The Map Will Tighten Before It's Over

Up until the Republican convention, observers of this election had been told that this election is going to be different, that the winner is going to "redraw" the electoral map.  After all, between 2000 and 2004, only three states switched sides (New Hampshire, Iowa and New Mexico).  It really was time for a change, and we heard some variant of this idea from really smart guys, like Chuck Todd and Larry Sabato.(1);You know the refrain: close elections are the exception, not the rule; and rarely has the electoral map showed so much long-term equilibrium.  The Cleveland-Harrison map of 1884-1892 was remade by McKinley in 1896; Truman's upset did not resemble the Roosevelt victories of 1940-1944.  So, the "theory" goes, we can discard our preconceived notion of blue and red states, as there are several states that can potentially switch sides.  

    Then came the selection of Sarah Palin, and the race tightened--primarily due to an energized Republican base.  Just like that, we were back to the traditional swing states: Pennsylvania, Michigan, Ohio and Florida.  This CW lasted about two weeks, until the onset of the current financial meltdown; in its wake, we are told that Barack Obama is once again poised to redraw the map.  The idea seems plausible--but it is probably wrong.  This essay will suggest that, in addition to polling data or economic statistics, there are four sociological factors to consider when modeling voter behavior in so-called swing states (including, of course, the potential for a Bradley Effect).  

Full Disclosure: I am a Democrat, but I know the first rule of the social sciences is to recognize the difference between normative and positive--between how we would like the world to work, and how it really works.  Which is another way of saying: the idea that Obama ever had a serious shot in Georgia, or that he still has a real shot in Montana, is seriously misguided.  It is one thing for pundits to engage in speculation; all it does is to annoy the informed reader or viewer.  But this lack of tough-mindedness can have serious strategic implications for determining where a campaign allocates its resources.  

I am not here to argue that the red-blue divide is static, and will not change in our lifetimes.  But 2008 is still not the year.  Chuck Todd suggests that by October 15th--the date of the last presidential debate--we will pretty much know where the election stands.  Not quite--but by late October, only a handful of states will still be in play in the race for 270 electoral votes. These are: Virginia, Ohio, Colorado, Nevada, and possibly Florida, North Carolina and Missouri.  Seven "red" states, 100 electoral votes--that's it.  (Pennsylvania, New Hampshire and Wisconsin lean Obama; West Virginia, Indiana and Montana lean McCain.  These will not be tossup states.)  By contrast, in 2004 there were twelve battleground states; of the seven current tossups, only Ohio, Colorado, Nevada and Florida were battleground states in 2004.(2)

    How can I be so confident?  Because the problem with many pollsters and pundits is they look at a series of polls, or even a moving average, and then extrapolate from there.  That is how a 48-45 McCain/Palin lead in Indiana or North Carolina, or the same Obama/Biden lead in Minnesota or Pennsylvania, becomes a `toss up." This type of modeling tries to be quantitative rather than qualitative (as if quantitative equals "objective" and qualitative equals "subjective.")  As such, these models tend to underplay the demography and political culture of a specific state.  On Real Clear Politics, any state with an average margin of less than five points is listed as a toss up.  Now, RCP does offer a "No Toss States" Map, but it is based simply on extrapolating from the hard numbers, i.e., it assumes that undecideds will break the same way as the polls.  This ignores what Todd calls the "soft underbelly"-- people who may be unenthusiastic about talking of pollsters, but will still vote the way they always have once inside a voting booth (Nate Silver of calls this the "Shy Tory" Factor).

It is quite probable that Barack Obama, from a neighboring state, will perform better in Indiana than Al Gore or John Kerry, and that he will likely win his western neighbor, Iowa.  At the same time, a strong (> ten point) Democratic win in Pennsylvania is probably out of reach.  But let's not kid ourselves: John McCain will win Indiana and Barack Obama was never going to lose Pennsylvania.  Southern Indiana is too culturally conservative, and Democrats in Pennsylvania are too well-organized, to reasonably predict different outcomes.  To put it another way: If, by late October, Obama is still within striking distance in North Carolina, Indiana or Montana, he will not only win; it means Republicans will receive a beat down, up and down the ticket.  Conversely, if McCain does regain a foothold in states like Minnesota, Wisconsin or Pennsylvania, Obama is in big trouble.  (Not that McCain would have to win any one--but to force Obama to bleed resources in these states might be fatal.)

