Is the Sky Really Falling? Understanding Obama’s Decline in Approval Ratings

As heralded by a spate of recent media stories, Obama’s approval ratings, as gauged by numerous polls released in the last couple of weeks, have exhibited an almost steady decline since his inauguration. Here is the trend line, based on calculations used at (Pollster approval here):

As you can see, Obama has dropped from a high of near 70% approval to about 53%, on average, today. Predictably, the major news outlets have interpreted this drop as cause for alarm. The following analysis, from David Kuhn at RealClearPolitics (see article here) is not atypical of this recent coverage:

“In other words, Obama’s got problems. Health care reform has come up against the rocks and the cop and prof race debacle has also likely taken some toll. Then there is the unemployment rate, almost assured to soon reach double digits…In the end, approval rating also concerns symbolism. Once a president falls below 50 percent he can no longer say he has the majority of the public behind him–a knockdown blow for a president’s legislative ambitions.”

Almost all the assumptions built into this statement – that the drop in Obama’s approval reflects public dissatisfaction with Obama’s policies, and that this drop will hurt his legislative ambitions – are wrong, or at least misleading.  In fact, as long time readers of this blog will remember, I predicted at the start of his presidency – long before the fight over health care – that Obama’s poll numbers would be in the mid-50’s range by this point, precisely where they are now.  How did I know this?  It helps to understand what drives poll numbers this early in a presidency. Simply stated, the drop in Obama’s numbers has almost nothing to do with healthcare, or unemployment figures, and everything to do with the underlying distribution of partisan sentiment in this country.  Remember, Obama was elected with about 53% of the popular vote – that represents his “natural” level of support; these are the people who are predisposed to support his policy choices on almost all issues.  Now, this level can fluctuate a few percentage points in response to specific events – so-called rally events, or their antitheses – but these tend to be short-term fluctuations that dissipate as the news coverage cycles to different events.  On the whole, under “normal” politics, we would expect Obama’s popularity to hover in the mid-50%, precisely where it is now.

However, this does NOT mean that a president’s approval ratings are destined to remain centered on the president’s popular vote percentage throughout his presidency.  Over the duration of a presidential term, poll numbers can undergo a sustained shift, but only in response to relatively constant news narrative, either positive or negative.  Typically, these more enduring shifts are linked to broad-gauge measures of the economy (inflation or unemployment numbers) or sustained military engagements.  This means that if the economy remains mired in a recession for the next 18 months, or there is a steady drumbeat of negative news emanating from Afghanistan, Obama’s numbers will look positively Bush II-like by the end of his presidency.  But we are a long way from there, as yet.

To get some perspective on the issue, consider what happened to previous presidents’ popularity across a similar time period. Here’s table listing the drop (or gain) in presidents’ approval ratings from their inauguration through August 1. (For comparison purposes, all data comes from Gallup poll ratings, using the closest poll to August 1 as the end date).

President Starting Approval (first poll after inauguration) Approval August 1 (or closest poll to date) Percent Popular Vote in Election Change from starting approval Change from popular vote
Obama 68% 56% 52.9% -12 +3.1
Bush II 57% 55% 47.9% -2 +7.1
Clinton 58% 43% 43% -15 0
Bush I 51% 60% 53.4% +9 +6.2
Reagan 51% 59% 50.7% +8 +8.3
Carter 66% 59% 50.1% -7 +8.9
Ford Took office in August with 70% approval
Nixon 59% 62% 43.4% +3 +18.6
Johnson Took office in November with 77% approval
Kennedy 72% 74% 49.7% +2 +24.3
Eisenhower 68% 73% 57.4% +5 +15.6

There are a number of ways to spin this data, of course. Obama’s critics harp on the steep decline, which historically has only been matched by Clinton’s first term drop.  Obama’s supporters, on the other hand, point to the relatively high approval ratings.  What I want you to see, however, is how, among presidents beginning with Carter, all tend to hover a few points above their “natural” support level seven months into their presidency, as indicated in the last column.  It reflects a natural equilibrium based on the initial electoral support combined with the additional support from the presidential honeymoon. Obama is on the low end of this score, but given the uncertainty of the polling data, he is certainly not an outlier.  Note that the pre-Carter presidents tend to operate under a different polling dynamic.  I’ll devote a different post to this, but I bet you can guess what happened post-Nixon to change presidential polling patterns.  Any takers?

The other important consideration is that approval ratings – contrary to what the media coverage suggests – are simply not that important, at least when it comes to passing legislation.  I hope to devote a more complete post on this topic, but the short story is that political scientists have found very little evidence that popularity is a significant predictor of legislative success in Congress, once one controls for other factors.  So Obama’s falling approval ratings are unlikely to have much impact on whether health care, or cap and trade legislation, gets through Congress.  The media likes to report approval numbers, because it is “hard” news, and because most of the major media outlets have their own polling affiliations with which to generate these stories.  But the raw polling numbers are simply not very good measures of presidential influence in Congress.

