Monthly Archives: October 2010

Rest In Peace Ted Sorensen

Ted Sorensen died today, a week after suffering a stroke. He was 82. Sorensen served for many years as John Kennedy’s chief staff assistant, beginning as a research aide to the newly elected Senator in 1953 and culminating with his role in Kennedy’s White House as Special Counsel.  Sorensen’s obituary as reported in various media outlets will undoubtedly cite his work drafting some of Kennedy’s most memorable speeches, including JFK’s celebrated 1961 inaugural address (“we shall pay any price, bear any burden, meet any hardship, support any friend, oppose any foe to assure the survival and the success of liberty”), as well as his assistance (how much assistance remains a matter of some controversy) in the writing of Kennedy’s Pulitzer Prize-winning Profiles in Courage.

But Sorensen was much more than Kennedy’s wordsmith. As JFK’s Special Counsel (a title originally given to FDR’s Samuel Rosenman, who created the position Sorensen inherited) Sorensen sat astride a key decision, or action forcing, process in Kennedy’s White House.  In contrast to the modern White House, in which aides are organized according to specialized functions (economic or national security adviser, speechwriter, liaison to interest groups, chief of staff, etc.)  Kennedy’s White House was modeled after Franklin Roosevelt’s. That meant aides were generalists who were organized not by functional specialty, but by the daily decisions that flowed into the Oval Office. Sorensen sat astride the most important of these “action forcing” processes: the drafting of all public documents by which JFK developed and “sold” his legislative program and related policies.  In addition to crafting speeches, this meant drafting legislation and related messages to Congress, issuing statements on enrolled legislation, and reviewing executive orders, to cite only the most important duties entrusted to Sorensen.  Because Kennedy – as did FDR – preferred to manage his own staff, there was no chief of staff in his White House.  This meant that Sorensen reported directly to JFK, and that his role as Special Counsel largely mimicked JFK’s perspective as president.  Sorensen’s duties stand in distinct contrast to those of modern White House aides, whose narrower jurisdiction means that they are unable to fully appreciate the President’s more holistic and broader perspective on decisionmaking. What the modern White House staff disaggregates due to functional specialization, the chief of staff must reassemble for the President’s understanding.  Kennedy, in contrast, dealt with a few senior aides, including Sorensen, whose duties more nearly aligned with his job as President; there was no chief of staff to manage the White House on JFK’s behalf.  The FDR-Kennedy model began to break down during Johnson’s presidency, when Sorensen’s successor Joe Califano shed the speechwriting role, and it was completely lost to history when Richard Nixon became president and installed the prototype, including a chief of staff, for what has become the modern White House staff.  In later years Sorensen would be highly critical of the proliferation of White House positions, particularly within the speech writing staff.

By his own admission, a part of Sorensen died with Kennedy’s assassination in November, 1963. For a decade, his life had been defined by his service to Kennedy – almost every waking moment was devoted to serving this one man. In a particularly poignant section of his memoirs, Counselor: A Life at the Edge of History (which I strongly recommend), Sorensen describes how helpless he felt because he was not there – as he had been for 10 years – to answer the questions – What will happen to my country? Who will take care of Jackie and my children? – that he imagined the President asked himself after the first assassin’s bullet hit him, but before the second (or third) fatal shot occurred.  Sorensen, looking back at JFK’s assassination after almost half a century, writes, “I do not know whether I have ever fully recovered from John. F. Kennedy’s death. Time passed. Love and laughter helped. But the deep sadness of that time remained, only to be reinforced five years later by the murder of his brother Robert. Those two senseless tragedies robbed me of my future.”  Although Sorensen briefly flirted with a return to public service – running unsuccessfully for the Senate in New York, and agreeing to head the CIA under Carter only to see his nomination blocked – he spent his remaining years working primarily in a New York-based law firm while protecting the Kennedy legacy, beginning with his memoirs of the Kennedy years in his book Kennedy.

I met Sorensen late in his life, when a stroke had robbed him of most of his vision. Although he was hard at work on his own memoirs, he graciously agreed to write a short piece for a book I was co-editing honoring the life of Richard Neustadt.  Sorensen and Neustadt had worked closely during Kennedy’s presidency, and Neustadt had been poised to join the White House staff when Kennedy was assassinated.  In my brief dealings with Sorensen, he was extremely cordial, and demonstrated no sense of entitlement or superiority that I have sometimes detected in other former White House aides.

When I heard of Sorensen’s death, I thought of the sacrifice that those who work for presidents often make.  In his memoirs, he admits that his first marriage fell apart largely because his life was primarily devoted to serving John F. Kennedy.  In the late summer, 1963, shortly before Kennedy’s assassination, his wife – who had already been separated from Sorensen for three years – moved back with their three sons from Washington to her previous home in Wisconsin.  In one particularly moving remembrance, Sorensen describes how he tried desperately to carve time out of his busy schedule to throw a baseball with one of his sons on the Washington mall.  After Kennedy’s death, with his family gone, Sorensen eventually decided to dedicate his life to keeping Kennedy’s legacy and ideals alive.

Rest in Peace, Mr. Sorensen.  I hope you found some comfort in your later years.

My House Forecast: Meet John Boehner, The Next House Speaker

That’s how I see it. The Republicans will regain control of the House on Tuesday.  If you want to know the reasons why in detail, I’ll be giving the last of my election talks today at 4 at the Ripton Community Church (the white church building just past and across the street from the Vermont Country store with the “IEC” sign – you know, that frozen water stuff).

But since many of you will be watching the bridge opening instead, and the less fortunate among you don’t have the pleasure of living here in God’s Green Mountains, let me give you the abbreviated explanation for why John Boehner is poised to be the next Speaker of the House.

As most of you know, historically the president’s party loses House seats during the first midterm after a presidential election.  The average loss across the seven first-term midterm elections in the period 1946-2006 in the House has been 17 seats (and slightly more than 1 seat in the Senate.)  The only exception to this rule since 1946 occurred in 2002, when the Republicans actually gained House seats. Most analysts believe this gain was largely due to the ability of George Bush and Republican candidates to nationalize the election by running on a national security platform in the aftermath of 9-11- an exceptional event.(Note that across all 16 midterms in the post-World War II period, the president’s party lost, on average, 24 House seats and 3.5 Senate seats.)

There are several reasons for this recurring midterm seat loss (I explore these factors more deeply in my talk), but the key point is that the loss is based on structural (or fundamental) factors that affect all first-term midterm elections to some degree, regardless of the president.  And it is these structural factors that political scientists attempt to measure when they construct their forecast models.  In an earlier post I discussed some of the House midterm predictions made by political scientists during Labor Day.  John Sides at the Monkey Cage has constructed a very useful chart summarizing some of these predictions, and adds in one by Nate Silver at 538.com as well (note that Silver is an economist so you probably want to take his prediction with the requisite dose of salt.)

There are a couple of things to keep in mind here. First, these forecasts are based on factors as of Labor Day, or earlier.  Several of the forecasters note that their estimates might change in the remaining two months prior to the election. (Silver seems to change his almost weekly, which is why I don’t put much stock in his underlying model).  Second, these red dots only show the mean estimate – not the confidence interval (a measure of the reliability of the estimate.) So, for example, although Jacobson’s model forecasts a Republican pickup of 43 seats, his confidence interval is +/- 27 seats – that is, the “true” seat pickup lies, with a specified probability, anywhere from 16 to 70 seats.  So, although most political scientists put the likelihood of a Republican takeover at 60-70%, this estimate is often subject to a great deal of uncertainty.

How are these models constructed? Most derive their forecast based on structural factors.  These usually include some measure of the economy (unemployment, change in disposable income, etc.); a measure of the number of seats in play, or the time in which the majority party has held power; and – in some cases – a measure of presidential popularity.

If most models measure the same set of factors, why are there such differences in the mean estimates (within the various confidence intervals) shown in this table?  Some of this has to do with differences in the models’ variables – e.g., do you use unemployment or disposable income to assess the economy?  But perhaps the biggest difference is whether the model incorporates a fourth factor – the generic ballot measure that asks voters whether they will vote Republican or Democrat in the midterm.  As I’ve discussed in several previous posts, that measure has been favoring the Republicans at historically high levels for much of this election season. The latest Gallup low-turnout model has Republicans up by 14% over Democrats.

The Real Clear Politics equivalent, which is a composite of several of these polls, has Republicans up by about 7% among likely voters.  Accordingly, the forecast models that incorporate the generic ballot (keeping in mind these were constructed more than a month ago when the generic ballot may have been different)  are predicting Republican seats gain in the 40+ seat range.  Those that do not – see the Cuzan and Lewis-Beck and Tien models – believe that Republicans will pick up seats closer to the historical average, and thus Democrats will retain control of the House.

So, where do I come down?  Note that I am a consumer of these forecast models, not a producer.  That is, although I will run some of these models on my own in order to see the consequences of tweaking some of the assumptions, I do not create my own Dickinson model.  Instead, my forecast starts with existing models, which I then modify based on what I am seeing in my own poke and soak-based methodology (surveying hundreds of polls, talking to my Mom, the local barber, what my former students working in campaigns across the country tell me, etc.) – a methodology I’ve honed through the years of election watching, and which led me to predict that Democrats would pick up 18 seats in 2006. They picked up 30.   So be forewarned!

My “methodology” leads me to believe, as I announced in my last talk, that the political science models are, for the most part, too conservative.  That is, most of them are underestimating the likely Republican seat pickup. In political scientists’ defense, there’s a good reason for this.  Political scientists (myself included) are inherently conservative; we believe that what happens in the future will not usually differ from what occurred in the past. That is, the dynamics that drove midterm outcomes in the period 1946-2006 aren’t likely to change much in 2010.  The forecast models are predicated on this assumption.  When these models get into trouble, then, is when they don’t account for new factors, or when existing factors operate in unusual ways.  In 1994, for instance, I can’t recall a single forecast model that predicted the 57-seat pickup by Republicans.  It was an unprecedented (in the post-World War II era) gain and it broke a 40-year period of Democrat control of the House.

My point is that you had to be an idiot, or a genius, to have predicted the Republican takeover in 1994. Most political scientists are neither.   Instead, they are smart people who are very good at understanding why regular patterns occur and in extrapolating from those regular patterns to forecast future outcomes.

So, what will happen Tuesday?  There are three factors that I think suggest this will be another “wave” election, closer to 1994 than to the other post-war midterms.  First, there are an extraordinary number of seats in play – Charlie Cook puts it at more than 90 – and most of them are held by Democrats.  In a typical year you might see half that number of seats in play.  So models that take the number of marginal seats in play into account (the Campbell model is one) are going to be more accurate, I think.  Second, I think the Tea Party movement is an indicator – not a cause, mind you! – that this election cycle is different.  As I’ve said repeatedly, most Americans are not members of the Tea Party, although perhaps a quarter to a third are sympathetic to it.   But the movement has, I think, tapped into a deep underlying fear widespread among many Americans that the nation is headed in the wrong direction.  Whether one agrees with that diagnosis or not, the fact remains that, according to most polls, independents are swinging Republican in large numbers compared to 2006 and 2008, and that Democrats are less enthusiastic than Republicans in this race.   In this regard, one of my bellwether election indicators remains Delaware.  The latest poll there (yes, it’s only one poll with all the methodological caveats you’ve heard from me  before) shows Christine (“I’m not a witch!/What Separation of Church and State?”) O’Donnell shaving 9 points off Chris Coons’ lead in the last week – and this was before the Halloween “trick and trick” story broke.  I’m not saying O’Donnell can win – she’s still down by double digits – but the fact that a Tea Party candidate with her credentials is even in hailing distance of the Democrat in this traditionally blue state speaks volumes in this election cycle.  There are other indicators along these lines – the latest Fox generic ballot, for example, has Republicans up big over Democrats, and polling data shows Angle solidifying her lead in Nevada and Rossi closing on Murray in Washington – that suggest the election might be breaking in Republicans’ favor.  Admittedly, this is a selective read of a very broad array of recent indicators, some of which favor Democrats (see West Virginia and California, for example). Underlying all this, however, is my belief that the financial crisis that hit in 2008, and which persists today in the guise of  an unemployment rate over 9%, a sluggish housing market, and general economic uncertainty, is one of those unprecedented factors with which the traditional forecast models don’t deal well.  In terms of previous election forecasts, these economic conditions are outliers, which leads me to believe that the models are not as useful in predicting the likely outcome in this cycle and that, contrary to some  models’ forecasts,  Republicans will pick up an unusually high number of House seats.

How many?  I’ll set the over/under at a 49-seat Republican gain.  Note that this may be extreme by political science standards, but it is very conservative when compared to what political pundits are predicting. RealClearPolitics, for example, has a mean estimate that has Republicans netting 65 seats.  I’ve heard some estimates as high as 75-80 seats. So in some respects I’m splitting the difference between the mean political science forecasts and the collective “conventional wisdom” espoused by political pundits.

So there you have it.  It’s time for you to weigh in again and test your electoral wisdom against mine.  As always, no wagering at home – all that’s at stake is an “It’s the Fundamentals, Stupid” t-shirt and my personal reputation. Lots of you participated in the Senate contest (hope everyone remembers to check the comments sections to see the Senate predictions.)  If you voted in the Senate contest, you can still participate in the House lottery as well.  Note also that the Senate forecast contest is still open – both remain open through Monday night.  Two t-shirts are better than one!

Don’t tarry – Election Day is around the corner.  And I’ll see some of you at 4 p.m. in the Ripton (nondenominational) Church today.

The Senate Forecast: Place Your Bets, Please

It’s time (actually past time!) for the midterm forecast/”It’s the fundamentals, stupid” t-shirt contest. As always, please, no money wagering at home. This should only be done by professionals.  All that’s at stake here is a t-shirt and, of course, my professional pride – oh, and control of the legislative agenda for the next two years.

Let me begin with the Senate. I should note that most political scientist forecast models focus on the House, not the Senate. (For example, among the five midterm forecast models presented in the recent issue of Political Science and Politics, only one includes a Senate forecast.)  In previous years the Senate forecasts have not always been very accurate.  I’m not sure why this is the case, although I can suggest explanations. Senate races are usually more competitive – they attract stronger challengers, who are often well funded (think Carly Fiorina or Linda McMahon). State populations – especially in large states – are more heterogeneous, making it difficult to assess how the population is reacting in the aggregate to the fundamentals.  There are fewer Senate races in any cycle, so perhaps it’s easier simply to assess each race individually, a la Charlie Cook or Stuart Rothenberg.

For whatever reason, the science of forecasting Senate races is not as well developed.  As a result, unlike the presidential forecast I’ve done in previous years (or the House one I will present later) this Senate forecast is not based on any specific model; there’s no underlying theory on which the projections are based.  It’s all seats-of-the-pants, race-by-race assessments which, frankly, you can find on a dozen other websites.  Nonetheless, we have a tradition to keep at the Presidential Power blog, so here goes.  Just remember – if my predictions go horribly wrong, it’s not an indictment of political science forecast models – it’s simply a reflection of my own lack of political insight.

Drum roll please!

I’ve set the over/under for the Republican gain in the Senate next Tuesday at 8 seats.  (Yes, for those of you who attended my talk last night, I’ve changed the prediction again. This is the real one. For now.)   Note that this is almost twice the Republican gain than what Alan Abramowitz’s Senate forecast model indicates based on the current generic ballot results, although Alan is careful to note the large margin of uncertainty surrounding his Labor Day Senate prediction.

Because there are only about a dozen competitive Senate seats (of 37) during this cycle, I can list the predicted Republican pickups, in rough order of likelihood (from most likely to I’m basically guessing).  They are:

1. North Dakota.  Democrat Byron Dorgan stepping down. Republican John Hoeven to beat Tracy Potter.

2. Arkansas.  Incumbent Democrat Blanche Lincoln to lose to John Boozman.

3. Indiana. Democrat Evan Bayh stepping down. Republican Dan Coats to beat Brad Ellsworth.

4. Wisconsin. Incumbent Democrat Russ Feingold to lose to Ron Johnson.

5. Pennsylvania. Incumbent Arlen Specter lost in the primary. Republican Pat Toomey to beat Joe Sestak.

6. Nevada.  Incumbent Democrat Harry Reid to lose to Sharron Angle.

7. Illinois. Democrat (and Blagojevich appointee) Burris stepping down. Republican Mark Kirk to beat Alexi Giannoulias.

8.  Colorado: Incumbent Michael Bennet to lose to Republican Ken Buck.

For what it’s worth, in the other closely contested Senate races, I think West Virginia stays Democrat with Manchin beating Raese, Murray retains her seat in Washington, Boxer holds on in California, Rubio keeps Florida’s Republican seat, Blumenthal holds Connecticut for the Democrats, Ayotte keeps New Hampshire’s seat Red, and Miller beats Murkowski in Alaska (with no party change in either case).

A word of caution with these Senate projections.  For races numbered 5-8 (and for that matter the Washington state and West Virginia races), the uncertainty surrounding the outcomes make my estimates not much better than what you would get with a coin flip.  The races are that close.  In the end, I went Republican on most of them because I think the fundamentals are heavily in Republicans’ favor this cycle.  It goes without saying (so why do I say it?) that much depends on who turns out, and in what numbers. Here the Democrat fundraising advantage may help at the margins although frankly I think that advantage – which is dwindling – will be worth more in the House races.

Ok, there you have it. If I’m right, the Democrats will retain control of the Senate, 51-49, although the working margin will be smaller than that on some issues. Let the betting begin.  As always, the person who comes closest to predicting the actual outcome wins this t-shirt.  In cases of ties, everyone wins or I’ll go with the first person who submits an entry.

Tomorrow, schedule permitting,  I’ll present the House predictions.  By way of preview, I think the situation is much worse for Democrats in this chamber.

Is It Tea Time In America?

For local readers, I’ll be at 51 Main tomorrow (Tuesday) night beginning at 7 p.m.  to give my latest version of my midterm election talk.  Although those who have attended previous talks dating back to August know my prediction regarding which party will control the House and Senate, I’m going to set the over-under seat projections for both chambers, and open up the voting to you.  As always, an “It’s the Fundamentals,  Stupid” t-shirt will be on the  line.

One issue I hope to address is the likely impact of the Tea Party movement on the midterm voting. Long-time readers know that for some time I have been arguing that this is a genuine conservative grass roots phenomenon that has the chance to significantly influence the outcome of the midterm elections. This view, to say the least, was not initially shared by most pundits, particularly those on the Left. They tended to disparage the Tea Party as an overhyped media-generated phenomenon, whose members were mostly racist kooks occupying the extreme rightwing fringe of American politics.

Thus, in April, 2009, Paul Krugman dismissed the movement as anything but a “spontaneous outpouring of public sentiment. They’re AstroTurf (fake grass roots) events, manufactured by the usual suspects” – that is, wealthy conservatives.

Similarly, E.J. Dionne, in this most recent article characterizes the Tea Party as a group of latter-day John Birchers that encompass perhaps the most extreme 10-15% of the electorate – a claim he has made before. Echoing Krugman, he also argued in an earlier column that the Tea Party was overhyped by the media and was fueled, at least in part, by racial animosity and bankrolled by billionaire conservatives.  Dionne concluded that it would have little impact on the midterm elections. Frank Rich, meanwhile, warned of the dangerous consequences (armed militias anyone?) should Republicans actually encourage this movement.  .

By now I hope I’ve persuaded you that these critics were evaluating the movement with their partisan hearts, not their minds. While they sought to dismiss the movement, the polling data and the electoral results to date paint a different picture, one that in my view supports my initial contention from a year ago that the movement would be an influential political force during this election cycle.

My view finds additional support in a just published Washington Post survey of the Tea Party movement. The highlights are as follows:

  1. The movement is a genuine grassroots phenomenon, built from the bottom up and not, as Krugman and Dionne argue, from the top down. Forty-two percent of the 600-plus chapters the Post contacted say they have no connection with any national organization. Thirty-two percent say they work mainly with the Tea Party Patriots, a national umbrella organization.  Less than 3% have contacts with the Republican Party. Almost 90% say their activities and strategies are entirely or predominantly local. In short, the movement is decentralized with chapters  spread across the nation and very small in size (more than half had less than 50 people at their last meeting). They meet in living rooms and coffee shops, not boardrooms.
  2. The movement lacks funding, and what money they have raised is almost entirely due to grass roots activity. On average, chapters raised less than $1,000 this year, and 95% of that came from individuals at the local level.
  3. Members lack political experience (over 80% said most of their members are new to politics).  Nor do they have any central leadership.  Indeed, they can’t agree on any single person who serves as the face of the Tea Party movement. (Palin comes the closest, with about 14% of those surveyed indicating that she best represents the movement.)
  4. The movement is conservative, but not – at least in professed allegiance – partisan. Members express no interest in forming a third party.  At the same time they are almost equally opposed to the Republican Party establishment as they are to the Democratic leadership. Less than 1% say their most important goal is electing Republicans to office.  Similarly, less than 1% cite countering Obama/the Democrats as the most important issue. Instead they would prefer to throw out incumbents from both parties.  Indeed, this has been perhaps their biggest impact during the current electoral cycle – in several high profile races they have succeeded in replacing the Republican establishment candidate with one backed by the Tea Party. Nonetheless, given a choice between a Republican or a Democrat in the general election, I think in most instances most Tea Partiers will support the Republican candidate.  Of those who say their chapter is involved in campaigning, 42% say they are working on behalf of Republicans, 32% say either party, and 22% say neither of the two major parties. But more than 80% of those campaigning for Republicans say they only campaign for Republicans who share their issues.
  5. The movement is predominantly focused on a single issue: reversing the trend toward “big government” as embodied in the size of the federal budget and the growth in the deficit.  Some 44% of respondents list some version of this issue as their primary concern. The next highest issue is protecting the Constitution, cited by only 11%. Ninety-nine percent say the economy is the most important factor explaining Tea Party support so far.
  6. The movement cares almost nothing about cultural issues.  One percent or less cite immigration, 10th amendment/states rights, gay marriage, abortion, or gun rights as the most important issue.

The Post story, combined with other polling data I have presented in several discussions on this topic, reinforce the basic picture I’ve painted before: the Tea Party movement is not a media-created, “astro-turf” movement – it is instead a grass-roots development rooted in the response to the government’s handling of the fiscal meltdown, the passage of the stimulus bill and health care, and a growing budget deficit, against the backdrop of  economic recession. But we shouldn’t over state the strength of the movement either.  It does not attract anything close to majority support within the electorate. Based on previous polling data, I estimate that only about 3% of voters have attended a Tea Party event. Perhaps 15%-25% of voters consider themselves aligned with, or supporters of, or approving the Tea Party movement (the number varies depending on question wording.) This compares, for example, to about 35% of adults who call themselves Democrats (combined strong or weak) and about 26% who say they are Republicans (based on the 2008 National Election data).  Of course, there is overlap between Tea Party and  partisan affiliation.

However, the potential impact of the movement belies its minority status.  First, by becoming actively involved in the nominating process, the Tea Party has already influenced the list of candidates from which voters must choose in the general election. And contrary to conventional wisdom, there is data indicating that although the Tea Party may consist of political amateurs, the candidates running under the Tea Party label are not as inexperienced as the media coverage suggests. Moreover, in low-turnout elections like the midterms, Tea Partiers are likely to be a larger proportion of the electorate than their minority status might suggest. Krugman et al can continue to whistle in the graveyard in the hope that the Tea Party movement is a media-created illusion.  But the facts to this point suggest otherwise.

Is it Tea Time in America?   Or is this simply a Tempest in a Tea Pot?  We’ll know in a week.

How Democrats Can Minimize Their Losses on November 2

How might Democrats minimize their losses on November 2?  Not by following the advice rendered by Sheri and Allan Rivlin, the authors of this Huffington Post article.

The Rivlins, echoing arguments that I find on many progressive websites, suggest that “something seems to be going terribly wrong with the Democratic message machine. To say we are off message is charitable, when a more accurate reading is that we are trying every message at once, which is another way of saying we are not committed to any message at all.”  To remedy this, they argue that Democrats should nationalize this election by framing it as a choice between the two parties’ economic plans.  In the Rivlins’ view, “The only thing voters should care about is who has the best plan to move forward, and our advantage is we have a plan and the Republicans don’t. The best evidence that we have a plan is all the statements by the other side that we have been doing too much. The best evidence that the Republicans don’t is their inability to point to any economic ideas that are new and different from Bush’s ideas.”

In my view, this is precisely the wrong approach for Democrats.  Fairly or not, the public – or at least that portion that is likely to vote in 10 days – perceives the Democrats as the party in charge at a time of stubbornly high 9% unemployment and sluggish economic growth.  Those facts are not going to change between now and November 2.  Simplistic or not, many voters view this election as either continuing on the current slow-growth path or trying something – anything – new.  Put another way, a good portion of the electorate – particularly the independents – that votes on Nov. 2 will frame this election as a referendum on the party in power – not a choice between competing economic plans. The Democrats aren’t going to lose the House because of bad messaging – they are going to lose it because of a bad economy, one that has not improved on their watch. As this recent survey indicates, most voters do not believe the economy is improving. Instead, “Americans are waiting for changes in the national and local employment statistics to signal that the downturn’s end is near.”  In the meantime they are increasing savings and cutting down on expenditures.

With this dismal economic context, what should Democrats do?  Incumbents – particularly those in the House – should do precisely the opposite of what the Rivlins suggest. That is, they should not do what Obama did in his recent radio address and try to frame this election as a choice between going forward under Democrat leadership or going back to the failed policies of Bush and the Republicans. A September Democrat poll conducted by Democratic pollsters Stan Greenberg and James Carville supports my claim. It found that “Voters are not moved by Democratic messages that say ‘go forward, not back,’ mention President Bush, compare then and now, or even that hint the economy is “showing signs of progress.” No matter how dramatically these messages set out the record of Republican obstructionism, their work for the wealthiest and Goldman Sachs, the millions of jobs lost and Democrats’ support for jobs, small business and new industries – these messages falter before the Republican attack. The messages get lower scores and lose voters. After hearing this battle of Republican and Democratic messages, 8 percent shift their vote to support the Republican, while only 5 percent move to the Democrats. We lose ground. These messages are helping the Republicans.”

In short, Democratic candidates for Congress lose support when they try to frame this election as a choice between going back to the Bush-Republican policies or continuing with Obama’s.  Rather than nationalize this election,   during the next ten days Democrat incumbents should distance themselves from partisan policy debates and instead work assiduously to localize this election by reminding voters of everything they’ve done for their district: earmarks, casework – all the constituency services that contribute to the traditional advantages that accrue to incumbents in both parties. At the same time, they should distance themselves from the Democratic Party leadership, including President Obama. His most useful role is as a fundraiser in strongly blue districts and states.  But if you are a House Democrat running in a marginal district, you do not want the President at your side right now.

There is some evidence that Democratic leaders are beginning to grasp this. Two days ago Nancy Pelosi, appearing on Charlie Rose , suggested that this was an election that would turn on local issues: “This is not a national election; it is district by district,” she said.  Obama’s press secretary Robert Gibbs, appearing on Meet the Press, agreed with Pelosi’s assertion by arguing that the midterm election would turn on local factors, not national ones.

Localizing House elections  at a time when midterms have become more nationalized  than at any time in the last half century is not going to be easy.  But at this point House Democrats don’t have much choice.