March 12, 2006
Last Week's Myths About 2006
By Jay Cost
The conventional wisdom has obviously congealed around the idea that the Republican Party is headed for trouble this November. Last week alone, I encountered nearly two dozen opinion pieces making the same argument. While I agree that the Republicans will lose seats in the House – probably about nine – I was also amazed at the reasons upon which so many professional pundits based their predictions. Many columnists seem downright naïve when it comes to congressional elections, content to repeat commonplace arguments without first checking if the facts fit the theories. In instance after instance, they do not. So many pundits last week were demonstrably wrong that, for the sake of sensible and prudent thinkers everywhere, a corrective is required.
Democrats Need To Unify Around Issues To Take the House; They Are Failing Because of Leadership Problems
I have two responses to this. First, even if the Democrats could rally around a set of issues, they could not take the House. The national trends do not favor it. The economy is too strong and Bush is not sufficiently unpopular. Democrats are not capable of making Bush more unpopular, and certainly not interested in weakening the economy. Further, the number of open seats does not favor it. 95% of House incumbents are running again, and the reelection rate of incumbents in the last three cycles has averaged 99%. Democrats cannot undo what is one of the most important, yet least appreciated, secular trends in American politics: the movement toward perfect incumbent retention. Further, the tight alignment of the electorate does not favor it. 93% of Republican members of Congress are in districts Bush won. What kind of issues could the Democrats put forward that could win red districts without losing blue districts?
Second, the Democrats are structurally incapable of unifying. I find it fascinating that people on both the left and the right blast the Democrats, even going so far in some instances to make moral critiques of the party leadership. The argument seems to be that the Democrats could unite around something, but due to either a lack of vision or a lack of willpower, they have been unsuccessful.
The intuition behind this is that the “normal” state of an American political party is unification. The Democrats themselves used to be unified, so the logic goes, and their lack of unity is their principal problem and a moral failing. In actuality, nothing could be further from the truth. American political parties are generally disunited, and the Democratic Party has been particularly disunited for quite a while. FDR Democrats have been a majority party for nearly 80 years; during that time, they have been a relatively loose affiliation of individuals who are tied together sometimes by issues, sometimes by history. Prior to Roosevelt, Democrats were more united, but they were also the minority party. Today’s pundits seem to demand a kind of programmatic, comprehensive unity – a new Democratic ideology that unites the most diverse members of the party, from San Francisco’s Nancy Pelosi to Montana’s Brian Schweitzer to Massachusett’s Ted Kennedy to Arkansas’s Mark Pryor. In other words, they want to get rid of the FDR Democratic Party, return to the William Jennings Bryan Democratic Party, but still keep majority status.
These pundits are expecting something that no individual or set of individuals can possibly deliver. The current state of the Democratic Party is not something that happened because Clinton triangulated, because Dean did not win the nomination, because Pelosi beat Steny Hoyer, or because congressional Democrats are too cozy with business interests. The current state of the Democratic Party is the product of a process that began with the events of October 29, 1929 – and the political settlement that followed. To blame Harry Reid and Nancy Pelosi for not uniting the party is ridiculous. How can you blame somebody for failing to single-handedly undo the fruits of a political alignment now in its 77th year?
The big problem is not a problem with the Democratic Party, which is operating as it has always operated. It is a problem for the Democratic Party – the combination of the Republican Party, the alignment of the electorate and the nature of congressional elections. The GOP has, since the FDR realignment, been much more ideologically unified than the Democratic Party. For most of that period, they were a minority party. However, the electorate has slowly but surely shifted such that the GOP, still as united as ever, is now also a majority party. This has created problems for the Democrats – as it is hard for a disunited party to outmaneuver a united one. On top of this, the Republican breakthrough of 1994 meant that the GOP now only needs to play defense to keep Congress – and our Madisonian system is much better suited to defense than to offense.
So, Democrats – go easy on Pelosi and Reid. Being in charge of the Democratic Party is a hard job. And, while you are at it, send a thank you note to Dick Gephardt, who did a much better job at holding the House Democrats together than he is given credit for. Further, mind the moral critiques, Republicans. If you had a caucus as diverse and unruly as the Democratic one, you would not do any better.
