knows, I'm not trying to draw definitive comparisons between the work here at FHQ and Jacobson's oft-cited body of work. Are there parallels? Yes, and I'll leave it at that. Sometimes the best models are the simplest ones. Parsimony counts and to some extent that is what FHQ provides with these electoral college analyses. And again, the reason I was hoping that the polls would tighten as we got closer to election day was to demonstrate just exactly how much better the more complex models were. My expectation was that there would be a noticeable difference between the two. But there wasn't; not in terms of projecting which states would go to which candidates. By other measures, the more complex models wiped the floor with FHQ (as, admittedly, they should have).3
--
The tie that binds all of these models -- if you really want to call the pre-algebra that FHQ does a model -- is a reliance on polling. And that raises a different question as we shift from reviewing 2012 to looking at 2016 and beyond. The quants "won" this one. But it was not without a wide-ranging -- and fruitful, I think -- discussion about the accuracy of polling. The one question that will continue to be worth asking is whether the seemingly perpetually dropping rates of response to public opinion polls continue to drop and what impact that will have. If that continues, then there would almost certainly have to be a tipping point where phone-based polls begin to more consistently miss the mark. The good news moving forward is that the online polls -- whether YouGov, or Google Consumer Surveys or Angus-Reid -- performed quite well in 2012; offering a ray of hope for something beyond phone polls in a time when cell phones are hard to reach and landlines are disappearing.
Still, we are now at a point where pollsters are talking about the "art of polling" as a means of differentiating from other pollsters instead of the overarching science of polling. That has implications. If all pollsters guess wrong about the underlying demographics of the electorate, all the polls are wrong. Of course, the incentive structure is such that pollsters want to find something of a niche that not only separates them from the competition to some extent but helps them crack the code of the true demographic breakdown of the electorate. [Then they can all herd at the end.]
The bottom line remains: these projections are only as good as the polling that serves as the sausage filling. If garbage goes in, then garbage is more likely to come out. On the other hand, if the polling is accurate, then so too are the projections.
--
1 I won't take any victory laps, but I will extend to all of those who have been both loyal and happenstance readers alike a very sincere thank you for spending some or all of election season with us. And yeah, that stretches back to late 2010. Thank you.
2 This is something Harry Enten of the Guardian mentioned via Twitter on Saturday and AAOPR more or less confirmed today.
3 One factor that should be noted here that may separate FHQ from the more involved models is polling variability. 2008 was witness to a great deal of polling variability. The margins in that open seat presidential election jumped around quite a bit more than in 2012 when an incumbent was involved. 2016, in some respects is shaping up as a repeat of 2008. That is even more true if both Hillary Clinton and Joe Biden pass on runs for the Democratic nomination. Both races would be -- at least from our vantage point here three years out -- wide open and influence the polling that is conducted across firms and across states. Yet, even with that unique situation, FHQ lagged just one correctly predicted state -- North Carolina -- behind FiveThirtyEight.
Thursday, November 1, 2012
Live by the 3, Die by the 3 (...and other thoughts on what November 6, 2012 is testing)
FHQ realizes that the race to 270 is down to its last five days and that both (partisan) sides are deeply mired in trench warfare in a seemingly close battle. The resultant pick-your-side, choose-a-winner mentality has also trickled down discussions of how exactly we quantify the state of the race through the polls. And that, I think, is a shame.
Who's right? Rasmussen or PPP?
OR
Which one's nailed it? The national or state polls?
As a social scientist I hate those questions. They aren't the right questions. And they aren't right because rarely are things as black and white as the eventual results we'll likely get next Tuesday (...or early Wednesday).1 [I'm reminded of Seth Masket's book title.] In other words, the tendency -- or the one I've seen permeate my Twitter feed this morning -- to say one is right and the other is wrong, completely eliminates the possibilities that we'll get something that falls in the middle. The true answer, it could be argued, is somewhere between Rasmussen and PPP or somewhere between the national and state polls.
To some extent, the various polling averages that are out there already account for the distance between the numbers from firms like PPP and Rasmussen. Throw them all in the pot and see what kind of soup you end up with. Fair enough. I'll address the state polls versus national polls issue here. Again, the "which one is right?" question seems utterly ridiculous to me. But I think it is clear that FHQ has cast its lot with the state polls; something that puts us -- fair or not -- into the "provides the Democrats with some solace" group of forecasters. The tie that binds all of those models/formulas2 is a reliance on the state polls.
