What If the Obama Campaign Didn’t Win Him the Election?

by John Sides on January 7, 2013 · 27 comments

in Campaigns and elections

In the wake of the 2012 election, it’s become commonplace to credit Obama’s “formidable” campaign for his victory.  I’ve already begun to challenge that narrative a bit.  This post will continue to do so.

As I understand the “formidable campaign” narrative, it’s that Obama campaign simply did a lot of things much better than the Romney campaign.  If so, then one possible implication is this:

Obama should have done better where the two campaigns went head-to-head, relative to places where neither side was campaigning.  That is, even though Obama was expected to lose votes in most states relative to a more favorable year like 2008, he should have done better in the battleground states, relative to non-battleground states, because the battleground states were where his campaign’s hypothesized prowess—in fundraising, messaging, GOTV, etc.—was manifest.  So did that happen?

No.

At a conference several weeks ago, I presented the graph below comparing Obama’s vote margin in 2008 and 2012.  The data are courtesy of the indefatigable David Wasserman.  The battleground states are the black dots (CO, FL, IA, MI, NC, NH, NV, OH, PA, VA, and WI).

If the Obama campaign really beat Romney that badly, you’d expect the battleground states to be “higher” on the vertical axis than the other states.  That is, you’d expect them to stand out as states where Obama did better relative to 2008.  But that’s not really true.  He lost 2.05 points in the battleground states relative to 2008 and 2.24 points in the other states—a difference of less than two-tenths of one percentage point (0.19).  A simple regression model confirms that, once you’ve taken Obama’s 2008 margin into account, his 2012 margin was no better or worse in the battleground states compared to the other states.

I was not the only one who made this graph.  See Gary King’s tweet.  See also this paper by Alan Abramowitz.  Abramowitz goes even further, showing that Obama’s advantage in terms of field offices was not correlated with outcomes in the states.  Abramowitz has also made a similar graph, comparing Obama’s 2008 and 2012 vote among different groups, using the exit polls:


Obama did a bit better among Latinos and a bit worse among Jews, but otherwise his coalition was nearly identical.  (Hence my skepticism about some of the early “Obama is in trouble with (group)” stories).

In sum, I haven’t yet seen good evidence that, where the two campaigns went head-to-head, Obama clearly got the better of Romney.

Some caveats.  First, to say that Obama’s campaign might not have won him the election doesn’t mean that his campaign wasn’t innovative.  See this great slidedeck from engagedc. Put differently, Obama’s campaign may have done many things better than previous campaigns—in terms of data analytics, digital, GOTV, etc.—even if the net effect of those things on Election Day wasn’t large.

Second, evaluating campaign effects depends on the underlying counterfactual. I want to stress that these graphs don’t mean that the campaign was irrelevant.  If Obama had never held a rally or spent a dime on ads or GOTV, there’s a very good chance he would have lost.  But of course, this counterfactual is also absurd.  And it’s not what the formidable campaign narrative has in mind, as far as I can tell.

A skeptic of this post might ask: what if the Obama campaign hadn’t been so innovative and formidable?  Wouldn’t he have done worse in the battleground states?  In other words, didn’t the Obama campaign need to be that much better than Romney’s in order for Obama to win?  I tend to doubt that, in part because Obama was favored by the underlying fundamentals.  He wasn’t actually “supposed” to lose, and so I’m not sure he needed to be a better campaigner to win.

A skeptic might also rejoin: you’re looking at votes, but many of Obama’s innovations were about fundraising, and without them, he wouldn’t have raised as much money and that would have put him at a disadvantage.  My response: it’s pretty clear that the Obama campaign learned how to better raise money by experimenting with emails, developing better fundraising apps, and the like.  What we don’t know—or maybe I should say, what I don’t know—is how much additional money they raised because of this learning.  We should also keep in mind that these likely a diminishing marginal return to campaign spending.  When you’re running a $1 billion campaign, a few million dollars here or there, even a few tens of millions, is not going to be crucial.