    So what factors should pollsters, pundits and strategists actually look at?

1. The Political Culture of a State.  Indiana provides us with a locus classicus. Ever since it went for its homeboy Wendell Willkie in 1940, the Hoosier state has voted for the Democrat exactly once--the LBJ landslide in 1964.  It may be part of the rust belt, but it is culturally rural and quite conservative.  Maybe Democrats could make inroads due to its large union presence.  But this year, the best they can do is to narrow the gap.  Yet what applies to Indiana also applies, to some extent, to Ohio.  Its rural areas are also quite conservative, and are traditionally Republican.  Long considered a swing state, since 1944 Ohio has shown a clear historical pattern.  Only southern Democrats actually win here: Truman, Johnson, Carter and Clinton twice.  To win Ohio, Obama will not only have to elevate Democratic turnout in cities like Cleveland and Akron, but avoid a wipeout in southern Ohio.  He can do both if he can connect his understanding of the current economic crisis to the Ohio voter (and maybe it is time to "let Biden be Biden").  But other formidable Democrats have tried and fallen short in Ohio, including Kennedy, Humphrey and Gore.  So it should surprise no one that Obama is performing only a bit better than did Kerry.  (Even Bill Clinton could not reach fifty percent of the popular vote in either of his victories here.)  

A more feasible goal for Democrats may be the three southwest targets:  Colorado, New Mexico and Nevada.  It seems that the Democratic Party has taken Thomas Schaller's advice, and is targeting states where people may be more receptive to their message.(3)  It seems that the Democratic Party has taken Thomas Schaller's advice, and is targeting states where people may be more receptive to their message. An extrapolation of Schaller's main argument may be useful here.   Schaller points out that, in many southern states, African-Americans have relatively high voter turnout rates.  However, in these same states, the larger the black population, the more white people vote Republican (as such, Schaller suggests that any attempt to boost African-American turnout in the South is probably doomed to electoral failure).  This type of ethnic bloc voting does not seem to occur in states like Colorado, New Mexico or Nevada, all of which have large and growing Latino populations.  However, their turnout rates pale historically compared to those of southern blacks.  (It is true that Latinos are not as overwhelmingly Democratic as blacks, but they still vote Democratic, and Obama is polling better than Kerry did in 2004.)  Democrats have been crowing about their registration and voter turnout efforts, and are making concerted efforts to register and turn out Latino (and Native American) voters.  At present (early October), Obama enjoys a real lead in New Mexico, a slight lead in Colorado and is virtually tied in Nevada.(4)  This one is simple: If the Democrats can succeed in boosting Latino turnout, they have a real shot at all three.  

2. Effect of Demographic Changes.  The effects of some demographic changes are intuitive.  Louisiana has probably moved from pink to red due to the shrinkage of its African-American population, whereas the explosive growth of northern Virginia will probably turn the state blue for the first time since 1964.  Other, long-term effects are less clear; these might include the growth of the Latino community in North Carolina, or LDS birthrates in Colorado and Nevada (it is Mormons, not white evangelicals, who are the most reliable Republican voting bloc).  But probably the most difficult states to model are states with high rates of immigration: Who is moving there?  A lack of attention to such demographic shifts can have serious consequences for a campaign.  In Florida in 2004, John Kerry received over 650,000 more votes than did Al Gore--and lost by 380,000.  Much of this margin of victory came from Republican voters who did not vote in Florida in 2000.(5)   I suggest that Florida and Nevada are, in this regard, the two most difficult states to model.

Yet possibly the most important demographic change will be the increased participation of young people (ages 18-29).   Not only have they shown greater enthusiasm in this election, but they maintain steady support for Obama.(6)  Of all pollsters, Ann Selzer employs models using the highest percentage of young voters, and not surprisingly, the healthiest leads for Obama.    If, in the end, this is a true realignment election--with Colorado, Florida, Indiana, Missouri, Nevada, North Carolina and Virginia all turning blue--this, rather than the current financial crisis, may prove to be the single biggest reason.  

3. Political Sociology 2008. Or, are they Thomas Frank Republicans, or Lincoln Chafee Republicans? We know Frank's thesis: Republicans get people in red states to vote their cultural biases over their economic interests.(7)   Yet the corollary may also be true; it may be part of the reason for recent blue success in the northeast, parts of Midwest and west coast.  There are lots of people in New Jersey, Michigan and California that vote blue, but they are not MoveOn liberals. Some may be receptive to school vouchers or partial privatization of social security.  What they are not receptive to is creationism, gay bashing, Global Warming denial or the repeal of Roe v. Wade.  There has been some good reportage about this phenomenon in Oakland County, Michigan.  These are not lunch-pail Reagan Democrats; rather, these are suburban yuppies concerned with property taxes and good schools.  They started voting Democratic in 1992.  (One Democratic strategist calls them Rockefeller Republicans, but they are actually Jerry Ford Republicans.)  On Friday October 3, the McCain campaign drastically reduced its presence in Michigan--it seems that the addition of Sarah Palin was not swaying the hockey moms in this economically troubled state.(8)    

Nate Silver suggests that Governor Palin's real appeal may be to a specific demographic: white women with kids at home. This may prove relevant in states like New Hampshire, Iowa or Wisconsin (Although, for all of Palin's bravura, her social attitudes probably take New Hampshire off any red map.)  It is important to recognize what the Palin selection really signals; an attempt to fire up the Republican base and engage another turnout war.  (It also suggests that Republicans believe that many of these new registered Democrats are not sold on Obama and are potential ticket-splitters.)

4. (Potential) Bradley Effect. First, let us dispose of what the Bradley effect is not; what transpired in the New Hampshire Democratic primary was not an example of it.  What seems to have happened is that pollsters under-sampled women voters (52 percent in the voter models, 58 percent of the actual turnout).  The effect seems to work like this: It is not people who state they will vote for an African-American who are lying; it is the undecideds that are reticent.  The poll numbers seem to predict the black candidate's actual percentage fairly well, as it did in Tom Bradley's second run, or Doug Wilder's win in Virginia.   In other words, if the black guy is ahead in the polls 47-41, he may lose 48-51.

The real concern is people (candid or not) who are reluctant to vote for Obama, but who would vote for a white Democrat.  Larry Sabato calls this "racial leakage," and suggests we have no way to precisely measure it.  It may turn out that Obama will not win a couple of northeastern states by the big margins of Gore and Kerry, because some white people may not vote for him.  But Obama will still win these states.  However, this may prove to be a tipping point in states like Missouri, North Carolina, Florida or even Ohio.(9)   Tip: As we get closer to the election, stop paying attention to the spread, and focus on Obama's percentage.  On October 25, it is better for Obama to be up 49-46 than 47-41 in any swing state.  

So where does the electoral map really stand?  In autumn of 2000, the late Tim Russert held up a famous sign with three words: "Florida Florida Florida." In 2004, he correctly predicted it would be decided by Ohio.  As of now, it seems likely that Obama will hold the 19 Kerry states and win back Iowa and New Mexico, for 264 electoral votes; McCain is leading in 24 states--including Indiana--for 174 electoral votes.(10)   Obama clearly has more room to maneuver--yet he still may not be able to close the deal in North Carolina, Florida or even Ohio.  Whatever the result, do not be surprised if Obama performs better in Colorado and Virginia than he does in Ohio and Florida.  But all he needs is one of these to put him over the top.

George Axiotakis (The Groundhog), was a history major and is a Democrat.


1. Cf. Abramowitz, Alan, Thomas E. Mann and Larry J. Sabato, "The Myth of a Toss-Up Election" (July 24, 2008) and Sabato Larry J., "Meet the New Map--Same as the Old Map (Almost)" (September 18, 2008):