Here at Davidson, and on some other college campuses, during basketball season there is a saying: "Live by the three, die by the three." The style that the Wildcat cagers use can be heavily reliant on making three point baskets if their inside-out game (usually against BCS conference opponents with sizable frontcourts) is not working. The same is true of any model or formula that uses as its basis either just state polls or just national polls. You live and die by the choices made. But that does not make one right and one wrong. It may make one righter and the other wronger in 2012, but that does not mean that that will be the case in four or eight years time when we may or may not see a convergence between the two sets of polls.
What this tells us is that in a close election, one in which the uncertainty of the outcome is high or very high, there is a higher likelihood of a divergence between those two sets of polls. [Actually, that is better as a testable hypothesis than as a statement. But I'm going to allow it to stand alone -- untested -- for the time being. It is something to consider.]
Fine then. We have a close election, uncertainty is high, and there is a discrepancy between the state and national polls with two emerging explanations. One is the one is right, one is wrong view. The other is what I am calling the convenient narrative. The convenient narrative holds that the numbers in the national polls are driven by enthusiastic Republicans in states that are overwhelmingly likely to vote Republican anyway. The other side of that narrative is that Obama is doing enough better than Romney in toss up states to amass more than enough electoral votes to retain the White House.
But let's take all the polarized camps above -- whether Rasmussen vs. PPP or state vs. national polls or convenient narrative vs. one right, one wrong -- and find some middle ground. Let's consider that the true outcome is or might be somewhere in between.
What does that mean?
Well, let's assume that FHQ -- or the underlying average of state polls really -- is overly rosy for Obama and the Democrats. Further, let's assume that the current Gallup numbers (Romney +5, October 22-28) are equally as optimistic but in Romney's and the Republicans' direction. Where is the middle ground there? Roughly, we can say that Romney is about +5 of that national poll average in Gallup. [Both Real Clear Politics and Pollster have the national average breakdown at a 47-47 tie as of this writing.] Eyeballing it, Virginia is right on that 47-47 number. Now, as FHQ has mentioned before, if Virginia is tied, then Obama will win the electoral college if the established order of states in the Electoral College Spectrum is correct (and it has held pretty steady throughout 2012). If that is the overlap between the state and national polls then FHQ is approximately two points more pro-Obama in how the numbers shake out. This is rough -- very rough not to mention somewhat incongruous -- folks, but that puts a seven point gap between FHQ on one end and Gallup on the other. What happens if we shift things three and a half points over toward the middle.
Now, FHQ does a lot of talking about how much 2012 looks like 2004. The above Spectrum looks an awful lot like what things looked like in the polls in 2004 but with there being a few more heavily Republican states now relative to then. Obviously, this puts Romney over the top if he were to win the series of toss up states that are on his side of the partisan line. It obviously also brings into play for the Republican challenger all the states that have been on the tips of everyone's tongues this last week; that series of (as FHQ has them categorized now) Obama lean states like Michigan, Oregon and Pennsylvania.
Of course, this is just one way in which we could reconcile the differences between the national and state polls. Gallup is five points out from the average of national polls at the moment. And perhaps it is not exactly fair to take one poll on one far end of the range of national polls and stack it up against a weighted average of state polls. In our effort to find middle ground a better approach -- not necessarily the best -- is to compare an average of state polls to and average of national polls. Again, if Virginia is the point of overlap between the two series (just as above), then that continues to put FHQ over about two points further in the Obama direction. Splitting that difference to find a middle ground means shifting the averages over by about a point. Contrary to what we saw above, that has a minimal effect. It pushes Florida over to the Romney side of the partisan line, puts Colorado on the Watch List as a state that is only a fraction of a point in Obama's direction, and essentially halves the lead Obama currently has in the FHQ weighted averages. That correction may just be bringing FHQ in line with some of the other polling aggregators out there. And that may not be properly adjusting for how wrong the state polls might be as compared to the national polls.
One final alternative could be to find some position between the two examples above. Instead of adjusting by 3.5 points as was the case in Alternative #1 or by a point as in Alternative #2, let's see what things look like if we adjust the state averages by 2.5 points toward the national polls.
That has the effect of firming up Arizona, Missouri and North Carolina for Romney. It also brings Florida, Colorado and Virginia over to the former Massachusetts governor's side of the partisan line. That reduces the president's cushion significantly. Iowa and Ohio become virtual ties and New Hampshire (and to a lesser extent Nevada) is not that far behind. Those are the Tier 2 states FHQ has been harping on for the last week. That Iowa/New Hampshire/Ohio group could be hugely consequential, but given the number of electoral votes Ohio holds, it would be the potentially decisive chip instead of either of the two other smaller states.
This alternative also brings those current Lean Obama states (Michigan, Oregon and Pennsylvania) into play but on the periphery of the category. Still, a toss up is a toss up. Finally, the Lean Obama group of states is really compressed; down to just two states. [The same compression can be seen on the Romney side of the partisan line as well.]