Finally, I want to emphasize—in light of my post on the early ads and Walter Shapiro’s response—that The Gamble and the other political science work on the 2012 election will include much more precise estimates of the effects of ads and other things.  What I’ve presented at 538 and in this post are very simple graphs that are designed to stoke interest and reaction.  They are iterations of research that will be much more involved (not that this makes our book or any other study beyond reproach, of course).

Stay tuned.

{ 27 comments }

Andrew Gelman January 7, 2013 at 9:27 pm

In the immortal words of Kevin Spacey, the greatest trick the Devil ever pulled was convincing the world he didn’t exist.

DavidT January 7, 2013 at 10:53 pm

I think Baudelaire said it first.

Andrew Gelman January 7, 2013 at 11:04 pm

Sorry, but as an American politics specialist, I speak only English.

Joseph Cera January 8, 2013 at 12:42 am

If conventions count as part of “campaigns” then I’d say Obama’s campaign had something to do with the 15-point spike in Gallup’s economic outlook index that occurred between September 2 and September 9.

Nadia Hassan January 8, 2013 at 12:43 am

Obama’s approval ratings also rose slightly after the conventions

John Sides January 8, 2013 at 1:01 am

Joseph: The spike in Gallup’s index was driven by Democrats and independents:

http://www.gallup.com/poll/158429/economic-confidence-best-2008.aspx

In the YouGov data, I unpacked independents to compare pure independents and partisan leaners. It confirmed a spike among Democratic identifiers and leaners but not among pure independents.

So I think this episode nicely illustrates how campaigns can rally partisans — a finding that dates all the way back to the original Berelson and Lazarsfeld studies. But it’s an open question whether this spike really netted Obama additional votes — i.e., what would be manifest in the figures above — or merely made people who were already going to vote for him conform related perceptions (like of the economy) to their vote intention. This is the rationalization process that Lenz and others have identified.

Erik M. January 8, 2013 at 12:51 am

Would it make a difference to distinguish between states that were battlegrounds in both 2008 and 2012, versus states that were heavily contested in one year but not the other? Even if Obama’s 2012 campaign did some things differently than his 2008 campaign, it could still be true that, say, outcampaigning McCain in OH in ’08 and outcampaigning Romney in OH in ’12 could add up to no net gain.

Jim P. January 8, 2013 at 2:01 am

Yes, exactly — using the 2008 election results as control data here introduces a potentially major bias, since it’s pretty widely accepted that the 2008 Obama campaign outperformed the McCain campaign as well. So not seeing a relative improvement in battleground states in 2012 only means that the edge the 2012 Obama campaign had over Romney wasn’t any larger than the edge the 2008 Obama campaign had over McCain.

Mark January 8, 2013 at 11:22 am

I don’t reach the same conclusion at all. Looking at Wasserman’s graph, it’s clear that Obama under-performed in 2012 relative to 2008 in all states, which the author acknowledges. However, in non-competitive states, the dots are widely dispersed and well below the 08-12 parity line. The battleground states are more tightly clustered toward the parity line indicating that the under-performance was less dramatic than in non-competitive states.

Wasserman’s data said the overall aggregate difference in votes from ’08 was -1.71 %. In battleground states Obama over-performed ’08 by 0.28% in aggregate. In non-swing states, Obama under-performed ’08 by -2..74%. This seems to indicate that Obama’s targeting of the swing states was successful since it reversed the overall trend of under-performance relative to ’08 that was evident in states where the campaign had no presence.

John Sides January 8, 2013 at 11:36 am

Mark: A few things here. First, it’s my graph, not Wasserman’s. He bears no responsibility for anything I argue here.

Second, in Wasserman’s data, you need to look at the second worksheet, where he tabulated the vote margins in percentage points. I think you are looking at whether the total number of votes increased in the swing vs. non-swing states. Whether the total number of votes increased doesn’t tell us anything about whether those votes went for Obama or Romney.

Matt Jarvis January 8, 2013 at 1:11 pm

John,
Clearly, the stories that said Obama was in trouble with Republicans, conservatives, and evangelicals had it right!

Walt French January 8, 2013 at 2:14 pm

Interesting analysis. But in ’08 Obama was playing offense in the battleground states — offering economic hope to hollowed-out economies — versus his having to play defense there in ’12. These states were battlegrounds in 2012 because without any particular shaping of themes, etc., both parties naturally expected them to be R.

The results could indeed be perfectly consistent with the notion that the re-election effort pulled out a victory.

Walt French January 8, 2013 at 2:28 pm

I think the 2nd chart is the more interesting one. Most dots are so close to the line as to be unremarkable, but we can make easy interpretations for the 3 standouts.

I interpret the big bump for “Democrats” as meaning that the “get out the vote” campaign offset both strenuous Republican voter-suppression efforts and Rs’ perception that D voters would be fatigued. Major issue; obvious success for Obama.

Conversely, the Rs seem to have found a successful theme in going after Obama as weak on Israel. Versus the Happy Warrior McCain, that should rank as a coup.

Finally, Romney lost in the Hispanic demo. Maybe this is that Romney didn’t carry the Bush halo. But if you lump in Asians, given that they, too, reportedly shifted even more sharply from party balance to Obama, the obvious interpretation is that voting citizens (ahem!) with obvious ties to immigrants found the Republican anti-immigrant and anti-meltingpot themes (“just borrow college money from your parents”) worthy of special approbation.

Khadijah January 9, 2013 at 10:23 am

So, in other words (and since Voter ID laws are *not* voter suppression by any reasonable standard — even the international observers commented on how much “trust” there is in US elections, which is a red flag to everyone other than a rabidly partisan Democrat — and since insisting that adherence to existing immigration laws is *not* anti-immigrant, again by any reasonable standard), the media won the election for Obama.

I’ll agree with that. :-)

Jfed_STL January 8, 2013 at 4:37 pm

Thank you for this analysis… it has defiantly stoked my interest!

But their is a major flaw in your metric.

50% +1 vote in a political district is a win… running up the margin does not matter.

Increasing voter turn out within sub-districts and demographic groups is important

Thank you for your work on this analysis, but let me nudge you in a different topic.

“TV Political Ad Revenue Up 68 Percent Over 2008″
http://www.responsemagazine.com/direct-response-marketing/news/tv-political-ad-revenue-68-percent-over-2008-4849

That’s right 68% more was spent on TV… and this did not increase voter turn out by 68%

Is this an example of diminishing returns on a dollar… or the death of TV adds?

tdiinva January 9, 2013 at 10:14 am

My two cents:

Close elections are won at the margin. The Obama campaign was much more successful in identifying, convincing and motivating the 2012 marginal voter. The low information voter was this electon’s marginal voter. The Obama campaign focused on unconventional informations sources to reach these voters and whereever possible get them to the polls before election day to lock in their votes. The Romney campagin completely ignored these voters because traditionally they don’t make it to the polls.

Khadijah January 9, 2013 at 10:17 am

This is excellent and appreciated. From my specific vantage point (and speaking anecdotally) I couldn’t grasp the “machine” narrative in the least. All subgroups I read and interacted with prior to the election were less enthusiastic about Obama than in 2008, clearly; yet, we seemed to be being to death with the narrative that Obama’s voter enthusiasm carried him through.

From the above it seems pretty clear. In 2008, he won with 365 electoral votes and 52.9% of the electorate to McCain’s 45.7%. In 2012, he won with 332 electoral votes and 51% of the vote to Romney’s 47.2%.

Outside of spinmeister world, that’s called an election where the incumbent suffered an erosion of support against an opponent who was more succcessful than his previous opponent.

IOW, he hung on to win despite the bad economy.