So, if Obama’s “natural” support hovers in the mid-50’s, why did he take office with sky-high ratings hovering near 70%?  That reflects a natural tendency among Americans to initially view a newly-elected president not through a partisan lens, but from the nonpartisan perspective of “president as national sovereign”.  Simply put, in this honeymoon period, the public’s initial reaction to the approval question tends to reflect their view of the presidency as an institution, rather than the politics of the man occupying that office.  In Obama’s case, that honeymoon was heightened, I think, by the public’s recognition that the election of a black man represented a milestone in the nation’s history.  Inevitably, however, this honeymoon fades, and the natural partisan distribution reemerges.  (And not every president takes office with a substantial honeymoon – I’ll devote another post to examining this issue.) This means that no matter what policy goals Obama pursued, and with what effectiveness, he was likely to lose popularity as long as those issues were viewed differently by Republicans and Democrats. And, on most major issues that reach the Oval Office, there are clear partisan differences; issues that are deeply popular with both Republicans and Democrats (Mom, apple pie, the Red Sox) typically don’t become the subject of presidential action. It’s the divisive ones with which presidents typically grapple.

We see, then, that the media narrative, with its “sky is falling” preoccupation with Obama’s poll numbers, has the story exactly backwards.  The real issue here is how utterly unremarkable Obama’s presidency has been when it comes to public opinion, further evidence that his election was anything but transformative.  Obama’s polling dynamics to date almost exactly mimic his recent predecessors’ approval ratings.  That is why a student of the presidency could have predicted these results back in January.

In fact, one of them did.


  1. So I have a quick clarification question about general poll numbers showing a decline of support (which is to be expected as time goes on), and poll numbers that indicate the growing concern over specific policy areas, and obviously here the health care debate is the primary example. Obama’s approval ratings have fallen; and then there are separate indications on Obama’s health plan. Given the length of this debate, and fractious partisan nature of the issue, can we conclude that the growing unease is to be expected, and is also insignificant as a factor in passing of the legislation? Or because poll numbers here might be a signal to individual representatives in their districts, are they more significant to pay attention to? In essence, I suppose it’s a popularity poll’s significance versus a more policy oriented one – who wins?

  2. Hi Matt,

    Really interesting to know there’s more going on than the Presidential honeymoon.

    I have the same question as Tarsi. And beyond that, is there any data on whether Presidential popularity affects the first Congressional elections? Or is it strictly the nation’s circumstances that drive that vote.

    On a slightly related matter, does rhetoric drive approval in any respect?

    What triggers this question is gradually coalescing complaints about truth-telling by the administration, which must be a perennial of the TV age, but usually has substantive considerations underlying it. In polling, is popular “trust” a different axis than “approval”/”confidence”? If so, a leading or lagging indicator?

    In this case, a couple of chinks in the armor highlighting the trend, namely 1) Below 50% Republican credence that Obama’s Honolulu birth certificate is the full story; 2) press about the economic team’s Goldman-colored glasses; and 3) the President’s notable dodge of the CBO’s tough refutation of the cost savings coming out of Congressional health care deals; 4) skirting Congress’ national security investigative agenda; 5) heat over taking down the ethics page on the White House site for a while.

    Ruth Marcus’ op/ed in the Washington Post this morning is another example, focused on Gibbs as Press Secretary, especially some implausible tax rhetoric.



  3. Tarsi and Marty,

    What I’m trying to get at here is that aggregate polling numbers aren’t really very useful indicators of how much “capital” a president possesses in any particular policy fight. For that, we need more focused questions centering on the policy under consideration. Even here, however, polling question wording is crucial for driving results. Thus, it is quite possible that a majority of people polled with support health care reform, but oppose health care reform that means their premiums/taxes will go up. That’s why polling on health care appears so volatile – much of the apparent inconsistency is the result of different question wording. But Obama’s declining aggregate approval isn’t particularly relevant to this issue.

    Marty raises some really interesting questions relating to the link between trust and approval. I might add to that questions regarding “right track/wrong track” as well – all tap into slightly different underlying dimensions that people might use when evaluating the Obama presidency to date. This is a fascinating topic that really deserves a separate post. As for the subsidiary “chinks in the amor” Marty lists – I’m not sure they are of equal significance when it comes to overall public evaluations of Obama’s presidency. My guess is the CBO numbers play a bigger role in evaluations of his health care policy (more accurately, Congressional Democrats’ health care), while the other issues get more to considerations of Obama as trustworthy leader. Without more evidence, I hesitate to roll them together as components of the overall decline in his approval. But I need to noodle on this a bit more. Great observations though…

  4. Oops – I meant “armor”, of course!

    As for presidential popularity and congressional results – there is data on that. I need to dig it up so I don’t get it wrong! But it also deserves a separate post. As a tease, note that as congressional elections are increasingly nationalized (and I have data showing this), they become more influenced by the same factors that drive presidential approval rates. What this means – and I need to get data to show this – is that presidential approval ratings should be a stronger predictor of midterm results in recent years. Even if this is true, however, there is still a strong local component in most congressional races. I’ll see if I can get actual data to show this.

Leave a Reply

Your email address will not be published. Required fields are marked *