Republican Disunity Indicates Trouble for the GOP
Today, we see the Republicans divided on issues like immigration and spending. This division seems to be a first in the post-1994 Republican Party. Unfortunately for Democratic partisans, their own history shows that internal divisions do not reduce a party to minority status.
The FDR Democratic Party has had unity problems since before the great architect himself passed away. Why do we remember fondly President Harry Truman and not President Henry Wallace? The former was a compromise vice-presidential candidate in 1944, designed to appease the conservative wing of the party. Since Roosevelt’s final election, the party has had a major split in its electoral coalition in 7 of the last 15 presidential elections. The first came when Strom Thurmond and the Dixiecrats abandoned Truman in 1948. The Democrats were severely disunited in the 1960s and 1970s – the split between the northern and southern wings of the party was the animating feature of American partisan politics during those years. It stifled Kennedy’s agenda. It utterly ruined Carter. Nixon was able to exploit it. Johnson was able to manage it. All the while – liberal success/conservative failure, conservative success/liberal failure – the Democrats held the House.
Contemporary Republican disagreements are nothing compared to what divided the Democrats during the 1960s, 70s and 80s. Even if they were, history shows that division has nothing to do with possession of power. The logic of congressional elections indicates this as well. Incumbents almost always win or lose by themselves. Their party label is second in the minds of the voters. The average midterm voter is not going to punish his member of Congress because the Republican Party does not agree on some issues. Sentiment toward the party does not even strongly affect midterm turnout – as midterm voters are the habitual ones who come to the polls every year. It is only in the on-year elections – where turnout is much greater due to the presidential contest – that a depressed party base is a real factor.
Accordingly, the Republican Party need not unite on the issue of immigration to win over the average midterm voter. It would be helpful for Bush in the marginal district here or there to increase his standing with the party’s base, but it is a bridge too far to argue that he needs to do that to keep the House. The Democrats were able to keep Congress for many decades despite being disunited on the key national issues of the day.
I would add to this that expecting your party to always be unified is like expecting the Steelers to win every Super Bowl: as much as you might yearn for it, the rules of the game are set up to make it virtually impossible. Just as our government is divided six ways from Sunday, so are the parties that manage our government. There are state party organizations, national party organizations, local party organizations, congressional parties, state legislative parties, the party base, the party elite, the party donors, party social clubs – the list goes on and on. There is nobody sitting atop this massive, diverse group of people with the capability of ordering them around. Given the incredible plurality of political interests in the nation, is it any wonder that the parties tend to internally disagree? Is it any wonder that they act similar to the way Madison, in Federalist #10, expected the government would act – interest counteracting interest such that nothing is accomplished? It seems to me that the real wonder is that the Republicans have managed to have a hand in government for the last 23 of 25 years without serious disagreements coming to the surface until now.
Current “Right Track” Poll Numbers Indicate Trouble for the GOP
The intuition behind this is that a positive right track number helps the President and his party; a negative one hurts the President and his party. The principle bit of evidence for this is 1994 – right track was heavily negative that year, and Clinton paid the price for it.
Once again, I have two responses. First, a positive right track rating does not necessarily help a president’s party, and a negative one does not necessarily hurt it. People with good memories will recall that this was one of the big arguments that the Kerry/Edwards people made in 2004. Remember May, 2004? I sure do: “Gallup has right track at -25! Bush is doomed unless that number goes positive!” Of course, it never did. By Election Day, right track was -9, and Bush still managed to increase his share of the vote. In 2002, Republicans won 49.2% of the House vote – this was an increase over their share in 2000 and their largest margin since 1994. Undergirding that was a tepid +1 right track. Right track was positive prior to the 2000 election, but that should have elevated Gore. In 1998, the Democrats failed to take the House, despite the “advantage” of right track being +25. In 1996, Gallup had right track at -26 before the election. Clinton won despite it.
Second, changes in right track/wrong track do not correspond to changes in the partisan share of the vote. In the last five cycles, an increase in the right track percentage – which should help the party of the president – has only corresponded to the “correct” result twice. The other three times, right track goes in one direction, final vote totals go in the other. In other words, this statistic has been weakly correlated with final electoral outcomes for the last decade. Random guessing would work better!