Steve January 9, 2013 at 10:18 am

We had an interesting post here where in raw numbers and percentages the country is trending Republican (tho not fast enough)…

http://commoncts.blogspot.com/2012/11/hey-state-controlled-media-country-is.html

Paradigm Lost January 9, 2013 at 10:34 am

Given the state of the economy, the persistent UN-popularity of his “signature” achievement, ObamaCare, and the continuing (and worsening) mess in foreign affairs, Obama rightfully should have LOST by at least 5 pts. The fact that he somehow managed to remain in office in spite of all that is a product of his campaign, nothing else.

scott ballard January 9, 2013 at 10:34 am

The deciding factor is who didn’t vote for Romney. Coutler correctly pointed out that Romney blew past McCain and Reagan in most demographics- he beat Cruz in Texas in the general population and Hispanics! So who didn’t vote for him? Evangelicals that showed up went R. But how many more stayed home? If you look at the numbers, I believe you’ll see why Romney lost.

vnohara January 9, 2013 at 11:28 am

I was hoping this article would identify the real masters who put up the puppet in the WH.

I guess it’s hard for Americans to accept that all we now have are puppets in WH.

The fact that Obama, the narcissist, had written a concession speech the day prior to Nov 6 is a big clue. The 11th hour negotiation went south for Romney. Perhaps there was one thing the master / firm wanted him to continue doing (which likely would not have helped the economy — Romney’s aim and specialty) that Romney refused to agree.

Obama has been saying things all factions of US voting body wanted/want to hear but he has bee following old plan. Obamacare wasn’t his plan, was not new. So on and so on. His sheep like to claim this shows his principle, core conviction, blah blah blah.

Well, core conviction where fighting for the skinny cats to me is living among the skinny cat, or at the very least holding back extravagance in consideration of the struggles of skinny cats. All the skunk has been doing is partying with the 1%ers he rails against and living kingly. A better fraud I have never known!

I will not lose sight of the fact that both parties belong to one ruling establishment and share one goal that is maintaining ruling power and longevity. US has grown from the different child into conventional maturing adult of the old world. American sheep may occupy different fields in US but are still sheep nonetheless.

Wheeler's Cat January 9, 2013 at 11:37 am

The deciding factor is demographic evolution. There just aren’t enough white people left in the electorate to deliver a victory to an all white party. Percent whites in 2012 was 72%. That is elephant in the room. Because of asymmetrical enthusiasm the GOP already does a great job of turning out the base. The GOP needs to integrate its base. Because in 2016 there will be only 70% whites (nonhispanic Caucasians) in the electorate, and a third of those whites are organic liberals. So no amount of GOTV magic is going to turn out base voters that don’t exist.

Elihu E. Wygant January 9, 2013 at 11:37 am

This was no ‘election’; it was merely ushering in, for another four years, the empty-suited puppet that the one-world government elites have currently chosen. If Obama should begin to waver, or tries to go his own way, he will soon be ‘replaced’. Not difficult to see this current scenario if you are not one of the sheep.

Matt January 9, 2013 at 12:23 pm

The regression line does not look like best fit somehow, with the vast majority of data points below the line. Also I would hesitate to draw conclusions with such a small dataset. But just looking at the scatter it seems like the battlegrounds are grouped tighter to the line than the rest of the system. And there are SO many factors that can influence these results, like the economic conditions of specific states, closeness of the race affecting turnout, demographic shifts in certain states etc; it’s especially hard with such a small dataset. Interesting, though.

John Sides January 9, 2013 at 1:21 pm

Matt: It’s not a regression line. It’s a 45-degree line meant to indicate where the data points would be if the state outcomes were identical in 2008 and 2012. That most points are below the line shows you that 2012 was a less favorable year for Obama than 2008.

Jerry Reasoner January 9, 2013 at 11:35 pm

If you did plot a regression line, the biggest battlegrounds of Florida, Ohio, and Virginia would lie above it. By my figuring Obama did about a percent of margin better in those states than you would estimate if you modeled the 2012 margin as a linear function of 2008 margin. Still only Florida was close enough to say that Obama won it by outdoing the overall trend there.

David Kimball January 11, 2013 at 5:03 pm

The converse of the growing narrative is that Romney ran a lousy campaign. But consider that Romney garnered more votes than the GOP Senate candidate in 27 of 33 states with a contested Senate race (put differently, Obama ran behind the Democratic Senate candidate in most states). If the Romney campaign was so bad, then there apparently were a lot of GOP Senate campaigns that were worse.

Comments on this entry are closed.

Previous post:

Next post: