Mass Partisan Resurgence and Candidate Polarization

I’m working on a paper discussing the impact of candidate polarization on party’s capacity to predict the vote. Most people think that if candidates polarize, partisanship will seem stronger. I disagree. I discuss why below. I’d love to hear some comments.

Evidence of partisan resurgence exists since the 1960’s and 1970’s among the United States electorate . Particularly, party seems to have a greater capacity to predict the vote now than ever seen in survey data. If true, the long lamented partisan dealignment has ended and a new partisan voter presents itself for analysis. Consider, however, that party’s capacity to predict the vote is partially dependent upon the ideological positions
of the major party candidates in an election. Why is this true? A series of hypotheticals will guide us.

Assume n voters with single-peaked ideological preferences along a unidimensional scale. Assume two parties that are in parity in the electorate; Republicans and Democrats. Most Republicans will be to the ideological right of the median voter while most Democrats will be to its left. Some, in the ideological center, will have ideological preferences that conflict with the traditionally held views of their chosen party. On the scale, for example, a few Republicans will be placed to the left of the median voter while a few Democrats will be placed to its right. I will call these voters cross-partisans. Assume stability in ideological positions and partisan allegiances among voters.

Onto this space, we can place the two major party’s political candidates. Voters will select one of the political candidates based on a calculus that weights ideological preferences and partisan allegiances. The weight given one or the other is likely to change depending on the position of the candidates. To start, take the scenario where both candidates strategically position themselves at the median voter. Voters have no incentive to weight ideological concerns given that the candidates do not differ ideologically. Thus, all weight in the voting calculus is assigned to partisan allegiances. Here, partisanship will be a very strong predictor of the vote.

However, the candidates may, given sincere policy preferences, move closer to their traditional ideological poles; the Democratic candidate moves left while the Republican moves right. Suppose they move symmetrically away from and equidistant to the median voter. Now ideological considerations play a role for voters. For most voters, ideology and partisanship are reinforcing. For cross-partisans, they are not. The vote decisions of these voters is dependent upon how much they weigh one or the other as candidates shift from the median voter. After a small shift, only a small number of cross-partisans may weigh ideology more than partisanship. This will, nonetheless, result in the apparent decline in the capacity for partisanship to predict the vote. As the candidates move further from the median voter and become more extreme, cross-partisans will become increasingly disenchanted with the candidates. They will incur representational loses which will have them assign greater weight to ideology over partisan allegiance. This will make them more likely to cross party lines and vote for the ideologically proximate candidate. In this scenario, also, polarizing candidates should more clearly result in an apparent decline in partisan voting.

Candidates may not, and often do not, polarize symmetrically. Rather, in elections, one party’s candidate may moderate, remain unchanged or polarize while, independently, the other party’s candidate faces the same options. While the partys act independently, these decisions are often made in a dynamic and strategic contexts. For example, the outcome of past elections helps to influence the decision as well as the concurrent activities of the opposed party. What can we say about the apparent strength of partisanship in predicting the vote when only one major party candidate shifts over successive elections while the other does not? Let’s suppose the Republican candidate is just to the right of the Democratic cross-partisans and s/he stays there. The Democratic candidate begins to the left of the Republican cross-partisans but moves left. The baseline result for the apparent predictive power of partisanship depends on how much the cross-partisans weigh ideology and party allegiance. If party allegiance is completely deterministic then party is a perfect predictor of the vote. If ideology plays a role, then some cross-partisans will cross party lines in their vote and the strength of party allegiance will wane. Relative to the resulting baseline (whatever it may be and dependent upon the voting calculus by cross-partisans), what happens as the Democratic candidate shifts left?

At some point, the Republican cross-partisans will have both ideological and partisan incentive to vote Republican. This is the point at which these voters are ideologically closer to the Republican candidate over the Democratic candidate. Now, liberal Democrats along with a shrinking number of Democratic cross-partisans (who weigh party a great deal) are voting Democratic while Republican cross-partisans, most of the Democratic cross-partisans (who are growing more disenchanted with the Democratic candidate for ideological reasons) and conservative Republicans are all voting Republican. The extent to which party seems to strengthen as a predictor of the vote depends upon the ratio between the number of Democratic cross-partisans who were voting Democratic before the candidate’s leftward shift but now vote Republican and the number of Republican cross-partisans that are now voting Republican given the shift. It is likely that such Republicans outnumber such Democrats (because Republican cross-partisans no longer face countervailing ideological and partisan forces while Democratic cross-partisans still do). If true, partisanship will seem to increase somewhat even though no voter changed position. This is in line with the current thinking in political behavior research.

As the Democratic candidate continues to shift leftward, an increasing number of Democrats will have to weigh ideology and partisan allegiance once the Republican candidate is ideologically proximate to them. Now, almost no Democratic cross-partisans vote Democratic and some left leaning Democrats are also settling for the Republican candidate. While the Democratic candidate earns only Democratic votes, the Republican candidates voters include all Republicans and a growing number of Democrats (cross-partisans and otherwise). As the ranks of Republican party voters gets contaminated with Democratic support, partisanship will seem to decline in its capacity to predict the vote. What all this shows, is that, by in large, polarizing candidates should result in an apparent weakening of party’s capacity to predict the vote.

Fitting Multilevel Models When Predictors and Group Effects Correlate

Andy and I (along with David Park, Boris Shor and others) have been working on various projects using multilevel models. We find these models are often optimal, particularly when dealing with small sample sizes in groups (individuals in states, students in schools, states in years, etc.). Many social scientists who come from an econometric background are skeptical of multilevel models because they model varying intercepts with error (often called random effects). With modeled varying intercepts, there’s the possibility that the predictors will correlate with the varying intercepts problematically. Andy and I wrote a paper discussing how this can be solved. Download file

Take a simple equation where some outcome is predicted by varying intercepts for groups and a covariate. Also assume that the covariate and the group effects correlate. A problem emerges because the correlation between the covariate and the varying intercepts falls into the error in the level 2, varying intercept regression equation. This error becomes part of the error in the level 1 model (as is evident when one substitutes the level 2 equation into a level 1 equation), and violates the Gauss-Markov assumption that predictors and errors cannot correlate. This violation will result in problematic estimates for the predictor in the level 1 equation. But there’s a straightforward solution to this problem. We can solve this problem of modeling with more modeling. One can take the correlating predictor, calculate its mean per group and include it in the level 2 modeled varying intercepts equation.

This new predictor will capture the problematic correlation before it falls into the level 2 error. It will also offer a substantive result which may be useful in the research. This method can be applied in any software that allows for model varying parameters (BUGS, R, STATA). I say varying parameters because all of the above can generalize to modeled varying coefficients.

Simulations can illustrate the problem and the solution further. First, we generate a random normal predictor of length 100 with a mean of 0 and a standard deviation of 2. Then, we generate an outcome (often called a dependent variable) that is equal to the predictor plus random normal noise with a mean of 0 and a standard deviation of 7. This ensures a strong, but not perfect, correlation between the two variables. Units effects are added to the outcome by adding a random normal component with a nonzero mean to each quarter of the data. So, for example, a set of random normal values with a mean of 1 and a tight standard deviation of .001 are added to the first 25 observations in the outcome. The means for the next three quarters are -1,-3 and 2. The standard deviations remain tight at .001.

We start by predicting the outcome by the varying unit effects and the explanatory variable free of unit effects. These results should not be problematic since the units and the predictor do not correlate. Next, we estimate the same equation but with a predictor that, like the outcome, varies across the units.

To see if the solution highlighted above works as promised, we run a third simulation where the correlation exists but the mean of the predictor per unit is included as a group-level predictor.

We estimate each equation 1000 times and record the coefficient and standard error of the key predictor. We plot a histogram of the t statistics (the coefficient divided by the standard error) calculated in each of the 1000 simulations.

forblog.png

The figure shows that the t statistic of beta for the model where the predictor does not correlate with the units tends to be smaller than the model where the correlation exists. The larger t statistic in the second plot results in an inflated sense of statistical significance in parameter analysis and a greater tendency to falsely reject the null hypothesis in research works, as mentioned earlier. This is the problem that researchers are cautioned against when estimating modeled varying intercepts. The problem is thought to be so severe, that this model is often cast
aside as inviable.

The third plot shows the same model but with the added level 2 predictor (the correlating predictor measured at its mean per group effect). The t statistic for the key predictor looks virtually identical as in the model with no correlation between the group effects and the predictor. The additional group-level covariate successfully accounted for the correlation before it fell into the group-level error term and
caused a problem.

Al Qaeda and Iraq: An Under Explored Link

The United States is amidst a war in Iraq. Entry into the war by the U.S. was justified on multiple grounds. One such justification was the connection between Al Qaeda and Iraq. Political leaders have described an Iraq that has provided aid and comfort to terrorist operatives. As late as June 17, 2004, Vice President Richard Cheney argued, “There clearly was a relationship. It’s been testified to. It goes back to the early ‘90s. It involves a whole series of contacts, high-level contacts with Osama Bin Laden and Iraqi intelligence officials.” The issue continues to be debated. What has gone largely unnoticed is a bit of history that closely links Al Qaeda with the situation in Iraq in a fashion rarely discussed.

The Al Qaeda/Iraq Link
In 1979, the Soviet Union began an effort to invade and conquer Afghanistan. The Afghanis and their allied groups (including Osama Bin Laden’s network and the United States) were successful in repelling the Soviets ten years later. Bin Laden’s associates set-up a theocratic dictatorship in Afghanistan ruled as an Islamic state. Bin Laden had such designs for other nations including his home nation of Saudi Arabia.
A strong requirement of the Bin Laden’s brand of Islam is that non-believers (often called “infidels” by Islamists) are unwelcome on land that is decreed holy. Saudi Arabia houses two of Islam’s holiest sites: Mecca and Medina. To avoid entry of non-believers onto this holiest of Islamic lands, Bin Laden offered the manpower, training and equipment he had procured in Afghanistan to protect Saudi Arabia following Iraq’s invasion of Kuwait; a small, neighboring country with great oil reserves and desirable access to the Persian Gulf. Bin Laden strongly resisted American or other “infidel” presence on this Arabian holy land even to protect against the designs of a secular and oppressive regime. The Saudi government, uncertain of Bin Laden’s capabilities and wary of an extremist presence on their land, rebuked his offer.
Nations from all over the world sent soldiers to the region to fight off Hussein’s invasion of Kuwait and fend off any threat to Saudi Arabia. Further, U.S. forces remained long after the Persian Gulf War was won. This may have been to defend against Hussein’s violations of U.N. sanctions (e.g., the no-fly zone), to more quickly secure the region should an aggressive regime act up (e.g., Iran), or simply inertia. Whatever the rationale for their continued presence, non-believers on holy land angered Bin Laden and his followers immensely. This represents the first of Bin Laden’s great laments against the west; great in that they would spark the first acts of violence specifically targeted at the U.S. The primacy he gave to this issue is evident from responses he gave during media interviews granted in the middle through later 1990s. He said,

When the American troops entered Saudi Arabia, the land of the two holy places [Mecca and Medina], there was a strong protest from the ulema [religious authorities] and from students of the sharia law all over the country against the interference of American troops. This big mistake by the Saudi regime of inviting the American troops revealed their deception. They had given their support to nations that were fighting against Muslims.

Now the people understand the speeches of the ulemas [Muslim scholars trained in Islamic law] in the mosques – that our country has become an American colony. They act decisively with every action to kick the Americans out of Saudi Arabia. What happened in Riyadh and Khobar [when 24 Americans were killed in two bombings] is clear evidence of the huge anger of Saudi people against America. The Saudis now know their real enemy is America.

I believe that sooner or later the Americans will leave Saudi Arabia and that the war declared by America against the Saudi people means war against all Muslims everywhere. Resistance against America will spread in many, many places in Muslim countries. Our trusted leaders, the ulema, have given us a fatwa that we must drive out the Americans. The solution to this crisis is the withdrawal of American troops … their military presence is an insult for the Saudi people.
The Independent, Interview by Robert Fisk, 1996

We have declared jihad against the US, because in our religion it is our duty to make jihad so that God’s word is the one exalted to the heights and so that we drive the Americans away from all Muslim countries. As for what you asked whether jihad is directed against US soldiers, the civilians in the land of the Two Holy Places [Saudi Arabia] or against the civilians in America, we have focused our declaration on striking at the soldiers in the country of The Two Holy Places. The country of the Two Holy Places has in our religion a peculiarity of its own over the other Muslim countries. In our religion, it is not permissible for any non-Muslim to stay in our country. Therefore, even though American civilians are not targeted in our plan, they must leave.

We ask about the main reason that called for this explosion [the bombings of United States troops in Riyadh and Dhahran]. This explosion was a reaction to a US provocation of the Muslim peoples, in which the US transgressed in its aggression until it reached the qibla of the Muslims in the whole world. So, the purpose of the two explosions is to get the American occupation out [of Arabia]. So if the U.S. does not want to kill its sons who are in the army, then it has to get out.

It is known that every action has its reaction. If the American presence continues, and that is an action, then it is natural for reactions to continue against this presence. In other words, explosions and killings of the American soldiers would continue. These are the troops who left their country and their families and came here with all arrogance to steal our oil and disgrace us, and attack our religion.
Interview by Peter Arnett, 1997

The call to wage war against America was made because America has spear-headed the crusade against the Islamic nation, sending tens of thousands of its troops to the land of the two Holy Mosques over and above its meddling in its affairs and its politics, and its support of the oppressive, corrupt and tyrannical regime that is in control.
Frontline, 1998

We can learn more of Bin Laden’s thinking from his own writings. Telling is the title of a key epistle he issued on August 1996, considered his first international fatwa . It was called the “Declaration of War against the Americans Occupying the Land of the Two Holy Places.” Naval War College Associate Professor Ahmed S. Hashim wrote, “…the focus of Bin Laden’s anger in the 1996 epistle was the continued American “occupation” of the land of the holy places…” In the declaration, Bin Laden wrote of the importance of unity among Muslims who found many reasons for in-fighting. He encouraged them to ignore their differences and to focus on one thing—removing “infidels” from the holy land. He wrote:
If there [is] more than one duty to be carried out, then the most important one should receive priority. Clearly after Belief (Imaan) there is no more important duty than pushing the American enemy out of the holy land. . . . The ill effect of ignoring these [minor] differences, at a given period of time, is much less than the ill effect of the occupation of the Muslims’ land by the great Kufr [unbelief].
In 1998, he wrote a declaration entitled, “World Islamic Front for Jihad against the Jews and Crusaders”. In it, he wrote:
Since God laid down the Arabian peninsula, created its desert, and surrounded it with its seas, no calamity has ever befallen it like these Crusader hosts that have spread in it like locusts, crowding its soil, eating its fruits, and destroying its verdure; and this at a time when the nations contend against the Muslims like diners jostling around a bowl of food.
Further evidence of the primacy with which Bin Laden viewed this issue exists. For example, a top international relations scholars, Robert Jervis, recently wrote, “Bin Laden had attacked American interests abroad and from early on sought to strike the U.S. homeland. His enmity stemmed primarily from the establishment of U.S. bases in Saudi Arabia…”
One may argue that this revealed lament may not reflect Bin Laden’s sincere sentiments. For example, the non-believer’s presence in the holy land may have been an excuse to take on the only remaining superpower after the Soviet Union’s demise. Just as Bin Laden believes the Soviet Union’s demise to be the fulfillment of Islamic destiny (and, therefore, his duty to have promoted), perhaps he believes the powerful U.S. must now be similarly humbled. Or, Bin Laden’s words may disguise a much more personal sort of vengeance for the substantial financial loss and forced ouster he suffered primarily at the hands of Saudi Arabia and the U.S. while taking refuge in Sudan.
While we cannot know for sure what drove Al Qaeda’s violent attacks toward the U.S. and Islamist laments are many (e.g., U.S. support for Israel), what can be said with certainty is that the U.S. presence in Saudi Arabia was, time and again, a foremost, expressed source of great consternation for Bin Laden and this followers. It was Hussein’s ambitions to rule a greater Arabia that brought the American soldiers to the Muslim holy land. The U.S. presence, perhaps more than anything else, sparked Bin Laden’s hatred of the U.S. and motivated the escalating tit for tat game between the west and Bin Laden’s terrorist organizations.

War with Iraq
For Al Qaeda, this culminated in massive attacks, organized for years that we have come to know as September Eleventh. For the United States and our (shifting) coalition of allies, the retaliation continues to this day. As an element of the retaliation, the U.S. has invaded and overseen regime change in Iraq. The rationale for this action included the threat Hussein posed to allies in and around his nation, stockpiles of WMD, efforts at attaining further WMD (such as the attempt to purchase banned missiles from North Korea), violations of United Nations sanctions, high-level contacts with terrorist organizations and others. Given a study of Bin Laden’s motivations, the most relevant rationale for removing Hussein from power may be to secure the region so that a U.S. military presence would no longer be needed. That’s just what happened in the area most in contention, Saudi Arabia.
In April 29, 2003, U.S. Secretary of Defense Donald Rumsfeld announced the withdrawal of U.S. troops in Saudi Arabia. However, he was unwilling to admit that the withdrawal had anything to do with the laments of Islamic terrorists. London Daily Telegraph reporter David Rennie wrote,

Withdrawal of “infidel” American forces from Saudi Arabia has been one of the demands of Osama bin Laden, although a senior U.S. military official said that this was “irrelevant”.

Despite American insistence that the withdrawal had not been “dictated” by Al Qaeda and that bin Laden was “irrelevant”, there can be little doubt that undercutting a central plank of Al Qaeda’s platform is one of several advantages offered by withdrawal from Saudi Arabia
(4/30/2003).

Al Qaeda targeted the U.S. when it did primarily because of “infidel” presence in the holy land. This presence was motivated by Iraqi aggression and, specifically, the effort to protect Saudi Arabia from meeting a fate similar to Kuwait.
In this way, perhaps Iraq and Al Qaeda are linked to a greater extent than has been discussed in public or elite discourse. This link does not show cooperation between Iraq and Bin Laden, as has been sought by some, but it does tie our policy toward Iraq to the effort to defeat Islamic terrorism.

Justification for U.S. Policy?
Does this link offer a rationale for the U.S.’s entry into Iraq? It does not seem so from the Bush administration’s rhetoric, which has been virtually absent on this issue. We can speculate as to why. Perhaps the Bush administration did not regard the military presence in Saudi Arabia as important when considering the U.S.’s post-9/11 policy toward Iraq. There were certainly other issues at play that provided rationales (informed or not) for our invasion.
On the other hand, it is possible that the administration was unwilling to focus on this issue because it would seem to grant Bin Laden a victory. His, perhaps, greatest contention with the U.S. had been resolved in a fashion that he had called for time and again; withdrawal of American soldiers from the holy land. This alone is too simpleminded in that Bin Laden certainly would not have wanted, as a trade-off, ten of thousands of troops in Iraq. Nonetheless, the presence of non-believers in Saudi Arabia is no more and this may have been more than U.S. officials would like to admit.
Certainly, from the Bush administrations point-of-view, military action to neutralize Hussein would have been deemed essential toward the long-term stabilization of the Middle East and, thus, the eventual defeat of Islamic terrorism. One may note that stabilization has hardly come as yet. However, an Iraq with Hussein and, later, his progeny in power along the threat they would have posed to the Saudis and other of the regions U.S. allies would likely have postponed the possibility of stabilization only longer.
Seen in this light, the Bush administrations targeting of Iraq may very well have stemmed from a larger strategy to defeat Bin Laden and his networks of terror. A strategy that included removing one of their main tools for recruitment and rationales for attack. The administration’s silence on this motive leave us only to conjecture as securing Saudi Arabia may simply have been an unintended (and as Jervis explains, ironic ) consequence of the war with Iraq.

Conclusion
The Al Qaeda/Iraq link, as described by officials in the Bush administration, has been an allusive one. The 9/11 Commission report found scant evidence of a possible attempt at an Al Qaeda/Iraq alliance. But, perhaps, this is the wrong sort of link to dwell upon. Action against Iraq in response to 9/11 can be better justified by a need to stabilize the region so that a U.S. military presence is not required on the holy land of Saudi Arabia. What is often ignored is that the removal of these forces is among the most important of Al Qaeda’s laments.
It is unclear how salient this link was in the minds of Bush administration officials. It certainly does not reveal itself in their rhetoric. There may be reasons for this (e.g., not wishing to seem as if the terrorists got what they wanted). Or, this link may not have calculated very highly in their deliberations. Regardless, the removal of Saddam Hussein, in that it secures Saudi Arabia, is a big step toward removing the impetus for further terrorist activities. Although removing one primary motivation for Islamic terrorism toward the west may have caused many others, it still remains to be seen if short-term sacrifices will beget long-term gains.

Midterm Balancing Still True?

Bob Erikson, Chris Wlezien and I have been working on a paper about midterm balancing. We see it as still very viable and find solid evidence that it happens. In what follows, I have copied in portions of the paper. They will reveal what we did, what we found and what it all means. The paper is under review at POQ.

In 1998, as the unpopular impeachment of President Bill Clinton was unfolding, Clinton’s Democrats gained seats in the House of Representatives. In 2002, in the shadow of 9/11, President George W. Bush’s Republicans gained House seats as well. These two recent instances might make it seem commonplace for the presidential party to gain House seats at midterm. Indeed, the early interpretations of 2002 by Jacobson (2003) and Campbell (2003) emphasized the theme that this election unfolded as normal politics. Jacobson even chose not to remark about the historical significance of the presidential party gaining seats. The historical pattern, of course, is that the presidential party loses seats at midterm. This in fact had been more than simply a pattern, and almost a deterministic law of politics. From 1842 through 1994, the presidential party gained seats (as a proportion of the total) only once—in 1934 as the FDR-led Democrats’ surged with a gain of nine seats. This was a spectacular run of 38 presidential party losses in 39 midterm elections. Clearly, forces are at work in American politics to diminish the electoral standing of the presidential party at midterm.

The two most common explanations for midterm loss come under the headings of “coattails” and “balancing.” The coattail explanation begins with the congressional vote in the presidential election year prior to the midterm election. It holds that the surge in support for the presidential winner (“coattails”) artificially inflates support for the presidential winner’s party in the presidential year (A. Campbell, 1966; Hinckley, 1967; J. Campbell, 1985; Campbell, 1991). The balancing explanation focuses on the congressional vote in the second of the two elections. It holds that the midterm electorate supports the out-party to push Congress in the opposite ideological direction of the president in order to achieve greater ideological balance in government (Erikson, 1988, 2002; Alesina and Rosenthal, 1995; Mebane, 2000; Mebane and Sekhon, 2002; see also Fiorina, 1996, regarding balancing more generally).
What do the violations of the midterm loss rule in 1998 and 2002 teach us about the competing theories of midterm loss? The coattail theory provides a handy excuse for the two exceptions. Since there evidently were no presidential coattails in either 1996 or 2000, there were no coattails to withdraw in 1998 or 2002. Balance theory, one could argue, was more discredited by the 1998 and 2002 exceptions (Campbell, 2003), particularly since moderate voters had every incentive to tack right in 1998 and left in 2002. Of course the theory could be salvaged by accounting for the effect of the Clinton impeachment in 1998 and 9/11 in 2002. While this type of post-hoc rationalization normally is to be discouraged, one could reasonably argue that these two events were abnormally large interventions that moved voters toward rather than away from the sitting president’s party at midterm.
In this paper, we offer a new test for ideological balancing as a source of midterm loss. For this test, we exploit survey researchers’ frequent monitoring of the “generic” congressional vote during midterm years. The generic poll question asks respondents which party they plan to vote for in the upcoming congressional election. These generic polls can provide crucial clues regarding the timing of the electorate’s shift from the presidential or “in” party to the “out” party. To begin with, consider respondents who are asked their congressional preferences early in the midterm year, at roughly the halfway mark of the election cycle. Unlikely to have given much if any advance thought to their vote for the House of Representatives at such an early date, they will offer poll responses that are too premature to incorporate beliefs about voting for the out-party as an ideological counterbalance to the president. Now, consider respondents asked for their generic vote on the eve of the election. By this point in time, their thoughts may have turned to the election and are likely to incorporate any cognitions they will make about the need for ideological balancing. Our analysis supports this hypothesis.

Recent midterm gains by the president’s party would seem to cast doubt on the notion that the electorate moves toward the out party in midterm elections as an ideological counterweight to the president. This paper provides new evidence in support of the balancing theory, by exploiting the predictive power of generic polls of congressional party preferences at midterm. At regular intervals throughout the midterm years from 1946 through 2002, pollsters have monitored congressional party preferences via their generic poll questions. Contrary to the frequent skepticism in the media, the generic polls are quite useful for forecasting midterm election outcomes, requiring only the proper discounting of the size of the partisan gap in the polls. As we have seen in this paper, the predictive power of generic polls is enhanced further by taking into account the party of the president. This provides the evidence for ideological balancing.
Our demonstration is simple. We regress the midterm vote on the generic polls plus a dummy variable for the presidential party. When the generic polls are measured early in the midterm year, the presidential party has a visible and highly significant negative “effect” on the vote. Over the course of the campaign, this effect declines toward zero. The only plausible interpretation is that (with 2002 a clear exception) the electorate becomes more sympathetic toward the out party as the campaign progresses. At the start of the midterm year, the electorate responds to the generic poll question with a party preference that does not take into account the party of the president. As the campaign progresses and voters focus more on the upcoming election, the electorate increasingly rejects the presidential party, a behavior that is not plausibly tied to presidential coattails two years before.
This growing attraction to the out-party, we contend, is due to the electorate increasingly focusing on the vote for Congress as a way of ideologically balancing the president. Some may deny the motivation that we attribute to the voters for their movement away from the presidential party during midterm campaigns. If the shift is not motivated by ideology and policy differences between the parties, of course, critics are invited to provide alternative explanations.
We close with a note of caution. We must be aware that the data provide a strong general rule but one that can have strong exceptions. Consider the generic polls of 2002. Plugging the generic polls from that campaign into the model, the Democrats “should” probably have controlled the House following the 2002 election instead of losing ground. But despite this sobering forecasting failure, the evidence is compelling–almost always, the out-party gains strength during the election year.