So, while it is true that right track was at this level when the GOP was swept into power in 1994, it is also true that it has been an extremely poor predictor of final vote totals. The fact that it was in sync with 1994 is not compelling, given the fact that it has been out of sync ever since. Even a broken clock is right twice a day.
Current “Congressional Job Approval” Poll Numbers Indicate Trouble for the GOP
Congressional job approval is also an extremely poor predictor of final vote totals – for exactly the same reasons. Positive job approval does not help the party controlling Congress, and vice-versa; positive changes in job approval do not help the party controlling Congress, and vice-versa.
Why are congressional job approval and right track/wrong track such poor predictors? It is because voters do not view congressional elections as a referendum on the state of the nation. Congressional elections are almost always a referendum on the incumbent. When they are not, it is usually because they are Senate elections (which can sometimes, but not always, become proxies for the national debate) or elections where there is no incumbent running.
So, accordingly, voters will tell Gallup that they think the nation is on the wrong track. They will tell AP-Ipsos that they disapprove of the way Congress is handling its job. Then, in December, they will tell the University of Michigan that they voted for their incumbent member of Congress – he or she, in their minds, is not part of the problem.
Bush’s Job Approval Will Cost Republicans the House Majority
Those who argue this are probably correct that Bush’s approval rating will be a factor. The intuition that lingers behind their assertions, however, is incorrect. Most pundits tend to assume that voters punish/reward members of Congress based upon what they think of the President. However, cross-sectional survey data indicates quite clearly that is not how voters think. Presidential job approval has an extremely complicated relationship to midterm congressional elections: individual voters do not claim that it matters, but it seems that aggregate seat changes depends in part upon it. Thus, referencing job approval minimally requires theoretical sophistication, which by and large has been lacking.
There are several major theories on congressional elections – two of which I find to be the most compelling in the post-1994 era. One factors presidential job approval, real changes in disposable income and the extent to which a party is above its historical average. This one predicts about 11 seats to switch hands. The other one factors the extent to which a party has to defend open seats, the extent to which the President increased the party’s vote total in the previous election, and real changes in disposable income. This one predicts about 8 seats to switch hands. One uses job approval, another does not. One takes into account open seats, another does not.
Both job approval and open seats seem to me to be important, and my inclination is to split the difference between them, putting the final total at 9-10. I am further inclined to drop this number a bit once again, to 8-9, because I think that the unique distribution of congressional seats (with only 17 Republicans being in districts Bush lost in 2004) probably means that Bush’s job approval is more efficient at holding congressional seats.
Job approval, then, is not the decisive factor. It is a possible factor among five. And its effect in November is measurable – and the measurement is that it will help cause about 9 net seats to change hands.
Current “Generic Congressional Ballot” Numbers Indicate Trouble for the GOP
In fairness to those who assert this, the generic congressional ballot used to be a good predictor of congressional voting outcomes – until, that is, the Republicans started winning. The generic question almost always and everywhere skews toward the Democrats. If one were to look at an archive of the 2004 polls, one would be amazed by the almost universal “blueness” of it – in comparison to the decidedly “red” outcome. Among those polls spelling doom and gloom for the GOP was an LA Times poll from Spring, 2004 that had the GOP down an embarrassing 19% to the Democrats. Final result? GOP +3%. That would be a pro-Democratic skew of 22% -- or the misgauging of more than one in five voters. Yikes – turns out it was the LA Times and not the GOP that should have been embarrassed.
If you look at the Gallup generic measure since 1994, you will see that the average poll skews toward the Democrats by an average of 6%. In 2002, the average skew was a whopping 8%. The final Gallup poll before the election has fared even worse. It has skewed toward the Democrats by 7% on average and by 10% in 2002.
Does this imply that we should correct the average generic congressional ballot poll by about 8%? Not really – once again, the correlation between the generic result and final electoral outcomes has generally been poor in the last 10 years. A party’s improvement in the generic vote has only corresponded to an improvement in its share of the vote once. We should take this as a sign that this poll is a poor reflection of voting intentions. This makes intuitive sense. American voters are not like their British counterparts on the other side of the pond. Partisanship comes second to individual candidates. In America, your average respondent will say, “Yeah, I want the Democrats!” in May, even in October. He will get to the ballot box in November, only recognize one name on the House ballot, recall that he likes that fellow, and vote for his Republican incumbent.
