I write this section very cautiously, because I am taking on one of my intellectual heroes. Let me say from the outset that Michael Barone has forgotten more than I’ve ever known about politics, and most everything that I do know about politics, I owe to my complete set of Almanacs of American Politics (1972-2006). Heck, I even own a dog-eared copy of “Our Country,” his (wonderful) political history of America from 1928 to 1988.
But I can’t agree with his repeated suggestion that there is a new list of target states for Obama and McCain, or that the map is likely to look especially different this time around from 2004. This isn’t to say that there won’t be slight differences – Obama’s strength in the Carolinas and the Northern Plains states may well be the real deal, but overall I think this base map is going to look very familiar when all is said and done.
The maps below are not the normal red/blue maps for the elections. They are PVI maps. These maps show the strength of the parties if the election was shifted to make it a fifty-fifty election. In other words, they control for the election-to-election changes in strength of the parties. If the Republicans win an election nationally 60-40, and the state is 60-40, the state shows up as purple. If a state is 50-50, it shows as blue. The reddest states are R+10 states, relatively speaking, while the bluest are D+10.
I do have some qualms with this approach, as it assumes the national trends are what swings states, while it may be the other way around. In other words, Reagan won the nation 59-41 in 1984 because even MA was voting Republican in 1984; he didn’t win MA because the nation was voting 59-41. To put it another way, one can say that MN, WI, and IA have moved to the right in the last two decades, as they voted for Democrats even in landslide years, but have been key battlegrounds in the previous elections. On the other hand, one could say that they have stayed in the same place, while other states have moved to the left and out of reach, making them the battlegrounds. Regardless, given the basic stability of the underlying maps, I think there is some merit to this approach.
Making The Modern Map
With that out of the way, let’s take a look at the most recent election map:
Note the salient features: A blue west coast with Oregon somewhat less blue. A bright red Western interior, with CO, NM, and NV as purple, and AZ reddish-purple. A column of purple abutting the Mississippi, getting progressively redder as we move south. A red Southeast, with VA, NC, and FL somewhat less red. A purple industrial Midwest, with IN and IL as outliers. And finally, a blue Northeast, with NH and to a lesser degree, ME as outliers.
How did this come to pass?
We begin 100 years ago, when the map looked quite different:
Note the solid blue Southern tier and the solid red Northern tier. I will also admit operator error here: AZ, NM and OK should not be states; their colors are holdovers from the 1932 map, which is when I decided to go back and do 1904. I’m too lazy to fix it now.
Regardless, in 1904, we still had the basic Civil War split to American politics. IN and NY were bluer than most states (notwithstanding New Yorker TR being at the top of the ticket), the latter owing to the Democrats’ dominance of Manhatten, the former due to the fact that Southern IN is basically an extension of the South.
From here, the map development is shown in this animated .gif
Note that 1920, 1924, and 1928 are variants of the same map, but with some important differences that I described above. 1920 is a Republican blowout year. In 1924 – another big R win – the strong showing of LaFollette’s Progressive party weakens Republicans in the Upper Midwest and California. In 1928, the heavy turnout of ethnic Catholics in NY, MA, and RI turned those states dark blue (Smith actually carried only MA and RI, and narrowly at that, but given the national environment, it was remarkable). That surge in turnout also elected, very narrowly, Franklin Delano Roosevelt as Governor of New York.
Roosevelt’s strategy again was to take the best of 1924 and 1928 and add it, along with economically distressed farmers, to the 1920 map. While NY, RI and MA did not swing as heavily as the rest of the country, they still went comfortably Democrat. Republicans lost the Democratic states from the 1920 map, the Progressive West and Upper Midwest, the farmers on the Plains, and much of ethnic New England.
Throughout the 30s and 40s, we see the Plains and Mountain West revert back to their Republican roots. In 1936, we also begin to witness Virginia being somewhat more Republican than the rest of the South – though it is still very, very Democratic. And in 1948 – the year the Civil Rights plank was added to the Democratic platform, we see the South begin to purple (even though Tom Dewey was an ardent civil rights supporter).
By the 1956 election, we see a map that looks much more like the modern map:
Except for the swath of states from Arkansas east to Georgia, the South has purpled, with Virginia the reddest. Remember, Ike carried five Southern states that year, and came a few points from adding NC and AR and thereby becoming the Republican since Reconstruction to carry a majority of the Confederate states. The Mountain West and Northern Plains are almost as red as the fading Yankee redoubts of ME and VT, while states like MA and RI are only slightly more Republican than FL and VA. This in the year that he was endorsed by Harlem minister and Congressman Adam Clayton Powell, and when Democrats ran a candidate who was uncomfortable with civil rights, unfriendly to labor unions, and who openly opposed using federal troops to enforce Brown v. Board.
