Monthly Archives: November 2009

Tipping Points, Health Care and Tuesday’s Results

I want to follow up on my earlier post regarding Tuesday’s election, because evidently there is still confusion regarding what actually explains the recurring pattern in which the president’s party loses the off year gubernatorial elections, and the impact those losses are likely to have this year on pending legislation, particularly health care. It turns out that even those who recognize the historical pattern behind Tuesday’s results are still drawing the wrong lessons. Consider the following from Taegen Goddard, who writes for the Political Wire “In both states, it seems pretty clear that the voters tend to show their disappointment with the new President by voting for the other party, no matter which party controls the White House. Most likely many people had some expectations from the newly (re)elected President, didn’t see them satisfied and wanted to send the incumbent a message. The correlation (12 out of 12 and 14 out of 16  [he’s referring to combined results in Virginia and New Jersey governors’ races dating back to the 1980’s]) is too strong for just chance. This year’s results should be interpreted in this light. While the results are not encouraging for President Obama, they are hardly surprising.”

But let’s be clear here: that’s not what I believe happened.  Voters in the two governors’ races did NOT, in the main, express their disappointment with Obama. The pattern primarily recurs because turnout dips in the off year election, not because people invariably become disappointed with whoever is president. In fact, this pattern happens even when presidents are very popular nationally. Keep in mind that turnout dropped by almost a million voters in both states compared to the presidential election year. In short, the results primarily reflect a smaller voting pool that – because the president is not on the ticket – tends to include fewer voters who supported the president’s party in the presidential election.

Now, what impact will these results have on pending legislation?  Consider the following two observations from writers for the Wall St. Journal and the Washington Post.  Kim Strassel writes in the WSJ: “So long as the president looked strong, those Blue Dogs and freshmen and swing-state senators would stick. Show them any sign of weakness, however, and rattled Dems would begin to care more about their own re-elections than they did their president. Tuesday, the White House hit that tipping point.”

In the Post, Perry Bacon reports that all 177 House Republicans plan to vote against the majority party’s health care bill.  Bacon writes: “The universal opposition to the health-care bill, which Congress could vote on as soon as Saturday, was long expected, but Republican wins in the Virginia and New Jersey gubernatorial races on Tuesday further emboldened the GOP in its stance.”

Tipping point?  Further emboldened?  Is Bacon suggesting that had the Democrats won the two governor’s races, some Republicans would have considered voting FOR the Democratic health care bill?  Is it really the case, as Strassel suggests, that Tuesday’s results suddenly reminded House Democrats to put their own reelection ahead of the president’s political interests? Remember back in January, when Obama was still being hailed as an agent of change, and with his popularity in the mid-60% range, he still couldn’t wring a single Republican House vote in support of his stimulus bill! (And we often forget that 11 House Democrats opposed the initial House stimulus bill and seven voted against the final conference bill.  Of the 11 House Democrats voting against the stimulus bill, 9 represent districts in which McCain beat Obama in 2008.)

My point is that the Republicans have already demonstrated a willingness to oppose legislation that they viewed as against their party principles. And blue dog Democrats have been battling the Democratic Party leadership long before Tuesday in an effort to moderate the fiscal impact of health care reform. The 40+ House Democrats representing districts that voted for McCain in 2008 didn’t need Tuesday’s results to understand what was at stake for them electorally.  (Note as well that some thirty-plus Republicans represent districts that went for Obama in 2008, and they too need to calibrate their votes accordingly.)

What we are seeing, I think, is another illustration of the structural bias in media coverage that tries to fit Tuesday’s results into a larger storyline.  But the evidence to support the claim that Tuesday’s results emboldened Republicans to oppose Obama and decreased support from moderate Democrats is, in my view, weak.   The political dynamics these articles cite were already in play.

With key votes coming up on health care, I’m going to try in my next post to see if we can make some sense of the politics of this legislation.

Meanwhile, here’s the data I promised on the New Jersey governors races dating back to the first election after the 1947 constitutional reform in that state which changed the governor’s term from three years to four.  You can see that the off-year loss pattern is a relatively more recent phenomenon.  Since 1949, it has held true in 10 of 16 gubernatorial races, including the last six in a row.  Frankly, I’m not sure why the pattern did not appear earlier, but my guess is that local issues in some of these years swamped the independent impact of the turnout factor.  But I’m open to anyone’s suggestions who knows something about New Jersey politics… .

Governor Election Party President’s Party Fits Pattern?
Alfred E. Driscoll 1949 Republican Democrat YES
Robert B. Meyner 1953 Democrat Republican YES
Meyner 1957 Democrat Republican YES
Richard J. Hughes 1961 Democrat Democrat NO
Hughes 1965 Democrat Democrat NO
William T. Cahill 1969 Republican Republican NO
Brendan Byrne 1973 Democrat Republican YES
Byrne 1977 Democrat Democrat NO
Thomas Kean 1981 Republican Republican NO
Kean 1985 Republican Republican NO
James Florio 1989 Democrat Republican YES
Christine Whitman 1993 Republican Democrat YES
Whitman 1997 – Resigned in 2001 Republican Democrat YES
Donald T. DiFrancesco 2001 (Appointed) Republican
James McGreevey 2001-04 Resigned Democrat Republican YES
Richard J. Codey 2004 (Appointed) Democrat Republican
Jon Corzine 2005 Democrat Republican YES
Chris Christie 2009 Republican Democrat YES

What Really Explains Tuesday’s Elections Results?

What explains Tuesday’s election results?  I had suggested in my posts leading up to the election that the outcomes would not be a referendum on the Obama presidency, but instead would be driven primarily by combination of local factors interacting with a historical pattern of low turnout in the off-year elections that tends to disadvantage the president’s party.  Journalists, in contrast – although often acknowledging the importance of these structural and local factors – nonetheless found it difficult to resist interpreting the results within a single dominant narrative that focused on Obama.  Although not universally accepted, this was the overriding theme of most of the post-election coverage that I read.

Let’s look at the evidence. Most significantly, we see that the outcome in both the New Jersey and Virginia governors’ races follows the historical pattern dating back to at least the 1970’s that I traced out in my earlier posts: the candidate of the president’s party – whether a Democrat or a Republican  – lost the election.  That makes six times in a row in New Jersey, and 7 out of the last 10 elections there that this has happened. In Virginia, the pattern is even stronger: in the last NINE governor’s races dating back to 1977, the candidate of the president’s party has lost (see chart below).

Why does this happen?  I said it is partly because turnout tends to be down in off-year elections, particularly among the president’s party.  Did that happen again this year?  Yes it did.  Let’s look first at New Jersey, where the Democrat incumbent Jon Corzine lost to Republican challenger Chris Christie.  In the 2008 presidential race, Democrats turned out in greater numbers than Republicans in New Jersey, by 44-28%, (with 28% independents). Last Tuesday the Democratic dominance in a historically “blue” state receded by 6%, to 41-31%, with no change in the overall proportion of independents. In total, 2.3 million people voted on Tuesday, down from 2008’s presidential election total of almost 3.9 million. So, again, we see the impact of structural factors in the form of lower voter turnout particularly among the president’s party coming into play.

But how do I know that, despite the low voter turnout, this wasn’t a referendum on Obama?  In New Jersey there are two pieces of evidence: first, according to exit polls, 57% of those who voted approved of the job Obama is doing as president.  Nonetheless, 19% of those who signaled their approval for the president still voted for the Republican, Christie. Second, 47% of those who voted thought corruption and property taxes were the two major issues driving their vote, and they went predominantly for Christie by about 67-25% in both cases.  Another 32% thought it was the economy and jobs, and here Corzine won, albeit by a smaller margin, 58-36%.  We see then, that local issues tended to dominate the thinking of Christie’s supporters more than Corzine’s; they weren’t voting against Obama or his policies.

We see the same story in Virginia. Here, fortunately, the exit polls asked voters for whom they had voted in the 2008 presidential election. Remember, Obama carried Virginia in 2008 over McCain by 52.6-46.3%, and the exit polls put the proportion of Democratic versus Republican voters then at 39-33%. But on Tuesday, it was McCain supporters who turned out in force, 51-43%, over Obama supporters, and the party proportions were almost reversed, with Republicans outnumbering Democratic voters 37-33%.   Turnout, moreover, was down from 2008; then more than 3.7 million people voted while on Tuesday turnout totaled a bit less than 2 million.  We see, then, much as I predicted, that structural factors linked to the historical pattern of lower turnout by the president’s supporters played a huge role in McDonnell’s victory, just as they did in New Jersey.  And McDonnell did best among those voters (47%) who thought the dominant issue was jobs and the economy and those (15%) who believed it was taxes – both issues that have a strong local dimension.

Here, to give you some historical sense of the underlying pattern that partly explains these results, are the outcomes of the Virginia governor’s races dating back to 1977. Pay particular attention to columns 3 and 4 which list the governor’s and president’s party affiliations.

Year Winner Party President’s Party %Vote
1977 John Dalton Republican Democrat (Carter) 55.9%
1981 Charles Robb Democrat Republican (Reagan) 53.5%
1985 Gerald Baliles Democrat Republican (Reagan) 55.2%
1989 Doug Wilder Democrat Republican (Bush I) 50.2%
1993 George Allen Republican Democrat (Clinton) 58.3%
1997 Jim Gilmore Republican Democrat (Clinton) 55.8%
2001 Mark Warner Democrat Republican (Bush II) 52.2%
2005 Tim Kaine Democrat Republican (Bush II) 51.7%
2009 Bob McDonnell Republican Democrat (Obama) 58.7%

Isn’t this cool?  Don’t you love political science?

I will discuss the New York congressional outcome in a separate post, but by way of foreshadowing my comments, I believe that result also turned on factors not directly related to Obama or his policies.

A final point: Republicans, and not a few journalists, are suggesting that Tuesday’s results will make it more difficult for Obama and congressional Democrats to pass his major legislative initiatives, particularly health care, in part because “blue dog” moderate Democrats will think twice about the implications of supporting potentially costly government initiatives.  I think this is wrong for two related reasons. First, Tuesday’s results turned primarily (although not exclusively) on local factors, against the backdrop of low voter turnout.  More importantly, however, moderate Democrats already have concerns about supporting these major legislative initiatives – two governors’ races did not provide any additional information in this regard. In other words, the independent impact of the Democrats’ losses in these two races on major policy debates in Congress is likely to be nil because it does nothing to change the calculations of legislators in other states who must decide whether to support these bills. Admittedly, it’s hard to see this in the blizzard of media stories regarding the alleged “momentum” and “signals” conveyed by these races.  By now, however, you should understand that this prevailing narrative fits the media’s structural needs to create an entertaining, newsworthy storyline.

The real story, I submit – and one that I find most exciting – is how well these outcomes seem to fit within the long-term historical pattern of election outcomes in off-year elections.  It suggests that we (political scientists) understand some of the underlying dynamics of these races – dynamics that tend to get lost in the punditry-driven hype surrounding the latest elections results.

It may be less newsworthy, but for a political scientist, it’s pretty exciting!

p.s. Note also my response to Olivier’s query in the comments section to my previous post.  Thanks also to my New Jersey and New York students who filled me in on some of the local issues in both states (as noted above, I’ll deal with New York separately).

Republicans Sweep – Obama to Resign!

Actually, they haven’t quite swept – Owens may take New York’s 23rd congressional district.  But, never mind – here are the headlines so far:

From MSNBC: “Election ’09 tests do not bode well for Obama” .

CNN asks, “Referendum on Obama?”  (Answer – It might as well be).

The New York Times puts it this way: “Republicans Sweep Top Races in Setback For Obama”.

The Washington Post warns: “Contests serve as warning to Democrats: It’s not 2008 anymore”  (To Dan Balz’ credit, he suggests that the results are not a referendum on Obama.)

Predictable, yes.  Accurate, no.

I’ll be on tomorrow (Wednesday) with a more in-depth analysis of the results, but for now, sit back and enjoy the show.  Pay particular interest to how Democrats have decided to spin the turn of events in NY’s 23rd district – that it shows how Republicans are eating their own by forcing moderates out of the party.  Compare that to their reaction to when Joe Lieberman lost in the Democratic primary in his last Senate race – then it was a blow for party purity!  Republicans, meanwhile, are jumping for joy not just because of the victories, but for the margin of victory and the turnout figures.


At least they could have waited for the election results

The off-year elections are still a day away but that hasn’t prevented pundits from already doing exactly what I predicted they would do after the election: interpreting the (anticipated) results as a referendum on Obama’s policies and as a bellwether for the future of American politics.  Steve Lombardi, at, suggests in a story today that, “many may be under-interpreting the meaning of a GOP sweep tomorrow. Yes, a lot can and will happen between now and Election Day 2010, but make no mistake: Republicans are likely to sweep all three races tomorrow and that does say something about the direction of the country and voter perceptions of the economy.” That’s right. We are under-interpreting the results.  Lombardo goes on to write, “Obama is personally popular but voters remain unsure of the effectiveness of his policies. That is why his personal popularity does not necessarily translate into help for either Corzine or Deeds. Poll after poll shows that the President is well-liked but voters are not yet convinced that his policies are moving the country in the right direction.”

I don’t doubt that Republicans may sweep all three races, and that a large part of their victories will be due to voter dissatisfaction with the economy.  But, for the reasons I laid out in my last post, I don’t agree with Lombardi’s implicit assumption that the elections will turn on voters’ views of Obama, or his policies.  Rather than repeat those points, I want to extend them here by responding to a couple of your emails regarding my claim that these results are more likely to be driven by structural factors, particularly the ebb in voter turnout characteristic of off-year elections.  You asked:

  1. If the disproportionate decline in turnout from the president’s supporters largely explains the tendency for his party to lose the off-year election, shouldn’t we also see that trend in New Jersey?

In fact, we do – I probably should have discussed this in my last post.  The same pattern has held, by my quick count, in the last five NJ governor’s races, and in 6 of the last 9.  I haven’t gone back before the 1973 race, which is when I started my Virginia count.

  1. If the structural explanation holds true, shouldn’t we see this type of pattern – the president’s party losing the governorship in Virginia and New Jersey – before the 1970s’ as well?

Not in Virginia – remember that until the 1960’s, Virginia was part of the solid Democratic south – no Republican could expect to hold state wide office.  If I get a chance, I’ll try to look up the data for New Jersey prior to 1974.

Interestingly, the House special election in New York’s 23rd congressional district provides something of a test for my thesis that local fundamentals, and not views toward Obama, will drive the results.  Obama won this largely Republican district in 2008, 52%-47%, and as of this past Oct. 31, he was still very popular there. According to a Siena poll released on that date,  59% of those surveyed had a favorable view of Obama, with only 37% viewing him unfavorable.  And yet Republican Doug Hoffman was up 5% over his Democratic rival. How can this be?  A PPP poll released today suggests the answer – they polled likely voters and found that “While Barack Obama won a narrow victory in NY-23 last year, those planning to vote in this race supported John McCain by a 51-43 margin. Obama’s approval rating with likely voters is just 39%.”  While the Siena poll is also of likely voters, they start with a random digit dial survey, and then pare the respondents down.  PPP, in contrast, surveys only people who are on the registered voter list, so are more likely to be surveying those who will actually vote.  If I’m reading these two polls correctly, then, the PPP poll is more likely to be capturing those likely to vote tomorrow – and among that subset of voters, Obama is less popular.   This doesn’t mean his lack of popularity will cost Owens the race – it means that the voting pool for tomorrow’s special election will consist of a higher proportion of McCain voters who are predisposed to vote Republican (or Conservative in this case).

If Hoffman wins, then, it will be in part because of the drop in turnout among those who voted for the president the year before – the same factors that have historically hurt the president’s party in the off-year governors’ races in Virginia and New Jersey.  In contrast to what pundits would have you believe, one cannot blame tomorrow’s results on voter backlash to Obama or his policies, and one certainly can’t use them to forecast the 2010 midterm results.

I’ll try to post a quick response to the results tomorrow night.   Meanwhile, when the chorus begins warming up with the bellwether song, remember my warning… .

Will Tuesday be a referendum on Obama?

In my last couple of posts I explored the degree to which the national news media can be considered biased. I argued that if bias exists, it is not likely a function of journalists’ own political ideology, which admittedly tends for most to lean left, but instead reflects a “structural” bias rooted in the economics of the news business, particularly the need to maintain an audience and to generate stories within a rapidly accelerating and increasingly competitive news cycle. Today I want to explore an example of that structural bias in action, by focusing on the coverage of three elections: governors’ races in New Jersey and Virginia, and the special congressional election in New York’s 23rd district, all to be held this coming Tuesday.  The dominant news narrative suggests that the outcome of these elections ought to be interpreted as a referendum on the Obama administration.  Consider David Broder’s take, which captures the prevailing narrative quite well: “The first key votes of the Obama era take place this week, not on the floor of the House or Senate, where health-care legislation still languishes, but in Virginia, New Jersey and northern New York state, where President Obama’s endorsements of threatened Democratic candidates will test his political clout a year after his own election.”  Or Michael Barone’s here: “In other words, the 2009 contests are a reasonably fair test of the strength and durability of the Democratic majority that Obama and his ticket-mates assembled in 2008, a majority that was only made possible by gains in hitherto Republican territory.”

This narrative serves journalists well, because it allows them to tie three otherwise disparate events into a unified story that provides an easy–to-assess gauge of Obama’s political clout.  Moreover, these races, according to this narrative, may provide early signs regarding the likely outcome of the 2010 midterm elections, as this story suggests:

“If Republicans seize the governors’ mansions in both states, the embarrassment will be acute. That is just what happened in both New Jersey and Virginia back in 1993 before the Republicans seized control of the US Congress the following year, dealing a crippling blow to the newly minted Democratic president of the time, Bill Clinton.”  So, according to the prevailing narrative,  the results are significant on several levels – as both an indicator of Obama’s clout and an early signal regarding the 2010 midterm elections.  Obama’s supporters, like E.J. Dionne, fearing Republican victories in all three races, are already trying to craft an explanation that fits within this broader narrative – namely, that Democrats who lose did so because they didn’t tie themselves closely enough to Obama (see story here).

On the other hand, the White House is already distancing itself from Tuesday’s anticipated results for fear that, in the event of a potential Republican sweep, the prevailing media narrative will be that Obama’s political clout has weakened. At the same time, however, Obama is making last minute visits to both Virginia and New Jersey in an effort to prevent that occurrence.

There is only one problem with this dominant narrative: it is almost certainly wrong.  Rather than a referendum on Obama’s political clout, these three races will be largely decided by factors that have nothing to do with Obama or his policies – although they will, in the case of the two governor’s races, have something to do with the presidency.

Consider the upstate New York congressional race first. In New York, the 23rd congressional district has been represented by a Republican dating back at least to 1993 (as far back as I’ve looked) and no Democrat has received as much as 40% of the popular vote in that time. Even in 2006, the year of the Democratic surge in the House, incumbent Republican John McHugh crushed his Democratic opponent in the 23rd district by more than 30%. Polls suggest that the current Democrat candidate, Bill Owens, is not going to top the 40% mark either, and that his only hope for victory is that Republican voters split their votes between conservative candidate Doug Hoffman and Republican Dede Scozzafavo, who in the last day pulled out of the race precisely to prevent that from happening.  Polls suggest there has been a last minute surge to Hoffman, as Republicans coalesce behind his candidacy.  In short, as those of you live across Lake Champlain can attest, this is Republican territory and has been for far longer than Obama has been president.  He can’t be blamed for Owens’s defeat. Similarly, if Owens were to win, it would have little to do with Obama and everything to do with the split vote among Republicans.   Indeed, Hoffman’s advertising, which I get to see from across the Lake, has been trying to tie Owens to Nancy Pelosi – not to Obama!  Moreover, should Owens win he would almost certainly be defeated in two years by a Republican candidate.

What about the two governors’ races – can’t they be reasonably viewed as a referendum on Obama and his policies?  Only tangentially, and not for the reasons suggested by the dominant news narrative. Both elections will likely turn on whether the depth of the anti-incumbent sentiment fueled by job loss and the economic downturn can offset demographic trends that have favored Democrats in recent years.  In New Jersey, the incumbent Democrat Jon Corzine has lost popularity after a divisive budget battle with the state legislature against the backdrop of the economic slump and charges of corruption and a crumbling road system. But what about Virginia?  Here there is no incumbent in the race, since the current Governor Tim Kaine, a Democrat, is prevented by state term limits from running again. That leaves an open race between Democrat Creigh Deeds and Republican Bob McDonnell – surely this can be viewed as a referendum on Obama, particularly in a state that is trending Democratic and which he won in the 2008 election?

I think not. To understand my reasoning, note the following pattern: since 1977, the party that won the White House the year before has lost the Virginia governor’s race every time, regardless of the president’s party. (Virginia began its off year elections in 1857, New Jersey began its in 1847.) A year ago, Obama won the presidency, so if the pattern holds, the Republican candidate, McDonnell, ought to win election as Virginia’s next governor.  And, if polls are an accurate barometer, he is poised to win handily. If so, this would mean this pattern has held for nine straight gubernatorial elections. In political science, that’s as close to a “law” explaining a political phenomenon that one is going to find!  But what’s the basis of this pattern?  It primarily reflects structural dynamics derived from the different levels of voter turnout in presidential versus gubernatorial election years. Take a look at the following chart (source here):

Note that in presidential years, turnout is much higher than during the off-year elections. The decline in turnout in the governor’s race, I argue, occurs disproportionately among those voters who supported the winning presidential candidate during the presidential election.  It follows, then, that the opposing party candidate benefits from the lower turnout in the governor’s race the following year.  And that, I believe, is why McDonnell is likely to win on Tuesday: the decline in turnout from the previous year is likely to be greater among Democrats than Republicans.

If McDonnell wins in Virginia, then, it ought not to be seen as an indication that Obama’s political clout has lessened (or that Deeds failed to use Obama effectively). To believe that we need to believe that every president has suffered a similar loss in clout dating back to Carter in 1976.  A simpler explanation for the pattern is that turnout invariably favors the candidate of the out party – regardless of who the president might be.  This may not make for a compelling news story – it tells us nothing about Obama’s “clout”, or about what’s likely to happen in 2010.  But for a political scientist, it’s the story that matters.  (As further evidence that the Virginia and New Jersey results don’t tell us much about midterm results, note that both states elected Democrats in 2001 – and Republicans went on to gain seats in the 2002 congressional midterm elections)

Come Tuesday, then, when the news reports begin writing the final paragraph to the prevailing narrative that sees these elections as a referendum on Obama, you will know better.  It’s all about the fundamentals.

Addendum (7:40 pm): the New York Times is reporting that, in a surprise twist, Republican candidate  Dede Scozzafava is endorsing her Democratic opponent Bill Owens in New York’s 23d district race.   I have absolutely no idea what impact this endorsement will have on the race, but it does provide further evidence supporting Republican claims that Scozzafava’s views were out of step with Republican voters in the district.