Post from October, 2012

Is 2012 Election a 2004 Repeat?

Tuesday, 23. October 2012 22:43

There have been a number of analysts who have compared the 2012 and the 2004 presidential elections. In both cases, relatively weak incumbents with troubling economies faced milquetoast, older and rich white guys. Both campaigns featured strong, negative attacks on the challenger with a point of emphasis in turning a percieved strength (Kerry’s status as a military veteran | Romney’s substantial business experience) into an electoral weakness. Both elections were expected to be close. Both elections are expected to have hinged on the same swing states, and in particular: Ohio. But is this the same race? Let’s take a look at both races for the president, comparing the RCP averages for 2004 and 2012 in real time from August 17th to October 23:

RCP Average of Bush v. Kerry and Obama v. Romney

As we can see, there are some simillarities between the two contests. Both contests were very tight in August, only to see the incumbent open up a significantly large lead in September. This was followed by an October surge by the challenger to once again make the race close. However, there are also some significant differences between the two races. Over the full course of this cut of the data, Bush out performs Obama by 0.39 percentage points, while Romney outperforms Kerry by 1.24 percentage points. Considering how close the 2004 election ended up being (Kerry lost Ohio and the presidency by just over 2 percentage points), this is good news for Romney going into the November election. But beyond the overall difference, there is even better news in the comparison between the two elections…as Romney’s differential over Kerry increases from month to month while Obama has increasingly trailed the Bush totals as we move from August to October:

candidate differential

Note how in August the election was looking very good for Team Obama: they were outperforming Bush during the same time period by 1.22 percentage points and Romney was doing worse than Kerry by 0.59 percentage points. That all changed in the beginning of September. For the month, Obama underperformed Bush by 0.67 percentage points, a near two point swing in the wrong direction for Obama with respect to Bush. While Romney overperformed Kerry in the month of September by 1.44 points, a full two point swing in the right direction for Romney. So even in a month where Obama dominated Romney and the media narrative was that Obama had the election well in hand, if we take 2004 as the baseline, Romney was doing substantially better as a challenger while Obama was doing substantially worse as an incumbent. The month of October continues that trend, but it only becomes larger. In October, Obama is underperforming Bush now by a full percentage point, while Romney is besting Kerry by a full two percentage points. Note, if we were to apply that as a weight to the 2004 election results, then Kerry would have bested Bush in Ohio by a percentage point and gone on to win the presidency.

So while it is true that, in broad strokes, there are striking simillarities between the 2004 and 2012 presidential elections, the incumbent in 2012 is performing significantly below the standard set by the incumbent in 2004, while the challenger in 2012 is substantially besting the performance of the challenger in 2004.

Category:PoliSciPundit | Comments Off on Is 2012 Election a 2004 Repeat? | Author:

A Preference Cascade – Part II

Tuesday, 23. October 2012 16:08

As I mentioned in Part I, there is mounting evidence that the first presidential debate initiated (or at least correlated) with a preference cascade for Mitt Romney in the 2012 presidential election. First let’s look at the evidence which suggests this might very well have happened. The first presidential debate was on October 3rd, 2012. So in order to assess the debate as an inflection point, we want to look at polling on or before October 2nd on the one hand in comparison to polling on or after October 4th. With tracking polls we will want to compare polling on October 2nd with polling reported all they way out to October 10th in the case of 7-day averages. Finally, we will want to look at the current polls. Our questions: 1) Was there a Romney surge post-debate? 2) Has that surge continued/snowballed or has it receded? I have included all polls cited in the RCP average where there is at least a poll before the first debate and a poll after the first debate. Ideally, we will look at polls immediately before, immediately after, and current polling.

polling evidence of preference cascade

This table depicts the Romney deficit or surplus in support for president in polling before the debate, after the debate, and currently. The trend columns indicate the difference in Romney support between the pre-debate numbers and the post-debate numbers as well as the pre-debate numbers and the current numbers. A positive number (green) in the trend column means that Romney has gained support between the two polling periods. Note, I compare within polls so that we can observe trends irrespective of the sampling technique and likely voter screen differences between the included polls.

One fact that should be immediately apparent is that there has unquestionably been a Romney surge. No poll included above shows Obama gaining, within their poll, over the last four weeks. The best Obama does is in the Hartford/Courant and IBD/TPP polls, both of which show no change in the race. Among the polls that have a pre-debate poll and a current poll on the race, the average Romney surge is 3.7%. That’s strong evidence that Romney’s surge out of the first debate has been durable — not a bump based on favorable media coverage. Note, the media consensus over the last 3 debates (1 VP, 2 Presidential) have been that the Obama/Biden ticket “won” the debate (at least narrowly). Despite this, Romney has seen his surge in the polls stabilize. In fact, it is even better than that for Romney if we focus on the polls that have a pre-debate, post-debate, and current poll in the above table. Those five polls are here:

preference cascade evidence 2

Note that among these polls, the trend to the current polling (which has baked into it the VP and 2nd presidential debate performances) from the pre-first debate polls is larger than the trend to the polling immediately after the debate. The average gain from the pre-debate polls to the immediate post-debate polls is 4%. However, the average gain from the pre-debate polls to the current polls is 5.33%. Romney is doing better now than he was in the immediate aftermath of his unamimous first debate victory by a little less than 1.5% points — a 26% improvement on his post-debate surge. This suggests that despite two Obama/Biden “victories” in the debates, the Romney surge has, in fact, deepened. Why might this be? A big part of the story is the misunderstanding of what the first debate performance was about. Alot of the debate analysis has focused on “winning on points”…in other words, analysts have looked at the debates as a competition. If one participant “out scores” the other, then that participant won. The consensus after the first debate was that Romney won — i.e. out-performed Obama. He certainly did that, but that is not what instigated the surge towards Romney in the polls. What instigated the surge was a preference cascade – an inflection point that established Romney as a credible alternative to Obama.

