About this Blog

RealClearPolitics HorseRaceBlog

By Jay Cost

HorseRaceBlog Home Page --> June 2008

A Frame of Reference for the General Election

Hillary Clinton's concession speech on Saturday marked the end of an amazing primary nomination contest.

Now, for the general. What to expect?

Political scientists have shown that "macro conditions" - presidential job approval, the state of the economy, how many terms a party has been in office - are extremely influential determinants of election outcomes. This year, if we were to adopt a strict "macro perspective," we would conclude that the Republicans are going down, and it will not be pretty.

Of course, these models are retrospective in character. Predictions for the future are based entirely upon past results, usually dating back to at least the 1948 election. Thus, almost all of the observations upon which the models are built occurred prior to the information age.

This fact may be of relevance this cycle.

Nowadays, partisan actors are aware of these macro models. Even if they have never read the scholarly articles where the models are offered and tested - they get the gist. They know which factors matter and how they matter. What is more, they have instant access to key points of data, receiving "real time" updates on everything from Bush's job approval to the unemployment rate. This is just one consequence of the explosion of information technology.

It points to a question. What does widespread knowledge of these models imply? Since political actors know how macro forces structure and condition political outcomes - can they use that knowledge to gain leverage over those outcomes? Can they master the political winds just as sailors have mastered the seas?

Both parties seem to be trying. Both seem to have changed their typical electoral strategies in light of extensive knowledge of the macro environment. These changes might be sufficient to alter the expected outcome.

Let's consider each party in turn.

The Republican Party seems to understand that it faces a bear market. After all, it has nominated a bear market candidate. John McCain is not the first, second, or even third choice of most Republicans. However, they believe he has cultivated a stable image as an anti-Bush Republican. Whether this belief is accurate, we do not know for sure. What we do know is that there is a non-trivial probability that it is accurate. Therefore, we can conclude that nominating McCain was the safe choice for the party, given the macro environment and the party's goal of electoral victory.

I would highlight the uniqueness of this behavior. Contrast the Republicans this year to the Democrats in 1920 and 1968. In 1920, the Democrats ran behind the League of Nations and, by extension, Woodrow Wilson. They were shellacked at the ballot box. In 1968, Lyndon Johnson's vice-president won the nomination on the first ballot despite the fact that his boss's job approval rating stood at just 35%.

How could the Republican Party behave in a different manner? After all, the party's organization is nothing like a Wall Street firm. Ideally speaking, a firm is structured so that relevant information is efficiently provided to decision-makers who formulate strategies to achieve the firm's goal. Republicans do possess a shared goal - electoral victory - but their party is not organized for the sophisticated pursuit of that goal. Instead, the party organization is a federated hodge-podge full of obscure procedures that nobody has heard of. How did it make a decision like this?

We might conclude that it was just dumb luck. The party got lucky with McCain. I find that explanation insufficient - though luck might have played a role. I would point to how knowledgeable people in the party are nowadays. Republicans of all stripes - party leaders, party activists, and even average partisans - know a great deal about the position the party is in. Organizational inefficiency should not matter as much when people generally feel the same way.

My sense is that most Republicans intuited that they might well lose the upcoming election, and so were less inclined to take unnecessary chances. In other words, their extensive knowledge of the macro structure made them relatively risk averse. They would take a candidate who, as president, will give them a lower expected "partisan payoff" if it means a higher probability of that candidate winning. The Republicans also knew that, though they didn't much like McCain, he had a seemingly good image among independents, who are the voters they need in a year like this. All in all, the party exhibited a kind of grudging unanimity in both its aversion to risk and its assessment of the situation: "McCain is a safer, but less satisfying, bet for us; in an ominous time such as this, he is the best choice."

We can tell a similar story about the Democrats, though they are quite different from their Republican counterparts. In many respects, they remain a more heterogeneous group. They are also in a much different situation than the GOP, and so their aversion to risk seems to have been different.

In the last seventy-five years, there has been a method by which Democrats have won. Victorious Democrats have succeeded by making (whom we inefficiently refer to as) "working class whites" comfortable that the party was "on their side." Even Kennedy did this. That is why his primary victory in West Virginia was so important. Rarely have victorious Democrats won white voters outright. Instead, they split them with the Republicans. This has been a crucial component of Democratic victory.

This year, the Democrats nominated a candidate who seems less able to do this than his former opponent. Obama lost the white vote in most primaries, often by lopsided margins. In light of the history of the party, he seemed not to be the safer candidate.

As we all know, there were two factions in the Democratic electorate that were equally divided. The superdelegates had to break the tie. They sided with Obama. Why did they do this? Why not go with the safer bet?

My sense is that it occurred for many reasons, including the following two. First, the party elite (and the intellectual class of the party, broadly defined) preferred Obama to Clinton. Either because of his appeal, or lingering Clinton fatigue, they wanted him over her. Second, thanks to easy access to relevant data, they believed they could afford the additional risk he might bring. Their knowledge of the macro environment made them more willing to take on a candidate with a possibly lower likelihood of victory because they expect a higher expected partisan payoff should he win. This is known as risk-accepting or risk-tolerating behavior; it is different from the GOP's orientation. The Democrats generally expect to win, so they are inclined to take a little electoral risk to get a candidate they really like.

So, both parties manifested signs of sophisticated thinking in pursuit of a goal - even though the manifestations were quite different. In both cases, information made a crucial difference. Party actors had up-to-date knowledge of relevant variables, and acted in light of that knowledge. The Republicans were aware of their dire straits and, accordingly, made a risk-averse choice. The Democrats were aware of how favored they are and made a risk-accepting choice. An important precondition of all this is the information age. It is simply easier for people of all classes to acquire information nowadays, and thus easier for them to make sophisticated choices like this.

Interestingly, while they are pursuing different goals in different manners, both parties are putting the same kind of stress on the electoral system. Republicans looked at the macro structure and determined that McCain might turn a probable defeat into a possible victory. Democrats looked at the same structure and determined that while Obama could probably not pull off an enormous win, he could still win in a year like this. Both candidates were thus selected with an eye to having the final result closer than the macro models predict. Both are testing the tensile strength of the macro structure of electoral conflict. Republicans picked a guy they don't like but who might pull the upset. Democrats (with the final say) are trading a possible landslide for their first choice.

What does this mean for November?

Ultimately, it comes down to whether human agency can affect processes that are largely governed by macro conditions. Presidential elections are one such process. We know they are governed in part by vague, impersonal forces. To what extent are they also governed by the actions of human beings? We have seen both parties try to influence this election. Both have positioned themselves in light of what they know. Will their positionings make a difference, or is the outcome already written in the stars?

That's a big question when you think about it - and it is one that we only have a weak understanding of. Personally, I have read but a handful of books that integrate structural forces with individual endeavor into a cohesive narrative of political change. Most works are either stories of personalities or stories of structures. Rarely does anybody look carefully at both.

So, I don't know what all this strategic positioning means for November. It could mean nothing. It could mean everything. We'll have to wait to see.

Personally, I wouldn't have it any other way. The surprise is part of the fun!

-Jay Cost

Obama Wins On Points

Last night's events were a microcosm of this whole nomination battle. Barack Obama obtained the endorsement of a sufficient number of superdelegates to clinch the nomination. This was despite the fact that he and Clinton - once again - split the contests at stake. Clinton carried South Dakota easily. Obama carried Montana easily.

In the wake of the 1968 convention in Chicago - the Democratic Party opened the nomination process to the public at large. The Republicans followed suit a few years later, and today there is a wide and deep sense that what V.O. Key called "the party in the electorate" chooses each party's presidential nominee.

This year, the Democratic party in the electorate has split right down the middle. We saw that last night, just as we have seen it all season long. If you count up the votes from all contests where both candidates were on the ballot, and include caucus estimates, you come up with Obama having a lead in the popular vote of 151,844 votes out of nearly 36 million cast for the two of them. However, that excludes Michigan, a state with up to 2.3 million Democratic voters that did choose between the two candidates. So, as it stands there is no way to know whom the party in the electorate generally prefers.

This speaks to an important point. There is a thing called public opinion. It is what it is, whether we are aware of it or not. It is "out there" somewhere. We only have imperfect ways to measure it. We have public opinion polls, which as we all know are imperfect. We also have elections, which like polls are metrics for gauging public opinion. These can be imperfect, too. When the difference between support for candidates is very small, it may be that the electoral process cannot determine which candidate's support is greater. After all, the electoral process is the creation of human beings with their own interests and agendas. It is possible for measurement problems to occur. Something like this happened this year. There was an excruciatingly close division between the candidates - and the imperfection of the Democratic nomination process, wherein Michigan was not fully included in the contest, means that we are unable to determine who actually had the greater support. As far as we know, the vote was split.

Thus, Obama has won the Democratic nomination not because his voting coalition is larger than Clinton's. As best we can tell, they are of equal size. Instead, Obama has won because his coalition is more efficient at producing delegates than Clinton's coalition. Obama's relatively narrow vote lead has produced a relatively wide pledged delegate lead, which has in turn produced an even wider lead in superdelegates. The following chart indicates this point by measuring the number of votes per pledged delegate. The idea here is that, the lower the number of votes per pledged delegate, the more delegates a single vote produces for the candidate, and therefore the more efficient a candidate's coalition is.

Votes Per Pledged Delegate.gif

As we can see - Obama's voters are worth more delegates. Put precisely, there are 10,237 voters for every Obama pledged delegate and 10,807 voters for every Clinton pledged delegate. That's a difference in Obama's favor of 570 voters per delegate. That might not seem like it would make a big difference, but it most certainly has. If the "votes per pledged delegate" metric were equal for Clinton and Obama - Obama's pledged delegate lead would drop from 106 to 12.

[Note that the popular vote used in the above chart does not include the Michigan vote while the delegate counts do include the Michigan delegates. This was done to account for the fact that the Rules and Bylaws Committee did not use the Michigan vote to estimate the delegate allocation. If we were to include the Michigan vote by allocating to Obama the uncommitted, Obama's voters actually become more efficient.]

Does any of this mean that Clinton, not Obama, "should" be the nominee? No. By our imperfect metrics for measuring the opinions of the public, we must conclude that there is no clear public choice.

So, Obama has scored what amounts to a win on points. He did not score a knockout. Clinton's invocation of "18 million votes" last night reminded me of Jake LaMotta's taunt of Sugar Ray Robinson in Raging Bull, "You didn't get me down, Ray!" Indeed, Obama won the nomination on a night that Clinton still managed to win another contest.

From this, I would suggest that, as a prelude to unifying the party, both sides need to be a little modest.

The Clinton people need to recognize that it is not coincidence that Obama's vote was more efficient. I have discussed this before. Part of this had to do with the fact that the delegate allocation system contains biases that happened to favor Obama. However, part of it had to do with the fact that the Obama campaign had a better understanding of the system. It found the possibilities and made the most of them. What's more, the Clinton campaign let it do this. Simply put, Obama out-maneuvered Clinton. Clinton supporters need to respect this.

Meanwhile, Obama supporters need to recognize that their candidate is the victor not because he put together a majority coalition, but because he out-maneuvered Clinton. This was a highly intelligent strategy, but it was not a grand feat of majority building. Obama supporters need to recognize that their candidate won not because "the people had their say," but because his campaign out-smarted her campaign. Accordingly, they need to respect the candidate whom they could not beat in a straight-up fight for votes.

-Jay Cost

A Review of Obama's Voting Coalition, Part IV

I'd like to bring my analysis of Obama's voting coalition to a conclusion today by tying together two points made last week. On Wednesday, we observed that voting groups tend to vary their support for one candidate or the other according to region. On Thursday, we noted that Clinton has performed better in the swing states.

The assertion I would like to make today is that this is not coincidence. It is not that Clinton has won the border states, the industrial Midwest, and the swing states. It is that because she has won the border states and the industrial Midwest that she has won more swing states. She is not uniquely capable in swing states. Instead, swing states are clustered in regions that favor Clinton. As we move into regions known to be more favorable to Obama, above all the Pacific West, we find him doing better in those swing states.

To demonstrate this, let's break down demographic groups in just the swing states by region. This time, we'll use broader Census Bureau definitions of region - Northeast, South, Midwest, and West. We'll also take a broad definition of swing states. We'll include all states that have held primaries in 2008 that Bill Clinton won in 1996, and that the Democrats either lost in 2004 or won by less than 5%.

We'll start with gender and age among whites.

Whites By Gender and Age.gif

Three observations are of note.

First, within regions, we see differences in support by demographic group. For instance, white seniors in the Northeast are much less partial to Obama than white youths in the Northeast.

Second, within demographic groups, we see differences in support by region. White voters in any given group are most likely to support Obama in the West, least likely to support him in the South, with the Northeast and the Midwest coming in the middle. We saw something like this for all states in Part II; now we see that it applies to the swing states just as much.

Third, notice that the Midwest appears to be in the middle. This is a deceptive position because there is a wide variation within the region. The following chart has the details for the Midwest.

Midwestern Swing States.gif

What we see here is that Michigan, Ohio, and Missouri tend to fall between the South and the Northeast. The region as a whole is "pulled" toward Obama by the fact that he performed strongly in Wisconsin.

With this caveat in mind, let's examine Obama's swing state performance by region and socioeconomic status.

Socioeconomic Status.gif

We find the same type of results. There is intra-regional variation by demographic groups - e.g. voters in a given region with college degrees are more inclined to Obama than voters in the region without college degrees. There is also intra-group variation by region. The college educated in the West are more inclined to Obama than the college educated in the South.

Once again, we see the regions fall in the same order. Holding socioeconomic status constant, the South is least inclined to Obama, the West the most, and the Northeast and Midwest falling in the middle. If we broke the Midwest down by states, we would again see Wisconsin most favorable to Obama, Ohio and Michigan least favorable to him, Missouri in the middle.

Finally, let's look at voters by region and type of area (urban, suburban, rural).

Residential Area .gif

Again, we find that within a region, type of living area has an effect on vote choice. We also find that within a type of living area, region has an effect. As per usual, the West is most partial to Obama, the South the least, and the Midwest and Northeast fall somewhere in the middle. The only exception is urban populations, all of which seem equally inclined to Obama. This is probably due to the fact that African Americans tend to be concentrated in urban areas regardless of region.

So, what can we conclude? While it is truthful to assert that Clinton has done better in the swing states, this seems to be largely due to the fact that swing states are in her best regions. Regional differences seem to account for a great deal in the variations in vote choice - be it in swing states or otherwise.

-Jay Cost