1960 resembles the modern map even more.
We see a purple West coast, with OR as the most Republican state. We see a reddish Mountain West and Northern Plains, with NV and NM “bluer” than the rest of the region. The industrial Midwest and Upper Midwest are purple, with OH, IN (and IA, at that time) the reddest. VA and TN are well redder than the rest of the South. MA, RI, and NY are very blue. If not for the deep South and VT/ME, a casual observer might mistake this for the modern map. But it was drawn four years even before the passage of the Civil Rights Act of 1964.
And so it goes. In the 1964 landslide, we see a purple West Coast (with OR somewhere to the right of CA and WA), a red interior (though now CO joins NM and NV as “purple” states), a bright red South, a bluish-purple industrial Midwest and Upper Midwest (now with OH as purple and IN as reddish), and a blue New England, except perhaps for NH, which is pretty purple. Again, this is the basic map we see today. It only needed some refining.
The two Nixon elections offer more of the same: OR to the right on a purple West Coast, NM, CO and NV (and at the time, MT) as purple states in the red Mountain West, a reddish-purple South, a bluish-purple Midwest and Upper Midwest, and a Northeast that is increasingly blue, with the exception of New England. Indeed, but for the blued Upper Midwest and Northern Plains states, due to McGovern’s presence on the ticket, and the purpled New England (save MA, which was blue), and OR being to CA and WA’s left, one could be forgiven for mistaking this for the modern map.
1976 and 1980 saw a somewhat different map, thanks in part to Jimmy Carter’s strength in the South, and Ford’ last gasp of appeal to dying New England Yankees. Reagan’s 1980 map looks quite similar to Ford’s 1976 map, except with Republican performance improved substantially. In 1984 and 1988, the map reverts to form, while in 1992, Clinton performs well in the South.
But by 1996, the present map is very much in place. Note especially the Mountain West, with CO, NM, AZ and NV barely redder than the nation (roughly the same relative position that they’d occupied since the 1960s, with the exception of AZ), a column of red from TX to ND, a purpler column along the MS river (although Clinton obviously ran better in AR than did either Kerry or Gore), a red South with a purple FL, and now, a less-red VA, a purplish-blue industrial Midwest (save IN), with IL the bluest state and OH quite purple, and a very blue New England, with the exception of NH. This is a moderate version of the map we saw in 2000 and 2004. The blue states got bluer, the red states got redder, but the map remained the same.
The point is, the present map didn’t occur overnight. There was no radical re-drawing that led to the present map. Sure, certain candidates had strengths and weaknesses that other candidates did not, but the basic dynamic was the same. I do not, therefore, anticipate that there will be any radical redrawing of the map in 2008. Obama may run better in the Upper Plains states (though I don’t think he will) and the Carolinas (much more plausible), and McCain may perform relatively better in the Mountain West.
But Obama’s Up 14 In Pennsylvania!!!
But what of the changing polls? After all, polls show Obama close in NC and ND and leading big in NM!
First, one should be skeptical of polls from the summertime. People are still sorting out their preferences. Allow me to give you a long list of some polls that came out in the summer of 2004:
July 18 ASU (Kerry 42, Bush 41)
Aug 1 Market Solutions (Bush 48, Kerry 45)
June 30 Rasmussen (Kerry 46 Bush 45)
July 12 SUSA (Bush 49 Kerry 45)
July 31 Ras (Bush 46, Kerry 47)
Aug 22 SUSA (Bush 48, Kerry 47)
Oct 20 Op. Rsrch (Bush 48, Kerry 48)
August 18 SUSA (Kerry 49, Bush 46)
Polls never had Bush with more than a 1-point lead until late September, see also the October 6 Gallup poll showing a tie.
June 23 ARG (Kerry 47, Bush 46)
June 27 Quinnipiac (Kerry 46, Bush 44)
June 30 Rasmussen (Kerry 48, Bush 43)
July 12 SUSA (Kerry 47, Bush 44)
July 15 ARG (Kerry 49, Bush 45)
July 19 Research 2K (Kerry 49, Bush 44)
Aug 5 ARG (Kerry 50, Bush 43)
Aug 10 Q (Kerry 49, Bush 42)
July 26 SUSA (Bush 52, Kerry 42)
June 30 Ras (Bush 45, Kerry 46)
July 31 Ras (Bush 44, Kerry 48)
Aug 24 SUSA (Bush 44, Kerry 49)
Sept 22 SUSA (Bush 47, Kerry 46)
July 1 SUSA (Kerry 51, Bush 41)
July 8 ARG (Kerry 50, Bush 43)
July 31 Ras (Kerry 50, Bush 44)
Aug 4 SUSA (Kerry 52, Bush 41)
And let’s not forget some of the grandest outlying polls, coming weeks before the election:
Oct 18 Ward (Bush 43.4, Kerry 42.6)
Oct 20 SMS (Bush 46, Kerry 45)
Sept 19 SUSA (Bush 48, Kerry 48)
Oct 25 Q (Bush 46, Kerry 46)
Oct 26 Strategic Vision (Bush 44, Kerry 44)
Anyway, I think my point is made, but this leaves out August polling showing Kerry up 14 in PA, Bush only up 7 in SC, up two in TN (and up six in September), and Kerry up only 10 in VT. Plus polling from September showing NY within 5, IL within 10, and WV a two-point matchup.
Most polling last cycle also showed VA to be within 5 or so, and NC within 4-6 until late September.
The bottom line is that, as late as early September, Kerry was leading in states worth a total of 310 electoral votes, and was within three points in states totaling another 60. This includes states like AR, MO, WV, TN, and FL, that didn’t end up being particularly close. We know how that worked out. Polling ebbs and flows. But it has little to do this early out with how the map eventually looks.
So we come now to the latest polling, which Barone argues shows new targets for McCain and Obama. Mostly for Obama. This survey of the polling data for the past two months shows some changes. But I don’t believe that it demonstrates a fundamentally different map.
Instead, what we’re seeing is an America that is much more sour on the Republican party than it was four years ago. There has been an absolute shift. Whether this is a long-term shift is a debate for next time.
But there is no denying that, in the short term, there has been a shift. But there has not been a relative shift. The states line up pretty much as they did in 2004, with some exceptions.
In 2004, on my blog myelectionanalysis.com, I employed a method of weighting state polls by age and historical accuracy. It is similar to the method employed at Poblano’s site. I’ve resurrected it. By weighting the national polls, as of Tuesday, Obama was leading McCain by about five points. By weighting the state polls, then in turn weighting them by population, you get an aggregate result of Obama up on McCain by about . . . five points.
And we get some astonishing results, which Barone picks up on. The Upper Plains states are close. Obama is up 22 in California, and 13 in MN. McCain is only up 4 in NC.
First, remember the 2004 results of summer polling above.
But what does the underlying map look like, vis-à-vis 2004? Has there really been a fundamental change?
We can do this by adding 7 points to McCain’s margins in every state. That shifts the national result back to R+2.5.
The results are seen in the Table below. The first column is the state, and the second column is the 2004 Bush margin. The third column is the current McCain margin. After the break, we see McCain’s margin if the result is shifted 7 points in his favor. That allows us to really compare the results from 2004 to see if there is a more fundamental shift in the states. The final column then subtracts the weighted McCain results from Bush’s results.
The results are pretty astounding. I’ve shaded states that move Republican by more than five points from 2004 in red, those that move Democratic by more than five points in blue. NOTE: I don’t know why this table didn’t work out so great here. But if you highlight the blue cells with your cursor, you can see the underlying data.
|STATE||2004 Bush Margin||Current McCain Margin||Normalized to McCain +2.5||Change From 2004|
A majority of the states haven’t moved more than five points since 2004. Let me type that again so it sinks in: Relatively speaking, a majority of the states haven’t moved more than five points since 2004.
Looking at states where polls have moved more than ten points, we get the following: Toward Obama: AK, HI, ID, IN, MT, NE, ND, SC, SD, UT, WY Toward McCain: LA, MA
HI we can explain as Obama’s home state. LA has had an unfortunate internal re-arranging in the intervening years. MA is harder to explain, except that it was a heavy Perot state, and McCain may appeal to Perot voters uniquely, and, well, Deval Patrick, who is kind of an Obama prototype, isn’t faring too well up there. I suspect the changes in the Carolinas and IN are for real. Those states may well have moved to the left, especially IN. Or they may just be heavily Republican states that aren’t comfortable with McCain, but that will likely break for him in the end. The remaining states are all Mountain West states and Northern plains states. They’re small, heavily Republican states, almost all of which McCain is still winning comfortably. These were also not heavily contested primary states, so people there might still not know much about Obama. They are also very conservative states, so it is possible that there is some suspicion of McCain. In the end, I suspect undecideds won’t break heavily in Obama’s favor in these (still) heavily Republican states.
So what does the PVI map look like today, compared to 2004?
I’ll let you decide how similar they are. To me, 2008 is looking like an exaggeration of 2004, which was an exaggeration of 1996. There’s no evidence of a single group pulling away from the Republican coalition, except possibly in IN. McCain’s problem is that the GOP is down nationally, not that there is any particular group splitting off of the GOP coalition. The fundamental red/blue map is intact (and in many ways, better for the GOP than 2004).