--
What does all of this mean? Well, it is a rough glance at how we can reconcile the difference between the state and national polls without assuming that one of them is right and the other is necessarily wrong. More or less, this is an attempt to split the difference in some way, shape or form between the two; to give us some alternatives -- and likely alternatives -- to consider as we wend our way toward Tuesday.
--
1 This assumes that something other than an electoral college/popular vote split, or a disputed election/recall occurs.
2 The weighted average FHQ uses can hardly be considered a model by conventional academic or statistical standards. It is a simple weighted average; no more, no less. Furthermore, it is a weighted average with some limitations. Mainly, the formula tends to lag just a little behind where the polls have moved. That is good and bad as we have discussed before. On the one hand, it grounds what can be wild fluctuations in the polls overall. On the other, the average tends to be a little less reactive than, perhaps, reflects reality, but definitely a little less than FHQ would like.
Who's right? Rasmussen or PPP?
OR
Which one's nailed it? The national or state polls?
As a social scientist I hate those questions. They aren't the right questions. And they aren't right because rarely are things as black and white as the eventual results we'll likely get next Tuesday (...or early Wednesday).1 [I'm reminded of Seth Masket's book title.] In other words, the tendency -- or the one I've seen permeate my Twitter feed this morning -- to say one is right and the other is wrong, completely eliminates the possibilities that we'll get something that falls in the middle. The true answer, it could be argued, is somewhere between Rasmussen and PPP or somewhere between the national and state polls.
To some extent, the various polling averages that are out there already account for the distance between the numbers from firms like PPP and Rasmussen. Throw them all in the pot and see what kind of soup you end up with. Fair enough. I'll address the state polls versus national polls issue here. Again, the "which one is right?" question seems utterly ridiculous to me. But I think it is clear that FHQ has cast its lot with the state polls; something that puts us -- fair or not -- into the "provides the Democrats with some solace" group of forecasters. The tie that binds all of those models/formulas2 is a reliance on the state polls.
Here at Davidson, and on some other college campuses, during basketball season there is a saying: "Live by the three, die by the three." The style that the Wildcat cagers use can be heavily reliant on making three point baskets if their inside-out game (usually against BCS conference opponents with sizable frontcourts) is not working. The same is true of any model or formula that uses as its basis either just state polls or just national polls. You live and die by the choices made. But that does not make one right and one wrong. It may make one righter and the other wronger in 2012, but that does not mean that that will be the case in four or eight years time when we may or may not see a convergence between the two sets of polls.
What this tells us is that in a close election, one in which the uncertainty of the outcome is high or very high, there is a higher likelihood of a divergence between those two sets of polls. [Actually, that is better as a testable hypothesis than as a statement. But I'm going to allow it to stand alone -- untested -- for the time being. It is something to consider.]
Fine then. We have a close election, uncertainty is high, and there is a discrepancy between the state and national polls with two emerging explanations. One is the one is right, one is wrong view. The other is what I am calling the convenient narrative. The convenient narrative holds that the numbers in the national polls are driven by enthusiastic Republicans in states that are overwhelmingly likely to vote Republican anyway. The other side of that narrative is that Obama is doing enough better than Romney in toss up states to amass more than enough electoral votes to retain the White House.
But let's take all the polarized camps above -- whether Rasmussen vs. PPP or state vs. national polls or convenient narrative vs. one right, one wrong -- and find some middle ground. Let's consider that the true outcome is or might be somewhere in between.
What does that mean?
Well, let's assume that FHQ -- or the underlying average of state polls really -- is overly rosy for Obama and the Democrats. Further, let's assume that the current Gallup numbers (Romney +5, October 22-28) are equally as optimistic but in Romney's and the Republicans' direction. Where is the middle ground there? Roughly, we can say that Romney is about +5 of that national poll average in Gallup. [Both Real Clear Politics and Pollster have the national average breakdown at a 47-47 tie as of this writing.] Eyeballing it, Virginia is right on that 47-47 number. Now, as FHQ has mentioned before, if Virginia is tied, then Obama will win the electoral college if the established order of states in the Electoral College Spectrum is correct (and it has held pretty steady throughout 2012). If that is the overlap between the state and national polls then FHQ is approximately two points more pro-Obama in how the numbers shake out. This is rough -- very rough not to mention somewhat incongruous -- folks, but that puts a seven point gap between FHQ on one end and Gallup on the other. What happens if we shift things three and a half points over toward the middle.
The Electoral College Spectrum1 -- Alternative #1
| ||||
VT-3
(6)2
|
WA-12
(158)
|
NH-4
(257/285)
|
GA-16
(167)
|
ND-3
(55)
|
HI-4
(10)
|
NJ-14
(172)
|
OH-183
(275/281)
|
SD-3
(151)
|
KY-8
(52)
|
NY-29
(39)
|
CT-7
(179)
|
IA-6
(281/263)
|
IN-11
(148)
|
AL-9
(44)
|
RI-4
(43)
|
NM-5
(184)
|
VA-13
(294/257)
|
SC-9
(137)
|
KS-6
(35)
|
MD-10
(53)
|
MN-10
(194)
|
CO-9
(303/244)
|
TN-11
(128)
|
AR-6
(29)
|
IL-20
(73)
|
OR-7
(201/344)
|
FL-29
(332/235)
|
NE-5
(117)
|
AK-3
(23)
|
MA-11
(84)
|
PA-20
(221/337)
|
NC-15
(206)
|
TX-38
(112)
|
OK-7
(20)
|
CA-55
(139)
|
MI-16
(237/317)
|
AZ-11
(191)
|
WV-5
(74)
|
ID-4
(13)
|
DE-3
(142)
|
WI-10
(247/301)
|
MO-10
(180)
|
LA-8
(69)
|
WY-3
(9)
|
ME-4
(146)
|
NV-6
(253/291)
|
MT-3
(170)
|
MS-6
(61)
|
UT-6
(6)
|
1 Follow the link for a detailed explanation on how to read the Electoral College Spectrum.
2 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, Romney won all the states up to and including Ohio (all Obama's toss up states plus Ohio), he would have 281 electoral votes. Romney's numbers are only totaled through the states he would need in order to get to 270. In those cases, Obama's number is on the left and Romney's is on the right in italics. 3 Ohio is the state where Obama crosses the 270 electoral vote threshold to win the presidential election. That line is referred to as the victory line. | ||||
Now, FHQ does a lot of talking about how much 2012 looks like 2004. The above Spectrum looks an awful lot like what things looked like in the polls in 2004 but with there being a few more heavily Republican states now relative to then. Obviously, this puts Romney over the top if he were to win the series of toss up states that are on his side of the partisan line. It obviously also brings into play for the Republican challenger all the states that have been on the tips of everyone's tongues this last week; that series of (as FHQ has them categorized now) Obama lean states like Michigan, Oregon and Pennsylvania.
Of course, this is just one way in which we could reconcile the differences between the national and state polls. Gallup is five points out from the average of national polls at the moment. And perhaps it is not exactly fair to take one poll on one far end of the range of national polls and stack it up against a weighted average of state polls. In our effort to find middle ground a better approach -- not necessarily the best -- is to compare an average of state polls to and average of national polls. Again, if Virginia is the point of overlap between the two series (just as above), then that continues to put FHQ over about two points further in the Obama direction. Splitting that difference to find a middle ground means shifting the averages over by about a point. Contrary to what we saw above, that has a minimal effect. It pushes Florida over to the Romney side of the partisan line, puts Colorado on the Watch List as a state that is only a fraction of a point in Obama's direction, and essentially halves the lead Obama currently has in the FHQ weighted averages. That correction may just be bringing FHQ in line with some of the other polling aggregators out there. And that may not be properly adjusting for how wrong the state polls might be as compared to the national polls.
The Electoral College Spectrum1 -- Alternative #2
| ||||
VT-3
(6)2
|
WA-12
(158)
|
NH-4
(257)
|
GA-16
(167)
|
ND-3
(55)
|
HI-4
(10)
|
NJ-14
(172)
|
OH-183
(275/281)
|
SD-3
(151)
|
KY-8
(52)
|
NY-29
(39)
|
CT-7
(179)
|
IA-6
(281/263)
|
IN-11
(148)
|
AL-9
(44)
|
RI-4
(43)
|
NM-5
(184)
|
VA-13
(294/257)
|
SC-9
(137)
|
KS-6
(35)
|
MD-10
(53)
|
MN-10
(194)
|
CO-9
(303/244)
|
TN-11
(128)
|
AR-6
(29)
|
IL-20
(73)
|
OR-7
(201)
|
FL-29
(332/235)
|
NE-5
(117)
|
AK-3
(23)
|
MA-11
(84)
|
PA-20
(221)
|
NC-15
(206)
|
TX-38
(112)
|
OK-7
(20)
|
CA-55
(139)
|
MI-16
(237)
|
AZ-11
(191)
|
WV-5
(74)
|
ID-4
(13)
|
DE-3
(142)
|
WI-10
(247)
|
MO-10
(180)
|
LA-8
(69)
|
WY-3
(9)
|
ME-4
(146)
|
NV-6
(253)
|
MT-3
(170)
|
MS-6
(61)
|
UT-6
(6)
|
1 Follow the link for a detailed explanation on how to read the Electoral College Spectrum.
2 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, Romney won all the states up to and including Ohio (all Obama's toss up states plus Ohio), he would have 281 electoral votes. Romney's numbers are only totaled through the states he would need in order to get to 270. In those cases, Obama's number is on the left and Romney's is on the right in italics. 3 Ohio is the state where Obama crosses the 270 electoral vote threshold to win the presidential election. That line is referred to as the victory line. | ||||
One final alternative could be to find some position between the two examples above. Instead of adjusting by 3.5 points as was the case in Alternative #1 or by a point as in Alternative #2, let's see what things look like if we adjust the state averages by 2.5 points toward the national polls.