What many missed in the coverage of Obama’s lead up until the first debate was the fact that, despite that lead, Obama never topped 50% in the RCP average. Let’s look at the trend here:

As you can see, most of the Obama’s lead in the RCP average was built on depressing Romney’s numbers below 47%…not on Obama getting above 50%. Only at one time in the race over the last 6 months did Obama approach 50%, and that was a consequence of the bump he recieved after the Democratic National Convention. Those numbers quickly dissapated, with Obama returning to that 47% equilibrium. In contrast, Romney’s numbers are improving even 2 weeks out from that debate performance. The answer to the puzzle lies in this observation: A majority of the voting public had already decided they did not support President Obama. What they had not decided, before the debate, was whether or not they could support Romney as an acceptable alternative. What that first debate performance accomplished for Romney wasn’t a victory on points — it wouldn’t have mattered if Obama had performed well (as he did in the second and third debates). What mattered is that Romney painted himself as a viable alternative to Obama — and that’s exactly what a majority of the electorate is looking for. And that’s why things are looking up for Romney…and very gloomy for President Obama. And it is also why the subsequent debates and debate performances have failed to slow/suspend the Romney surge.

Category:PoliSciPundit | Comments Off on A Preference Cascade – Part II | Author:

A Bad Lip Reading of 1st Debate

Thursday, 18. October 2012 6:32

It’s party time, chumps.

Category:PoliSciPundit | Comments Off on A Bad Lip Reading of 1st Debate | Author:

Why Obama is in Trouble

Wednesday, 17. October 2012 9:45

The difference between a 48 state blow-out and a statistical tie 3 weeks out in one graph:

Reagan v. Obama

Category:PoliSciPundit | Comments Off on Why Obama is in Trouble | Author:

The Family Guy Debate

Wednesday, 17. October 2012 8:56

This parody is just a little too close to reality for comfort…

Category:PoliSciPundit | Comments Off on The Family Guy Debate | Author:

2012 Presidential Election – Preference Cascade

Tuesday, 16. October 2012 9:11

A preference cascade in public opinion, or what is sometimes called an information cascade (Sunstein and Kuran build on the concept with their idea of an “availability cascade” with respect to the popularization of “novel” ideas), is an inflection point in public perceptions (in this case, of presidential candidate Mitt Romney) whereby a shift in public opinion builds on and reinforces itself…leading to a major and durable shift in public opinion. There is growing evidence that we may be witnessing a preference cascade in the 2012 presidential election. The idea here is that the unknown quantity in the 2012 presidential election was Mitt Romney — was he a viable president? Team Obama spent the summer (and some 150 million dollars) painting a portrait of Mitt Romney to the contrary — a cold, politically inastute, corrupt vulture capitalist with radical designs on the welfare state and a set of policies designed to cater to the wealthy at the expense of the poor and the middle class. However, one of the reasons most of the histrionics about campaign finance and political advertising is overblown is that voters, even low-information voters, discount information from partisan sources. That, of course, doesn’t mean they ignore it…and in a vaccum such information can sway voter opinion. But in a competition between political advertising vs. political news, the later has more credibility and hence is a more powerful informational trigger.

The first debate of the presidential campaign provided just such an opportunity for voters to view the candidates (or to witness the candidates in action through the filiter of media coverage), and given the near universal opinion that Romney won the debate going away, it constitutes a political information data point that had the potential to exert real influence over the course of this election. Not necessarily because the conventional wisdom is that Romney won (although that helps), but because of how he won — passing the “presidential” threshold test and contrasting the “real” Mitt with the Obama campaign’s caricature of him in political advertising. Aiding this process was the fact that the first presidential debate came at a time when low-information, undecided voters were just beginning to pay attention to the race. But while this makes a neat anecdotal story, what evidence is there that the Romney public opinion surge is real and durable rather than a mere “bump” from the favorable coverage of his first debate performance? The most significant evidence has been the fact that Romney’s lead (Mitt-mentum) has not only persisted through the VP debate, but has also continued to build. In other words, Romney has been gaining in the polls even into this week…a day before the second debate. The fundamentals of the US economy have always meant that Obama would be a weak incumbent…and his persistent inability to get within range of 50% support for his re-election (with a few exceptions) has afforded Romney the opportunity to mount a competitive challenge. If this is a preference cascade, then it bodes very ill for Obama’s reelection prospects. With opinion on Obama’s first term already baked in, the singular variable in the election was opinion on Romney. If Obama’s efforts to make Romney toxic have failed, then odds are we will inaugurate a new president in 2013. In my next post, I will discuss the public opinion evidence which suggests that there has been an inflection point in the race and a preference cascade in favor of Mitt Romney.

Category:PoliSciPundit | Comments Off on 2012 Presidential Election – Preference Cascade | Author:

Looming Debt-meggedon

Wednesday, 10. October 2012 7:58


Category:PoliSciPundit | Comments Off on Looming Debt-meggedon | Author:

Economic Malaise

Friday, 5. October 2012 9:55

Despite the unemployment rate dropping three tenths of a point in the latest jobs report (to 7.8 from 8.1), the economy only added just over 100,000 jobs…not enough to keep up with population growth. The rate drop is a consequence of a mid-year adjustment and the record-low workforce participation rate, as you can see in the second graph below.

Category:PoliSciPundit | Comments Off on Economic Malaise | Author:

Presidential Debate 2012

Thursday, 4. October 2012 0:45

Category:PoliSciPundit | Comments Off on Presidential Debate 2012 | Author:

%d bloggers like this: