Where Should Barack Obama Be Campaigning?

The story--whether it is true or not I do not know--is that in 1988 Michael Dukakis said that he wanted to go campaign in Western Massachusetts in September, at the Massachusetts State Fair in Springfield. A nd his campaign manager Susan Estrich--anxious to tell the candidate what he wantd to hear and not what he needed to hear--said that that was a wonderful idea.

It wasn't. There are state fairs everywhere. There are additional votes to be gained by visiting places personally. And some of those votes count more than others.

Which votes count the most? It's not the votes in states that the polls say are evenly divided that count the most. It's the votes in those states that will be evenly divided if the election turns out to be close that count the most. John McCain spent n weekend campaigning in Arizona. But if Arizona is close--if John McCain actually needs to spend time in Arizona to shore up the votes there--then John McCain is probably already and also behind in Florida, Ohio, Nevada, Missouri, Arkansas, Colorado, Virginia, Tennessee, Louisiana, and other states with a total of 375 electoral votes. For John McCain to carry Arizona and thus collect 163 rather than 153 electoral votes is simply not very important.

Which states will be close if the election is close? Take the average difference in the two-party vote in the 2000 and 2004 elections. Correct for the fact that in this election our African-American citizens are likely, for the first time, to turn out in proportion to their numbers--and to turn out for Barack Obama--by adding 1/10 of the African-American share of the population to the average two-party difference. Normalize, so that all of these state-level numbers together add up to zero.

We then have three groups of states:

I. States Which Barack Obama Is Likely to Win by More than 4% in a Close Election:

From D.C. to Wisconsin, adding up to 248 electoral votes.

II. States Which Barack Obama Is Likely to Lose by More than 4% in a Close Election:

From Utah to Arizona, adding up to 172 electoral votes.

*III. The States in the Middle:

New Hampshire, Iowa, New Mexico, Florida, Ohio, Nevada, Missouri, Arkansas, Colorado, Virginia, and Tennessee--with 118 electoral votes.

Some of these states--Florida, for example--may be harder for Barack Obama to win than this very rough cut at the numbers would predict.

But if I were running the Obama campaign, I would not bother to send the candidate to California or Washington or Pennsylvania or Oregon--if those states are close enough for personal campaigning to make the a difference, the election is already lost. And I would also not bother to send the candidate to Arizona, West Virginia, Georgia, Mississippi, Indiana, Texas, or Kansas--if those states are close enough for personal campaigning to matter, the election is already won.

Comments

[Joe Sommer]

This is true for a close race. But if the race isn't close, Obama will want to win the Senate. That means he has to campaign constantly in places like MS and TX. There are other advantages to campaigning hard in red states. But I do agree--there is no point in wasting time in blue states, unless they are like ME or MN, with a Senate candidate who needs help.

[David]

Do personal appearances have a big impact on fundraising? That's one reason I could imagine, though I don't know if it applies.

If Obama can campaign in red states hard enough to force McCain to spend money there, it could be worth it even if the state is lost. Who knows, maybe Obama can also tap a few donors there?

[ogmb]

The "reduce decimals" button is unknown in Evans Hall?

[alkali]

This oversimplifies. Surely the Dakotas, the Carolinas, Kansas and Alaska are worth some attention, and perhaps others, and perhaps Texas is worth a test run just by virtue of its size. On ther othe hand, it is probable that at least Arkansas will be a difficult get for the Obama campaign. But the principle is sound, and it's surely what the Obama campaign is doing, even if they are not using this precise methodology.

[Bruce Wilder]

There is a subtle, strategic argument about the interaction of message and where you force your opponent to campaign.

There's a pushme-pullme aspect to assembling the support of a political coalition. If you concentrate too much on a few borderline States, which are middle-of-the-national-ideological-road, you can make it too easy for your opponent to mobilize his base, while not alienating the vital independent voters.