The Electoral College Spectrum1 -- Alternative #3
| ||||
VT-3
(6)2
|
WA-12
(158)
|
NH-4
(257)
|
GA-16
(167)
|
ND-3
(55)
|
HI-4
(10)
|
NJ-14
(172)
|
OH-183
(275/281)
|
SD-3
(151)
|
KY-8
(52)
|
NY-29
(39)
|
CT-7
(179)
|
IA-6
(281/263)
|
IN-11
(148)
|
AL-9
(44)
|
RI-4
(43)
|
NM-5
(184)
|
VA-13
(294/257)
|
SC-9
(137)
|
KS-6
(35)
|
MD-10
(53)
|
MN-10
(194)
|
CO-9
(303/244)
|
TN-11
(128)
|
AR-6
(29)
|
IL-20
(73)
|
OR-7
(201)
|
FL-29
(332/235)
|
NE-5
(117)
|
AK-3
(23)
|
MA-11
(84)
|
PA-20
(221)
|
NC-15
(206)
|
TX-38
(112)
|
OK-7
(20)
|
CA-55
(139)
|
MI-16
(237)
|
AZ-11
(191)
|
WV-5
(74)
|
ID-4
(13)
|
DE-3
(142)
|
WI-10
(247)
|
MO-10
(180)
|
LA-8
(69)
|
WY-3
(9)
|
ME-4
(146)
|
NV-6
(253)
|
MT-3
(170)
|
MS-6
(61)
|
UT-6
(6)
|
1 Follow the link for a detailed explanation on how to read the Electoral College Spectrum.
2 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, Romney won all the states up to and including Ohio (all Obama's toss up states plus Ohio), he would have 281 electoral votes. Romney's numbers are only totaled through the states he would need in order to get to 270. In those cases, Obama's number is on the left and Romney's is on the right in italics. 3 Ohio is the state where Obama crosses the 270 electoral vote threshold to win the presidential election. That line is referred to as the victory line. | ||||
That has the effect of firming up Arizona, Missouri and North Carolina for Romney. It also brings Florida, Colorado and Virginia over to the former Massachusetts governor's side of the partisan line. That reduces the president's cushion significantly. Iowa and Ohio become virtual ties and New Hampshire (and to a lesser extent Nevada) is not that far behind. Those are the Tier 2 states FHQ has been harping on for the last week. That Iowa/New Hampshire/Ohio group could be hugely consequential, but given the number of electoral votes Ohio holds, it would be the potentially decisive chip instead of either of the two other smaller states.
This alternative also brings those current Lean Obama states (Michigan, Oregon and Pennsylvania) into play but on the periphery of the category. Still, a toss up is a toss up. Finally, the Lean Obama group of states is really compressed; down to just two states. [The same compression can be seen on the Romney side of the partisan line as well.]
--
What does all of this mean? Well, it is a rough glance at how we can reconcile the difference between the state and national polls without assuming that one of them is right and the other is necessarily wrong. More or less, this is an attempt to split the difference in some way, shape or form between the two; to give us some alternatives -- and likely alternatives -- to consider as we wend our way toward Tuesday.
--
1 This assumes that something other than an electoral college/popular vote split, or a disputed election/recall occurs.
2 The weighted average FHQ uses can hardly be considered a model by conventional academic or statistical standards. It is a simple weighted average; no more, no less. Furthermore, it is a weighted average with some limitations. Mainly, the formula tends to lag just a little behind where the polls have moved. That is good and bad as we have discussed before. On the one hand, it grounds what can be wild fluctuations in the polls overall. On the other, the average tends to be a little less reactive than, perhaps, reflects reality, but definitely a little less than FHQ would like.
Tuesday, November 11, 2008
2008 Electoral College Wrap Up
"How did we do?"
In the last week, several of our electoral college analyst brethren have asked about the level of accuracy each achieved. Let's have a look at how well and/or poorly FHQ did in that regard. [What, you thought we were going to be any different?] Below, you see how the race actually played out on election day, except now we've added some gradations to reflect states where candidates won by a substantial margin or where the final spread between Obama and McCain ended up being narrow.