Last Week's Myths About 2006
By Jay Cost
The conventional wisdom has obviously congealed around the idea that the Republican Party is headed for trouble this November. Last week alone, I encountered nearly two dozen opinion pieces making the same argument. While I agree that the Republicans will lose seats in the House – probably about nine – I was also amazed at the reasons upon which so many professional pundits based their predictions. Many columnists seem downright naïve when it comes to congressional elections, content to repeat commonplace arguments without first checking if the facts fit the theories. In instance after instance, they do not. So many pundits last week were demonstrably wrong that, for the sake of sensible and prudent thinkers everywhere, a corrective is required.
Democrats Need To Unify Around Issues To Take the House; They Are Failing Because of Leadership Problems
I have two responses to this. First, even if the Democrats could rally around a set of issues, they could not take the House. The national trends do not favor it. The economy is too strong and Bush is not sufficiently unpopular. Democrats are not capable of making Bush more unpopular, and certainly not interested in weakening the economy. Further, the number of open seats does not favor it. 95% of House incumbents are running again, and the reelection rate of incumbents in the last three cycles has averaged 99%. Democrats cannot undo what is one of the most important, yet least appreciated, secular trends in American politics: the movement toward perfect incumbent retention. Further, the tight alignment of the electorate does not favor it. 93% of Republican members of Congress are in districts Bush won. What kind of issues could the Democrats put forward that could win red districts without losing blue districts?
Second, the Democrats are structurally incapable of unifying. I find it fascinating that people on both the left and the right blast the Democrats, even going so far in some instances to make moral critiques of the party leadership. The argument seems to be that the Democrats could unite around something, but due to either a lack of vision or a lack of willpower, they have been unsuccessful.
The intuition behind this is that the “normal” state of an American political party is unification. The Democrats themselves used to be unified, so the logic goes, and their lack of unity is their principal problem and a moral failing. In actuality, nothing could be further from the truth. American political parties are generally disunited, and the Democratic Party has been particularly disunited for quite a while. FDR Democrats have been a majority party for nearly 80 years; during that time, they have been a relatively loose affiliation of individuals who are tied together sometimes by issues, sometimes by history. Prior to Roosevelt, Democrats were more united, but they were also the minority party. Today’s pundits seem to demand a kind of programmatic, comprehensive unity – a new Democratic ideology that unites the most diverse members of the party, from San Francisco’s Nancy Pelosi to Montana’s Brian Schweitzer to Massachusett’s Ted Kennedy to Arkansas’s Mark Pryor. In other words, they want to get rid of the FDR Democratic Party, return to the William Jennings Bryan Democratic Party, but still keep majority status.
These pundits are expecting something that no individual or set of individuals can possibly deliver. The current state of the Democratic Party is not something that happened because Clinton triangulated, because Dean did not win the nomination, because Pelosi beat Steny Hoyer, or because congressional Democrats are too cozy with business interests. The current state of the Democratic Party is the product of a process that began with the events of October 29, 1929 – and the political settlement that followed. To blame Harry Reid and Nancy Pelosi for not uniting the party is ridiculous. How can you blame somebody for failing to single-handedly undo the fruits of a political alignment now in its 77th year?
The big problem is not a problem with the Democratic Party, which is operating as it has always operated. It is a problem for the Democratic Party – the combination of the Republican Party, the alignment of the electorate and the nature of congressional elections. The GOP has, since the FDR realignment, been much more ideologically unified than the Democratic Party. For most of that period, they were a minority party. However, the electorate has slowly but surely shifted such that the GOP, still as united as ever, is now also a majority party. This has created problems for the Democrats – as it is hard for a disunited party to outmaneuver a united one. On top of this, the Republican breakthrough of 1994 meant that the GOP now only needs to play defense to keep Congress – and our Madisonian system is much better suited to defense than to offense.
So, Democrats – go easy on Pelosi and Reid. Being in charge of the Democratic Party is a hard job. And, while you are at it, send a thank you note to Dick Gephardt, who did a much better job at holding the House Democrats together than he is given credit for. Further, mind the moral critiques, Republicans. If you had a caucus as diverse and unruly as the Democratic one, you would not do any better.
Republican Disunity Indicates Trouble for the GOP
Today, we see the Republicans divided on issues like immigration and spending. This division seems to be a first in the post-1994 Republican Party. Unfortunately for Democratic partisans, their own history shows that internal divisions do not reduce a party to minority status.
The FDR Democratic Party has had unity problems since before the great architect himself passed away. Why do we remember fondly President Harry Truman and not President Henry Wallace? The former was a compromise vice-presidential candidate in 1944, designed to appease the conservative wing of the party. Since Roosevelt’s final election, the party has had a major split in its electoral coalition in 7 of the last 15 presidential elections. The first came when Strom Thurmond and the Dixiecrats abandoned Truman in 1948. The Democrats were severely disunited in the 1960s and 1970s – the split between the northern and southern wings of the party was the animating feature of American partisan politics during those years. It stifled Kennedy’s agenda. It utterly ruined Carter. Nixon was able to exploit it. Johnson was able to manage it. All the while – liberal success/conservative failure, conservative success/liberal failure – the Democrats held the House.
Contemporary Republican disagreements are nothing compared to what divided the Democrats during the 1960s, 70s and 80s. Even if they were, history shows that division has nothing to do with possession of power. The logic of congressional elections indicates this as well. Incumbents almost always win or lose by themselves. Their party label is second in the minds of the voters. The average midterm voter is not going to punish his member of Congress because the Republican Party does not agree on some issues. Sentiment toward the party does not even strongly affect midterm turnout – as midterm voters are the habitual ones who come to the polls every year. It is only in the on-year elections – where turnout is much greater due to the presidential contest – that a depressed party base is a real factor.
Accordingly, the Republican Party need not unite on the issue of immigration to win over the average midterm voter. It would be helpful for Bush in the marginal district here or there to increase his standing with the party’s base, but it is a bridge too far to argue that he needs to do that to keep the House. The Democrats were able to keep Congress for many decades despite being disunited on the key national issues of the day.
I would add to this that expecting your party to always be unified is like expecting the Steelers to win every Super Bowl: as much as you might yearn for it, the rules of the game are set up to make it virtually impossible. Just as our government is divided six ways from Sunday, so are the parties that manage our government. There are state party organizations, national party organizations, local party organizations, congressional parties, state legislative parties, the party base, the party elite, the party donors, party social clubs – the list goes on and on. There is nobody sitting atop this massive, diverse group of people with the capability of ordering them around. Given the incredible plurality of political interests in the nation, is it any wonder that the parties tend to internally disagree? Is it any wonder that they act similar to the way Madison, in Federalist #10, expected the government would act – interest counteracting interest such that nothing is accomplished? It seems to me that the real wonder is that the Republicans have managed to have a hand in government for the last 23 of 25 years without serious disagreements coming to the surface until now.
Current “Right Track” Poll Numbers Indicate Trouble for the GOP
The intuition behind this is that a positive right track number helps the President and his party; a negative one hurts the President and his party. The principle bit of evidence for this is 1994 – right track was heavily negative that year, and Clinton paid the price for it.
Once again, I have two responses. First, a positive right track rating does not necessarily help a president’s party, and a negative one does not necessarily hurt it. People with good memories will recall that this was one of the big arguments that the Kerry/Edwards people made in 2004. Remember May, 2004? I sure do: “Gallup has right track at -25! Bush is doomed unless that number goes positive!” Of course, it never did. By Election Day, right track was -9, and Bush still managed to increase his share of the vote. In 2002, Republicans won 49.2% of the House vote – this was an increase over their share in 2000 and their largest margin since 1994. Undergirding that was a tepid +1 right track. Right track was positive prior to the 2000 election, but that should have elevated Gore. In 1998, the Democrats failed to take the House, despite the “advantage” of right track being +25. In 1996, Gallup had right track at -26 before the election. Clinton won despite it.
Second, changes in right track/wrong track do not correspond to changes in the partisan share of the vote. In the last five cycles, an increase in the right track percentage – which should help the party of the president – has only corresponded to the “correct” result twice. The other three times, right track goes in one direction, final vote totals go in the other. In other words, this statistic has been weakly correlated with final electoral outcomes for the last decade. Random guessing would work better!