There's a good argument for Obama campaigning in Georgia and a few other States, where the Republican Party is repulsively conservative, and McCain, cultivating his mavericky moderateness, is not all that popular. McCain can not take Georgia for granted, because a demoralized Republican Party might not show up in sufficient numbers. But, if McCain goes to Georgia and says what he has to say, to solidify that base, his words will tend to repulse the critical voters he needs to win Ohio.

Republican operatives are saying, let McCain be McCain. Obama's people want to find ways to force McCain to be Bush. Making your Republican opponent campaign hard in a State with a very conservative Republican Party is a good way to do it.

[a student of economics]

This is a good start, but a significantly more sophisticated analysis can be found at fivethirtyeight.com which, among other things, considers 10,000 simulation runs to assess possible outcomes.

For instance:
1) recent polling, more detailed demographics and other factors should be used to update the set of states in the middle;
2) certain states are more likely to be pivotal in an election due to their size -- they may be hard to win, but if you do win (or lose) them, they are very likely to change the election;
3) the cost of investment, hence return on investment, varies across states.

Finally, it seems likely that there are interaction effects -- strategies that win one state likely affect probabilities in states that are geographically, demographically or otherwise similar.

[Brian]

Doesn't this imply that that the numbers of voters will essentially remained unchanged, in either proportion or total numbers, from 2004 to 2008? If we are registering more voters than the Republicans and can turn them out, why wouldn't this affect the equation?

Doesn't this also imply that there aren't any circumstances in individual states or regions that can affect the totals of one or two states in a group but not all? For instance, Indiana may be more competitive than Oklahoma or Utah because it's a Midwestern state that, for whatever reason, shows signs of being more competitive for Obama simply because of its proximity to Illinois?

Finally, doesn't this also argue against the idea of laying the groundwork for future races? Sure, Indiana or Georgia, or perhaps Texas, may be unlikely to flip this time. But if we can register a few hundred thousand voters in these states, we probably stand a better chance of making it a closer next time.

If the answers to the first two questions is something along the lines of "You don't understand statistics," I wouldn't be shocked. Still, I'd like how I am wrong, if I am indeed incorrect, explained.

[PSP]

I have no idea if the story is true, but I think it is factually wrong:

1) The Big-E is the New England Fair, not just the Mass State Fair, and
2) I doubt it was an irrational move on Dukakis' part. He was a Mass pol who could count voters, having been kicked out of office once already. I suspect he was afraid of losing Massachusetts too.

The Springfield Irish Machine had no use for Dukakis, and controlled the city back then (I left Mass in 2001, and they were still pretty much running things, but federal indictments were looming). A short search reveals "A larger group of protesters was on hand to greet Mr. Bush when he made an unscheduled trip to Springfield, Mass., to receive the endorsement of the International Brotherhood of Police Officers Local 364 and the Springfield Police Department Supervisors Association." fn. Sounds like the machine was coming out in full force to defeat him, and he needed to fight back.

The rest of Western Mass traditionally went republican anyway. cf. "The Last Hurrah". Even today, the hill town where I grew up votes 80-90% Republican.

Losing your home state is embarrassing for a presidential candidate.

fn. http://query.nytimes.com/gst/fullpage.html?res=940DE5D91730F932A35753C1A96E948260&sec=&spon=&pagewanted=1

[Zeno]

There is also the factor of national media attention. For example, Alaska is probably out of reach for Obama, but it's not impossible. Presidential campaigns usually ignore it. (Nixon wasted some time there in 1960.) However, the Democrats are running a strong challenge to Sen. Stevens, with Nick Begich (mayor of Anchorage) having a good chance of ousting him. An Obama appearance would boost the Democratic nominee, get national coverage, and thereby potentially affect the mindset of the nation's voters in a positive direction. ("Wow. Obama has a chance at Alaska? He must be doing really well!")

An eye on national coverage and opportunities to poach on traditionally enemy turf will take the candidates to venues other than just the teeter-totter states, some of which might topple into his column with the impact of let's-go-with-the-winner.