[Click Map to Enlarge]
As Nate Silver explained recently, Obama could have given 9.3 points on average back to John McCain in every state and still have come away from Tuesday night's election with an electoral college tie. We have spoken time and again about the electoral college cushion Obama had in this race, but we have done so in terms of how many states past the victory line Obama's campaign was able to push. If George W. Bush would have given 9.3 points to John Kerry in 2004, Kerry would have been able to snatch up Florida, Ohio, Nevada, Arkansas, West Virginia, Virginia and Colorado to get to around the same number of electoral votes (360 EVs) Obama had in 2008 (365 EVs).
We knew, for instance, that when and if New Hampshire and Pennsylvania went for Obama on election day, that all it was going to take, given where Obama was likely to do well the rest of the evening, to push Obama over the top was the victory line state (Colorado) or some state below it. So when Ohio fell into Obama's column the race was over. And that can be seen on the Electoral College Spectrum for the final results above.
Great, FHQ's weighted averages granted us the ability to see that Obama would win and what states to watch on election night. Lots of people saw that coming. But where did FHQ fail to capture in its average what actually happened on election day?
[Click Map to Enlarge]
There were a handful of states that FHQ missed (...as did several other outlets). Most notably, North Carolina and Indiana turned blue for the first time in decades. We had discussed North Carolina at length during the last month of the campaign and the Tar Heel state's average had crept closer and closer to a complete tie in that time. But it never moved into the blue for Obama. Between the average and the actual outcome, North Carolina moved about a point on election day. So, while North Carolina wasn't correctly predicted, the result wasn't out of left field either. The state was already close and on the Watch List for a potential switch toward Obama.
Indiana, on the other hand, was a bit of a surprise given where the graduated weighted average has the Hoosier state ranked on the Spectrum. Consistently on the McCain side of both Missouri and North Carolina, Indiana jumped over two points on election day (from where FHQ's average placed the state and where it ended up after the votes were counted). Heading into the day, Missouri looked much more likely to end up on Obama's side of the ledger than Indiana. What's strange is how both those states split their votes between the presidential and gubernatorial level. Missouri gave McCain its 11 electoral votes while electing a Democratic governor and Indiana provided Obama with a narrow margin and at the same time reelected a Republican governor. Yes, local factors played a role in each case, but that's still an interesting occurrence.
[The final electoral vote from Nebraska's 2nd congressional district was one that was never accounted for in our averages. Now that the first split allocation of electoral votes has occurred, that may be something that FHQ will have to attempt to factor in in subsequent cycles. But we'll talk about possible improvements momentarily.]
Also, when we compare the predicted map and Spectrum to the actual results above we find that while several states were correctly predicted, they were either more or less competitive than our averages would have let on. On the McCain end, Alaska, Arkansas, Louisiana and West Virginia ended up being less competitive than expected while South Dakota and Indiana and North Carolina, obviously, were closer to Obama on Tuesday than had been predicted.
On the Obama side, there were several states that were "off" in terms of how their averages and results matched up, but the rank ordering fell pretty much in line with what had been expected. Nevada ended up being much less competitive then the polling in the Silver state otherwise would have indicated. As UNLV political scientist, Dave Damore, told FHQ back in September, pollsters tend to oversample the the rural and more Republican areas of Nevada which in 2004 meant support Bush was overestimated in the polls conducted in the state. But even adding that 4-5 points to FHQ's average falls short of where the Silver state fell on November 4. What was the deal then? Well, it could be that we didn't have enough information on Nevada -- it certainly had fewer polls conducted within the state lines than many of the other toss up states -- or it could be that rural/Republican oversampling really overestimated McCain's support in the state.
[Click Figure to Enlarge]
But how well did FHQ's averages match up with where the individual states actually fell on election day? A simple bivariate regression with our averages as the explanatory variable and the actual results as the dependent variable show that the averages explained over 95% of the variation in the vote margins witnessed on election day. All 50 states are clustered pretty tightly around that regression line above. But how closely? And which states were problematic?
[Click Figure to Enlarge]
We can eyeball it or we can add a 95% confidence interval to the plot above. Sure, you can see that Alaska and Hawaii are outliers in that original scatterplot, but are there states that fall outside of that confidence interval? There are and we come full circle with the earlier discussion of Nevada. One of the potential problems with the Silver state that I mentioned was that there were fewer polls there than in other toss up states. If you look at the states that fall outside of the gray area in the second plot, you see that most of them are less competitive and thus less frequently polled states. That indicates that some sort of repeated simulation -- akin to what FiveThirtyEight, the Princeton Election Consortium or Hominid Views use -- could be useful in providing more information on those states and a greater level of confidence in their averages. Ah, something to work on for 2012. Isn't that just copying them? Yeah, but FHQ would remain different in that it would include all the older polls in a given cycle while the others phase them out gradually or focus on only the more recent ones.
On the whole, though, this first run in 2008 was a relatively successful one for FHQ in terms of the electoral college. 48 of the 50 states were correctly predicted with a simple weighted average and one of those two, North Carolina, was certainly within range of a switch heading into election day.
Recent Posts:
How Stuff Works: An Alaska Vacancy in the US Senate
More on the Georgia Senate Runoff
Omaha to Obama
In the last week, several of our electoral college analyst brethren have asked about the level of accuracy each achieved. Let's have a look at how well and/or poorly FHQ did in that regard. [What, you thought we were going to be any different?] Below, you see how the race actually played out on election day, except now we've added some gradations to reflect states where candidates won by a substantial margin or where the final spread between Obama and McCain ended up being narrow.
[Click Map to Enlarge]As Nate Silver explained recently, Obama could have given 9.3 points on average back to John McCain in every state and still have come away from Tuesday night's election with an electoral college tie. We have spoken time and again about the electoral college cushion Obama had in this race, but we have done so in terms of how many states past the victory line Obama's campaign was able to push. If George W. Bush would have given 9.3 points to John Kerry in 2004, Kerry would have been able to snatch up Florida, Ohio, Nevada, Arkansas, West Virginia, Virginia and Colorado to get to around the same number of electoral votes (360 EVs) Obama had in 2008 (365 EVs).
| The Electoral College Spectrum* | ||||
| HI-4 (7)** | ME-4 (157) | NH-4 (262/279) | GA-15 (159) | NE-4 (58) |
| VT-3 (10) | WA-11 (168) | IA-7 (269/274) | SD-3 (144) | KY-8 (54) |
| RI-4 (14) | MI-17 (185) | CO-9*** (278/269) | ND-3 (141) | LA-9 (46) |
| MA-12 (26) | OR-7 (192) | VA-13 (291/260) | AZ-10 (138) | AR-6 (37) |
| NY-31 (57) | NJ-15 (207) | OH-20 (311/247) | SC-8 (128) | AL-9 (31) |
| DE-3 (60) | NM-5 (212) | FL-27 (338/227) | TX-34 (120) | AK-3 (22) |
| IL-21 (81) | WI-10 (222) | IN-11 (349/200) | WV-5 (86) | ID-4 (19) |
| MD-10 (91) | NV-5 (227) | NC-15+1**** (365/189) | MS-6 (81) | UT-5 (15) |
| CA-55 (146) | PA-21 (248) | MO-11 (173) | TN-11 (75) | OK-7 (10) |
| CT-7 (153) | MN-10 (258) | MT-3 (162) | KS-6 (64) | WY-3 (3) |
| *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 Colorado (all Obama's toss up states plus Colorado), he would have 269 electoral votes. McCain's numbers are only totaled through the states he would have needed in order to get to 270. 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 crossed the 270 electoral vote threshold to win the presidential election. That line is referred to as the victory line. ****Nebraska allocates electoral votes based on statewide results and the results within each of its congressional districts. Nebraska's 2nd district voted for Barack Obama on November 4. | ||||
We knew, for instance, that when and if New Hampshire and Pennsylvania went for Obama on election day, that all it was going to take, given where Obama was likely to do well the rest of the evening, to push Obama over the top was the victory line state (Colorado) or some state below it. So when Ohio fell into Obama's column the race was over. And that can be seen on the Electoral College Spectrum for the final results above.
Great, FHQ's weighted averages granted us the ability to see that Obama would win and what states to watch on election night. Lots of people saw that coming. But where did FHQ fail to capture in its average what actually happened on election day?
[Click Map to Enlarge]There were a handful of states that FHQ missed (...as did several other outlets). Most notably, North Carolina and Indiana turned blue for the first time in decades. We had discussed North Carolina at length during the last month of the campaign and the Tar Heel state's average had crept closer and closer to a complete tie in that time. But it never moved into the blue for Obama. Between the average and the actual outcome, North Carolina moved about a point on election day. So, while North Carolina wasn't correctly predicted, the result wasn't out of left field either. The state was already close and on the Watch List for a potential switch toward Obama.
Indiana, on the other hand, was a bit of a surprise given where the graduated weighted average has the Hoosier state ranked on the Spectrum. Consistently on the McCain side of both Missouri and North Carolina, Indiana jumped over two points on election day (from where FHQ's average placed the state and where it ended up after the votes were counted). Heading into the day, Missouri looked much more likely to end up on Obama's side of the ledger than Indiana. What's strange is how both those states split their votes between the presidential and gubernatorial level. Missouri gave McCain its 11 electoral votes while electing a Democratic governor and Indiana provided Obama with a narrow margin and at the same time reelected a Republican governor. Yes, local factors played a role in each case, but that's still an interesting occurrence.
[The final electoral vote from Nebraska's 2nd congressional district was one that was never accounted for in our averages. Now that the first split allocation of electoral votes has occurred, that may be something that FHQ will have to attempt to factor in in subsequent cycles. But we'll talk about possible improvements momentarily.]
| The Electoral College Spectrum* | ||||
| HI-4 (7)** | ME-4 (157) | NM-5 (264) | ND-3 (381/160) | AK-3 (61) |
| VT-3 (10) | OR-7 (164) | CO-9*** (273/274) | GA-15 (157) | KY-8 (58) |
| DE-3 (13) | WA-11 (175) | VA-13 (286/265) | WV-5 (142) | TN-11 (50) |
| NY-31 (44) | NJ-15 (190) | NV-5 (291/252) | AZ-10 (137) | KS-6 (39) |
| IL-21 (65) | IA-7 (197) | OH-20 (311/247) | SD-3 (127) | NE-5 (33) |
| MD-10 (75) | WI-10 (207) | FL-27 (338/227) | LA-9 (124) | AL-9 (28) |
| RI-4 (79) | MN-10 (217) | NC-15 (353/200) | AR-6 (115) | WY-3 (19) |
| MA-12 (91) | PA-21 (238) | MO-11 (364/185) | TX-34 (109) | ID-4 (16) |
| CA-55 (146) | MI-17 (255) | IN-11 (375/174) | MS-6 (75) | UT-5 (12) |
| CT-7 (153) | NH-4 (259) | MT-3 (378/163) | SC-8 (69) | OK-7 (7) |
| *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 Colorado (all Obama's toss up states plus Colorado), he would have 274 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. | ||||
Also, when we compare the predicted map and Spectrum to the actual results above we find that while several states were correctly predicted, they were either more or less competitive than our averages would have let on. On the McCain end, Alaska, Arkansas, Louisiana and West Virginia ended up being less competitive than expected while South Dakota and Indiana and North Carolina, obviously, were closer to Obama on Tuesday than had been predicted.
On the Obama side, there were several states that were "off" in terms of how their averages and results matched up, but the rank ordering fell pretty much in line with what had been expected. Nevada ended up being much less competitive then the polling in the Silver state otherwise would have indicated. As UNLV political scientist, Dave Damore, told FHQ back in September, pollsters tend to oversample the the rural and more Republican areas of Nevada which in 2004 meant support Bush was overestimated in the polls conducted in the state. But even adding that 4-5 points to FHQ's average falls short of where the Silver state fell on November 4. What was the deal then? Well, it could be that we didn't have enough information on Nevada -- it certainly had fewer polls conducted within the state lines than many of the other toss up states -- or it could be that rural/Republican oversampling really overestimated McCain's support in the state.
[Click Figure to Enlarge]But how well did FHQ's averages match up with where the individual states actually fell on election day? A simple bivariate regression with our averages as the explanatory variable and the actual results as the dependent variable show that the averages explained over 95% of the variation in the vote margins witnessed on election day. All 50 states are clustered pretty tightly around that regression line above. But how closely? And which states were problematic?
[Click Figure to Enlarge]We can eyeball it or we can add a 95% confidence interval to the plot above. Sure, you can see that Alaska and Hawaii are outliers in that original scatterplot, but are there states that fall outside of that confidence interval? There are and we come full circle with the earlier discussion of Nevada. One of the potential problems with the Silver state that I mentioned was that there were fewer polls there than in other toss up states. If you look at the states that fall outside of the gray area in the second plot, you see that most of them are less competitive and thus less frequently polled states. That indicates that some sort of repeated simulation -- akin to what FiveThirtyEight, the Princeton Election Consortium or Hominid Views use -- could be useful in providing more information on those states and a greater level of confidence in their averages. Ah, something to work on for 2012. Isn't that just copying them? Yeah, but FHQ would remain different in that it would include all the older polls in a given cycle while the others phase them out gradually or focus on only the more recent ones.
On the whole, though, this first run in 2008 was a relatively successful one for FHQ in terms of the electoral college. 48 of the 50 states were correctly predicted with a simple weighted average and one of those two, North Carolina, was certainly within range of a switch heading into election day.
Recent Posts:
How Stuff Works: An Alaska Vacancy in the US Senate
More on the Georgia Senate Runoff
Omaha to Obama
Subscribe to:
Posts (Atom)