So, while it is true that right track was at this level when the GOP was swept into power in 1994, it is also true that it has been an extremely poor predictor of final vote totals. The fact that it was in sync with 1994 is not compelling, given the fact that it has been out of sync ever since. Even a broken clock is right twice a day.
Current “Congressional Job Approval” Poll Numbers Indicate Trouble for the GOP
Congressional job approval is also an extremely poor predictor of final vote totals – for exactly the same reasons. Positive job approval does not help the party controlling Congress, and vice-versa; positive changes in job approval do not help the party controlling Congress, and vice-versa.
Why are congressional job approval and right track/wrong track such poor predictors? It is because voters do not view congressional elections as a referendum on the state of the nation. Congressional elections are almost always a referendum on the incumbent. When they are not, it is usually because they are Senate elections (which can sometimes, but not always, become proxies for the national debate) or elections where there is no incumbent running.
So, accordingly, voters will tell Gallup that they think the nation is on the wrong track. They will tell AP-Ipsos that they disapprove of the way Congress is handling its job. Then, in December, they will tell the University of Michigan that they voted for their incumbent member of Congress – he or she, in their minds, is not part of the problem.
Bush’s Job Approval Will Cost Republicans the House Majority
Those who argue this are probably correct that Bush’s approval rating will be a factor. The intuition that lingers behind their assertions, however, is incorrect. Most pundits tend to assume that voters punish/reward members of Congress based upon what they think of the President. However, cross-sectional survey data indicates quite clearly that is not how voters think. Presidential job approval has an extremely complicated relationship to midterm congressional elections: individual voters do not claim that it matters, but it seems that aggregate seat changes depends in part upon it. Thus, referencing job approval minimally requires theoretical sophistication, which by and large has been lacking.
There are several major theories on congressional elections – two of which I find to be the most compelling in the post-1994 era. One factors presidential job approval, real changes in disposable income and the extent to which a party is above its historical average. This one predicts about 11 seats to switch hands. The other one factors the extent to which a party has to defend open seats, the extent to which the President increased the party’s vote total in the previous election, and real changes in disposable income. This one predicts about 8 seats to switch hands. One uses job approval, another does not. One takes into account open seats, another does not.
Both job approval and open seats seem to me to be important, and my inclination is to split the difference between them, putting the final total at 9-10. I am further inclined to drop this number a bit once again, to 8-9, because I think that the unique distribution of congressional seats (with only 17 Republicans being in districts Bush lost in 2004) probably means that Bush’s job approval is more efficient at holding congressional seats.
Job approval, then, is not the decisive factor. It is a possible factor among five. And its effect in November is measurable – and the measurement is that it will help cause about 9 net seats to change hands.
Current “Generic Congressional Ballot” Numbers Indicate Trouble for the GOP
In fairness to those who assert this, the generic congressional ballot used to be a good predictor of congressional voting outcomes – until, that is, the Republicans started winning. The generic question almost always and everywhere skews toward the Democrats. If one were to look at an archive of the 2004 polls, one would be amazed by the almost universal “blueness” of it – in comparison to the decidedly “red” outcome. Among those polls spelling doom and gloom for the GOP was an LA Times poll from Spring, 2004 that had the GOP down an embarrassing 19% to the Democrats. Final result? GOP +3%. That would be a pro-Democratic skew of 22% -- or the misgauging of more than one in five voters. Yikes – turns out it was the LA Times and not the GOP that should have been embarrassed.
If you look at the Gallup generic measure since 1994, you will see that the average poll skews toward the Democrats by an average of 6%. In 2002, the average skew was a whopping 8%. The final Gallup poll before the election has fared even worse. It has skewed toward the Democrats by 7% on average and by 10% in 2002.
Does this imply that we should correct the average generic congressional ballot poll by about 8%? Not really – once again, the correlation between the generic result and final electoral outcomes has generally been poor in the last 10 years. A party’s improvement in the generic vote has only corresponded to an improvement in its share of the vote once. We should take this as a sign that this poll is a poor reflection of voting intentions. This makes intuitive sense. American voters are not like their British counterparts on the other side of the pond. Partisanship comes second to individual candidates. In America, your average respondent will say, “Yeah, I want the Democrats!” in May, even in October. He will get to the ballot box in November, only recognize one name on the House ballot, recall that he likes that fellow, and vote for his Republican incumbent.