  1. Source: Leip Atlas.
  2. Schaller, Thomas, 2006. Whistling Past Dixie: How Democrat Can Win Without the South.  New York: Simon & Schuster.  
  3. Source: Leip Atlas.
  4. My own experience taught me something about this lack of attention to empirical data.  In 2004, I volunteered with a well-known progressive group that usually allies with the Democratic .Party.  I tried to tell anyone who would listen that Florida was fool's gold for Democrats, and why.  Everyone I spoke with was confident that Florida was following the northeast in turning blue.  It seemed to me that this optimism was based solely on the 2000 result, and not any subsequent empirical research (including Jeb Bush's easy 2002 reelection bid). The final 2004 vote in Florida was Bush 52.1%, Kerry 47%; this was a larger margin than almost all polls, including the exit polls!    
  5. Selzer polls in Indiana, Iowa and Michigan.  Unlike other pollsters, she does not employ the 2004 result as a benchmark, where voters 18-29 accounted for 17 percent of the electorate.  Source: Nate Silver,
  6. Frank, Thomas. 2004. What's the Matter with Kansas?  How Conservatives Won the Heart of America. New York: Henry Holt & Co.
  7. By contrast, the suburbs of Minneapolis-St. Paul have grown redder in recent years.  This is probably the result of an aversion to taxes rather than a new cultural conservatism.   Some polls do show McCain to be competitive in Minnesota, but this is the only state where McCain actually outspends Obama (and St. Paul was the site of the GOP convention).  
  8. One region where this nonsense does not seem to obtain is the upper Midwest, in the "battleground" states around Illinois (except Missouri): Wisconsin, Minnesota and Iowa.  Obama maintains small but steady leads in all three, including Minnesota.  Both Gore and Kerry had to fight like hell here, and Iowa turned red in 2004. Also, whatever happens in Indiana, Obama is performing better than any Democrat in recent memory.  
  9. One month out, my hunch is to give Obama Virginia, Colorado and Nevada, for 291 electoral votes (and the election); Missouri and North Carolina to McCain, for 200 electoral votes.   As such, Obama will not need Ohio or Florida to win.  

Tags: battleground states, Bradley Effect, democratic strategy, electoral map, predictions (all tags)



Re: Different Yes; Realignment, No.

This is not a "normal" election by any stretch of the imagination.  The discontent with the GOP is widespread and deep.  The GOP brand at this point is badly damaged.  An unpopular president, the worst economic crisis in many decades blamed on the GOP, the issues people care about the most are primarily Democratic strong suits.  This is an election cycle tailored for a realignment. It would be surprising, in this political climate, if it weren't, frankly.

The rule of thumb seems quite simple to me (and is borne out by current polling data):  Look at a state's voter registration records.  In Florida Democrats have a massive voter registration advantage over Republicans by almost 2 Million. Similarly they have strong advantages in other states that are currently "red."  Not coincidentally, Obama leads in all of those states.  Given how badly damaged and wounded the GOP is at this point it appears that the voter registration advantages coupled with an increased enthusiasm amongst Democratic voters and a simultaneous lethargy amongst GOP voters should be enough to bring an electoral victory to the Democrats in those states.  Indiana and West Virginia are probably out of reach, as was pointed out, but look for the states with a healthy Democratic registered voter ID advantage, and you have a state that is likely to flip to the Democrats.

Florida, Ohio, Virginia, North Carolina, Colorado as well as other states fit that bill.  Expect almost all of them to flip to the Democrats this time around, which will in turn bring about the classification of a realignment election for this particular cycle.

by devilrays 2008-10-13 08:25PM | 0 recs
Disagree: Obama wins Indiana

The surprise of the night (unless you count him not quite winning NC) will be Obama squeaking a win in IN by.2%

by Davidsfr 2008-10-13 09:00PM | 0 recs
It's a toss-up right now

Might tighten; might widen. At this point I think it could legitimately go either way. It depends on whether McCain can salvage something resembling a functioning campaign. If he can do something to get some traction then yes, it'll tighten.

On the other hand, right now, with McCain flailing and getting no traction whatsoever? If he can't put together a plan that's workable by Wednesday's debate, I think it's pretty much over. I don't expect to see much map tightening begin after that point.

At this point, I'd consider the Kerry states plus Iowa and New Mexico to be largely settled ground. The battlegrounds are Ohio, Florida, Virginia, North Carolina, Indiana, Missouri, Nevada, and Colorado, and possibly West Virginia, maybe North Dakota. An Obama win outside those states is unlikely (but not impossible, if McCain tanks further).

Here's my thinking on your specific points:

1) I agree; this is a factor. However, it has its limits. There are a number of structural factors which bear on this (hugely unpopular Republican President, state of the economy, vastly increased Democratic registration, etc).

Taking your specific example -- right now across recent polls Obama is running +2.8 in Ohio. There are 8 current polls showing an Obama lead; two showing a McCain lead. The latest McCain lead in weeks is +2. By comparison, in 2004, the best stretch for Kerry was 6-4 in polls and with Bush never failing to have at least a +4 within any two week period. Obama is in fact performing considerably better in Ohio than did Kerry.

Certainly I agree about your Western-state analysis, to a point. I just wouldn't consider two of the three as just "feasible goals". McCain hasn't shown a lead of any side in New Mexico in a month, and only two leads, one tie, out of 16 polls over the last month in Colorado. That's not "feasible goals", that's pretty much decided. Nevada is more in play, though even there, the polling seems to have settled out very pro-Obama.

#2 I agree with more. I do think young voters (and new voters) are going to be a major factor this year. Part of this is that Obama is a far more inspiring candidate than Gore or Kerry (look at the enthusiasm gap if you don't believe me); part of it is that Obama's GOTV effort is likely to be ridiculously superior to Gore's or Kerry's.

#3... agree and disagree. Certainly political sociology is a factor.

On the other hand, I'd generalize your Michigan comment about Palin and simplify the whole point to "Sarah Palin's only purpose at this point is to consolidate the base." Polling suggests she has an important secondary purpose -- to drive off people who aren't in the base. Nate's analysis was good -- but it was done while she still had a positive net favorability. Now that that's not the case, and her rallies have become entirely red-meat vitriol, I don't think she plays to anyone outside the base.

#4 I strongly disagree. I think you're vastly overstating the impact of a possible Bradley Effect (and this has been studied in great detail; see Nate Silver's piece If The Bradley Effect Is Gone, What Happened To It?.

In the primaries we saw undecideds break for Clinton no worse than 2-1 in almost every state -- often closer than that. I'd be amazed if we saw any undecideds break more than 2-1 for McCain; frankly, I expect undecideds to break for Obama for a variety of reasons. But even if you go with worst-case, in your 47-41 example I would expect to see it wind up no worse than 51-49.

You're also missing a very real "Reverse Bradley Affect", which is that Obama has in general pushed AA turnout to well beyond historic levels. In states with sizable AA populations this may inflate his totals several points.

Overall, Obama reaches 300 EV using only states at which an average of recent polling puts him about 50% (thus taking undecideds entirely out of the picture). Barring a major McCain recovery, that's the floor. I'd consider the likely range to be 330-360.

One last thought: suppose Obama wins some tossup states and loses others. Does that mean the map tightened? I think it just means they were really tossups.

by Texas Gray Wolf 2008-10-13 09:30PM | 0 recs
Re: The Map Will Tighten Before It's Over
One thing that has gotten some attention, but not enough (in my view) is Obama's ground game.
I believe that the ground game has been extraordinarily underestimated, even by those who grant it's effectiveness.
The Republican right has been wringing every last vote out of the evangelical voters for a while now. It was an integral part of Rove's brand of politics. That effort was not up and running until after the Palin pick, and it won't be up to full speed before the election.
Obama's ground game, on the other hand, is a thing to behold. It is running everywhere, and is vastly superior to anything seen in national politics in this nation . . . ever. It's fluid, organized, goal-oriented, and effective. The guys over at 538 have seen it in action across the country, and they have been floored by it.
It is simply impossible for any pundit to quantify the effect that this effort could have on the election because it HASN'T BEEN SEEN BEFORE. We have no idea whether the polls are accurately reflecting the results of these efforts or whether they've fallen woefully short.
ANY prognostication that doesn't take the possible game-changing effects of this effort are missing a pretty big piece of the puzzle.
by EvilAsh 2008-10-13 09:52PM | 0 recs
a nice meaty post, Groundhog

States might have tendancies, but in many States, it is only a lean tendancy which can be overcome by GOTV, macro factors (like the economy), actual/perceived candidate performance in the last weeks and momentum.

You seem to suggest that a large percentage of undecideds and moveables (about 10% of the total vote) will go to McCain.  Maybe yes, but maybe no.

Take Florida, it's ripe to be had by Obama (and the campaign thinks so too).  Obama has superior resources in the State and older voters who have leaned against Obama, may in enough numbers, go for Obama because of the Palin pick and say, McCain botching a question on Social Security on Wednesday night (which is very possible).  We don't know.  

I hope you come back to either gloat or take your lumps.

by mboehm 2008-10-14 12:35AM | 0 recs


Advertise Blogads