[save_the_rustbelt]

Obama has put Michigan in jeopardy, and Ohio should have been a sure thing but is not (remember the Kerry campaign?).

Why can't Obama get any traction on the economy? Good God.

Could he be a primary wonder and a general election flameout?

[save_the_rustbelt]

Obama has put Michigan in jeopardy, and Ohio should have been a sure thing but is not (remember the Kerry campaign?).

Why can't Obama get any traction on the economy? Good God.

Could he be a primary wonder and a general election flameout?

[save_the_rustbelt]

Obama has put Michigan in jeopardy, and Ohio should have been a sure thing but is not (remember the Kerry campaign?).

Why can't Obama get any traction on the economy? Good God.

Could he be a primary wonder and a general election flameout?

[Doctor Jay]

I doubt that number of votes moved by a personal appearance is very large. I think the votes moved by ads bought with money raised by personal appearances, and by media coverage of a personal appearance is larger.

Then there's the coattails and strategic reasons that others mention.

[David J. Balan]

David Stromberg has a paper on a closely related topic in the current issue of the American Economic Review.

[jerry]

"Where Should Barack Obama Be Campaigning?"
and
"Why can't Obama get any traction on the economy? Good God.
Could he be a primary wonder and a general election flameout?"

My suggestion has been ignored. Obama should be leading his campaign from the US Capitol, where he would be leading Congress in protecting consumers from credit companies, from the awful bankruptcy bill, from warrantless wiretapping, from an incompetent foreign policy debacle, from a private health insurance industry that gives us terrible health care, ....

If Obama had led, or would lead in Congress, it really wouldn't matter which state fairs he attended.

[bigTom]

I like your logic. My main quibbles are that if it looks like it won't be a close election, then the coattails effect for Congress and the Senate should become the priority. Secondly, I suspect there are diminishing returns, from multiple or extended campaigning in any particular state, i.e. if you've vistied Florida say 6times already, and have one more stop in your (time?) budget, it might pay to spend it on a state with a 5% expected difference, that you haven't visited at all. I.E. the marginal effect of visiting a given state likely diminishes with the number of visits. I expect that the act of never visiting a state (like those people don't count at all), could be pretty damaging. So maybe he should visit every state at least once, but make several visits to the ones identified by your method. Astudentofecon, might well be able to help identify an optimal strategy, assuming you can reasonably quantify things like the expected change (delta) in votes by another visit to a state, i.e. that you can reasonably calculate your odds of winning, and maximize it.

[bakho]

Part of my issue with your analysis is the effect of past campaining. If one candidate totally ignores a state, (Indiana in 2004) then the numbers might be much lower than otherwise. Bush beats Kerry about 60-40 in Indiana at the same time that Evan Bayh is trouncing his opponent 65-35. In some states a sizable portion of the electorate is voting candidates and not party. Usually, states are not in play because the party organization is too weak or non-existence. Because Indiana was a contested primary, Obama was forced to spend a lot of money and time to put together an Indiana organization. The hard part of building the organization is done. A lot of the volunteers are local. Abandon the effort? Or put a minimal amount of money in to keep it going? Personally, I think the money spent on Party organization and GOTV is better spent than negative TV ads.

[BillCross]

Isn't this the kerry strategy from 2004? It seems somewhat naive in that while the presidency is very important, but congress people are not exactly chopped liver in importance. For instance, the Alaska example above, Obama campaigning there will very likely have some effect down ticket and make electing the Democrat more likely.

he should campaign more in the close states, but ignoring the other states is a somewhat self-limiting strategy too.

[John]

This looks like a "median state theorem" stack up the electoral votes and if you win the middle one, the election is yours. I think this has Bob Shrum written all over it, unfortunately.

Campaigning is both offense and defense. I think the last congressional elections showed the value of putting on a full court press. Make McCain think Utah is in play...leak positive polls in MT, NV, SC, AK...This isn't a campaign for fine tactics...superior fundraising and a weak GOP candidate call more for the tactics of the Army of the Potomac smashing into the Army of Northern Virginia.

[Gene O'Grady]

I think some of this analysis is out of date -- Obama has a better chance in the Dakotas and Montana than in Arkansas and Tennessee. Further, the kind of publicity/coverage he can get out of just showing up and looking natural in those states carries over to the rest of the West and the Upper Midwest.

Plus states like Mississippi and Alaska where there are possibly winnable senate races and it may be a positive to have him show up and look presentable rather than remain the absent bogeyman. And eliminating the ability of the R's to campaign against the unspeakable absent can only have good long term effects.

[Bernard Yomtov]

Bruce Wilder says,

"Obama's people want to find ways to force McCain to be Bush. Making your Republican opponent campaign hard in a State with a very conservative Republican Party is a good way to do it."

I think that's a good point.

[Movie Guy]

Where Senator Obama should be in the forthcoming months will, in part, depend on his strategy. And that strategy needs to be updated...right about now.

.

[Jacob]

I really do think that Indiana is a swing state this time, which is a big surprise considering its solid record as a red state for the last, what, 6 elections or so? It's actually pretty amazing that VA, NC, and Indiana are being considered as deciding states this time around. This election may break the previous patterns.

[James Wimberley]

Fundraising? On the traditional fat cat model, you only need to pass by a few pachykattic watering-holes like Hollywood and New York as well as the median states. But Obama's fundraising depends on mobilising a wide base, and his personal charisma is important to this. So he has to campaign more widely than votecounting strategery suggests.
BTW, this sort of calculation is one reason for abolishing the Electoral College (OK, making it a museum piece with proportional rules.) The proportional rules of the Democratic primary enfranchised voters - like black Democrats in the Derp South - who have no weight in the general.

[Darren]

James... what does "pachykattic" mean? Google returned a big "huh?"

[Barkley Rosser]

This is an analysis made on the basis of past elections with an arbitrary adjustment for African-American population. That means that it is indeed off on some of these. Tennessee has been mentioned, where Obama's chances look a whole lot worse based on current polls than a whole bunch of states that this argument says he should avoid. Thus, I just saw a recent poll showing him slightly ahead in NC, and that is a big state. He is not far behind in MS as in some others like MT. And, sure, the presence of competitive Senate races where Obama is in the lead should be a factor as well.

Yet another are details of geography and local media markets. Thus, WI is put as not needing to visit, albeit at the bottom of that list. But, good old LaCrosse in the southwestern part of the state on the Mississippi River across from Iowa and not far from MM, has a three state media market. In 2000 and 2004 the locals and their police got sick of the candidates constantly tying up everything in their lovely, small city. But Obama would probably be advised to make a couple of visits there anyway, and I am sure McCain will be making their taxpayers pay their police overtime for the same reason (and he really needs WI and MN, as well as the obviously very competitive Iowa, along with that competitive Senate race in MN).

[Jim]

And speaking of the Bankruptcy Bill the Senate passed under the leadership of Joe Biden, Senate-MBNA/BA and the support of HRC,Senate-Citi/Chase, I would hope that JB is not the VP nominee. Nor Evan Bayh, for goodness sake. (Appropos of nothing, upon graduating from college in June 1968, I made the case to his dad in his senate office that Vietnam was a civil war, but he still maintained at that time that it was NVN aggression)Gov. of Ariz, Ks, Pa, OK. Senator Brown, nice.

[Robert Waldmann]

You assume, roughly, that Whites aren't affected by Obama's race. It is sadly a fact that Whites in the states with high Black populations are more racist than Whites in other states (except for my home state of Maryland where many of us are Black and all of us are wonderful). Basically the 10% of Black population could be fiddled with.

One thing you could do is take current polls (pollster or fivethirtyeight.com) and add a swing to McCain.

Or how about age or ... well it would take a lot of time. I should do it and I won't.