Thursday, May 16, 2013

Presidential Approval Probably Isn't Stuck At Midrange

Ezra Klein poses a hypothesis I've heard a few times before at the end of a nice post earlier this week:
Tucked inside Abramowitz’s take is an interesting structural prediction about American politics: As party polarization increases and the persuadable middle dwindles, scandals and gaffes will become less meaningful as they’ll only be able to convince those who want to be convinced. It’s not clear that the IRS mess, Benghazi, or the DoJ/AP issue will rise to the level of being a useful test of this thesis. But something will.
It's tempting to believe that polarization is producing presidential approval results which are little more than reflections of partisanship. Barack Obama's Gallup approval history would seem to suggest that might be happening. His best rating of 69% ranks 10th best of the 12 presidents in the Gallup era, and is really in a tie for last with Nixon (67%), Reagan (68%), and perhaps Ford (71%). Everyone else had higher highs. And yet, at least to date, Obama's lowest point of 38% approval beats everyone but Ike and Kennedy. Obama's overall average to date is 49%; since the beginning of 2012, his weekly Gallup number has ranged higher than 55% and lower than 45% only once in each direction, and overall, outside of a brief honeymoon in 2009, he's been in that range almost his entire presidency.

And yet...there's no reason to believe that polarization has advanced significantly in the last fifteen years, but George W. Bush's approval ratings were all over the place. Bush holds the highest spike in the history of Gallup, hitting a soaring 90%. And when he crashed, Bush bottomed out at 25%, lower than everyone but Truman and Nixon. Bush's first term was, overall, fourth-highest (and also fourth farthest from 50%); his second term is tied for second-lowest (and also tied for second farthest from 50%).

Conclusion? Most likely, Obama's approval ratings have stayed in a narrow range mainly because there really haven't been that many events, and those that have happened have roughly balanced out. The economy has mostly been the same from 2010 on -- recovering, but not recovering enough to get people thinking "prosperity." The same, really, with foreign affairs; with the exception of the bin Laden killing, the administration can't really offer peace, but also isn't running up the level of policy disaster that occurred during the Bush presidency...good news (out of Iraq! overall casualties in Afghanistan way down!) seem to be balanced by bad (Benghazi -- the real event, not the phony scandal; the Boston bombing; various high-profile attacks in Afghanistan).

Granted, it could be that it's harder for events now to move the needle, but the reaction to events during the Bush administration just makes me very skeptical of that.


  1. This doesnt have anything to do with approval ratings per say, but as just a casual observer of politics it looks like the Bush presidency was the catalyst for increasing partisanship over the last 15 years. I'm interested in what your basing your statement "there's no reason to believe that polarization has advanced significantly in the last fifteen years" on, that runs counter to some of the changes I've seen in the media with the rise of conservative aligned press not to mention the increase in filibusters and decrease in Senate confirmations under Obama... And now we even have Mark Sanford getting re-elected. I would say the American people look to have internalized the "vote for the party not the man" attitude to a greater extent now than they had in the early 00s.

  2. This comment has been removed by the author.

    1. I had the same reaction as you. I think what we are getting at here is partly a matter of definition and partly a matter of analysis. The phenomena that many people put down to polarization JB would, I think, say are properly understood as a result of party dysfunction, particularly the dysfunction of the GOP. Therefore although GOP dysfunction has gotten worse over the last 15 years, the electorate is no more polarized.

      I am just a casual observer of all this, but it seems that this is one of those things political scientists argue about pretty intensely. I think Abramowitz, whose email Ezra Klein was referencing, does indeed see increasing and problematic polarization in the electorate (and the general population, which is an even bigger issue).

    2. There's definitely SOME polarization in the electorate: Abramowitz has been making that point pretty well recently.

      However, its relative. The electorate is more polarized than it was. Relative to our elected officials? Not even close.

  3. This is weird. I'm naturally inclined to side with Abramowitz & Klein on this one. But, then I went and overdid what I should have for a blog post comment. (I really don't want to grade these papers)

    I regressed approval on party ID (7 point scale) for every year in the NES data (1972-2008, with no 2006--thanks NSF!). I then saved the regression coefficients, beta weights, and r-squareds. I then regressed THOSE on the year. (Technically, I used the absolute value of the Bs and beta weights, so that its not being yanked around by the party of the president)

    Nothing to speak of.
    Bs: -0.00068
    Betas: 0.006546218
    R2s: 0.004226891

    There's one year that REALLY supports Abramowitz & Klein, 2004: R2 =.46, Beta .678.
    But then we drop right back to an R2 of .022 in 2008.

    (Yes, I applied OLS to approval, a big no-no, but it was easy)

  4. My gut reaction without looking at the data ... Democratic/liberal voters are more likely to at least occasionally express approval of Republican presidents than Republican/conservative voters of Democratic presidents because Republican/conservative voters have self-polarized over the last 20 years or so. Thus, Democratic presidents have a lower approval rate "ceiling" than Republican presidents.

    1. But that doesn't really explain why Obama and Clinton have higher "floors" than any of the Republican/conservative presidents of the polling era other than Eisenhower. Sort by highest approval and lowest disapproval here:

      To me, it looks more as if some combination of wars and the economy are the major driving factors behind the historic spikes in polling ceilings and floors for the presidents who had the highest spikes and the lowest floors. The presidents who have the highest ceilings mostly saw them spike during major attacks on the US or major victories in wars. Likewise, the presidents who have the lowest floors saw them come during long drawn out wars, foreign policy scandals, economic recessions, and/or high inflation. Insofar as partisanship drives the floors and ceilings, I think a major reason for that is that the two most recent Democratic presidents have seen fairly steady economic growth after coming into office and have not been responsible for any major foreign policy disasters, whereas the Bushes presided over major foreign policy incidents that spiked their polling numbers (quick Gulf War victory, 9/11) followed by a recession and in the case of the younger Bush a foreign policy disaster. Not that that means there isn't any partisanship involved, and if nothing else there is a genuine ideological difference between the Republican Party and the Democratic Party on the desirability of war as a foreign policy option which would result in a difference in polling variation which has nothing to do with voter partisanship.

    2. I think the floor is a different dynamic - you have to lose your base to have a really low approval rating, and that has more to do with competence than partisanship. Clinton and Obama were/are both widely seen as competent presidents so they never lost their base to any great or lasting degree. Whereas Bush 43 did lose some of his base in the last 2-3 years of his presidency.

      In terms of the ceiling, I just have a hard time imagining any Democratic president achieving a 90% approval rating, as Bush 41 did after the Gulf War and Bush 43 did after 9/11, under any circumstances. Obama managed a 69% approval rating entering office as a savior during the worst economic crisis in 75 years. He ordered and oversaw the raid that killed Bin Laden (America's Most Wanted for 10 years before that) and his approval bounced from the mid-40s to the mid-50s, then quickly came back down. We can never know for sure since each President governs in different times and circumstances, but I suspect a Republican president's approval would have been higher in the same circumstances.

    3. Perhaps. I'm sympathetic to the idea that even in the most optimistic scenarios (miraculous rapid economic recovery and a decisive and rapid win in a war evidently justified by an attack on United States soil), Obama would still not receive higher than 80% approval rating, especially because he's black and I have no doubt that racism continues to cloud some people's perceptions of him as a president. But still, I think the key point here is that achieving something close to a 90% approval rating is not something that a president does; it's something a president has happen to him. It's only happened a few times in the history of presidential polling, once at the end of World War II, once at the end of the Gulf War and in the midst of the breakup of the Soviet Union, and once after 9/11. Each time, it proved ephemeral, with the bounce disappearing completely within a year or two, as the effects of the short term bounces receded and the fundamentals of economic growth and long term foreign policy results and party identification ultimately drove these presidents' approval ratings. I think it's more instructive to write off the Bushes' 90% approval ratings as quirky artefacts capturing short term reactions to particularly major foreign policy events and instead look at the approval ratings of presidents like Reagan and Clinton who presided over more similar economic and foreign policy conditions at varying points during their presidencies and thus share similar approval rating trajectories.

  5. Granted, it could be that it's harder for events now to move the needle, but the reaction to events during the Bush administration just makes me very skeptical of that.

    I disagree with your interpretation of the evidence. The 24/7 media and proliferation of web outlets have only intensified since the Bush years. "News events" go from scandalous to boring much more quickly now.

    Society is enervated by the ubiquitous coverage. Let me coin the clumsy term "Chronic media fatigue syndrome."

    1. I'm skeptical of your "chronic media fatigue syndrome" explanation for reasons I started describing above. If you check the presidential approvals wikipedia page I linked in my previous comment, you'll see that Kennedy, Eisenhower, Reagan, Ford, Roosevelt, and Clinton are all within five points of Obama on the high minus low measure and thus had very little variability in their terms. These presidents all had fairly steady dropoffs between 27 and 36 points from highs in their honeymoon periods immediately after election to lows during periods in which the economy was slow. Nixon, Johnson, and Carter all have significantly larger dropoffs from normal honeymoon periods in which they enjoyed approval ratings in the high 60's and 70's which dropped down into the 20's and 30's amidst scandals (Watergate), foreign policy fiascos (Vietnam, oil embargo, hostage crisis), and economic decay (stagflation). These are all still roughly in line with the presidents who saw their 30-point drops, but a little worse because they enjoyed slightly higher approval ratings during their honeymoon periods and then saw very severe problems that happened during their presidencies. Then you have three presidents who stand alone: George H.W. Bush (Gulf War victory spike followed by recession causing his approval ratings to drop 60%), George W. Bush (9/11 spike followed by Iraq War and Afghanistan War and slow economic growth and 2007 inflation and 2008 mega-recession causing a 65% drop), and Truman (WWII spike followed by Korean War failures).

      There's no way of getting around the fact that Obama is basically in line with historical norms in terms of changes in approval ratings: a reasonably high honeymoon period after his election followed by a 30-point decline during a period in which the economy was very sluggish but not in freefall or hyperinflation, followed by some stabilization somewhere in between the high and low. The polling numbers and their trajectories are comparable to Clinton and Reagan, both peacetime presidents who presided over weak economies when their numbers bottomed out as well. Basically, from a historical standpoint, Bush's huge variations in approval ratings are an aberration. If something new happened AFTER Bush to make presidential approval ratings less variable, from an empirical standpoint, it basically just pushed us back to the historical norm established by Reagan and Clinton.

    2. That was helpful, compelling, and well-explained. Thanks!

  6. I wonder if the Bush years are really a fair test, just given how extreme they were. As in, it doesn't disprove the idea that increased partisanship (I don't feel like I have the full picture on voter versus elected official partisanship, so I'm not presuming that there actually IS increased partisanship, just granting the original premise) is making it harder for presidents to have wide swings in approval just because Bush did - most presidents aren't going to have a 9/11 or financial crisis or Iraq/Afghanistan equivalent. Obama certainly hasn't, Clinton didn't, H.W. Bush is a maybe, but he wasn't quite in the modern era in the sense we're discussing. So, trying to distill that a little better: it could be that partisanship is increasing and that's narrowing the range of normal presidential approval, but W. just wasn't a normal presidency at all, and so overshot the whole thing.

    I'm not arguing that, necessarily, but doesn't it largely work as a hypothetical, if you assume the Klein premise?

  7. I don't think it's quite right to say that George W. Bush's approval ratings were "all over the place." They were, on balance, pretty steadily downward:

    Already by 2004, James K. Galbraith was noting the steady predictability of this. The first-term spikes were 9/11, the start of the Iraq war, and the capture of Saddam. Then there's a bit of a re-election "bump," and then the downward trend resumes and continues through the second term, with a brief trough and brief bump in '06. Otherwise, it's basically the picture of a ship slowly taking on water, listing to the right (as it were), and finally disappearing under the 30% waterline in the crisis months at the end.

    Basically, this is the trend line of a bad president who caught an early break (as it were), but whose failures and unpopular policies created continual downward pressure on his ratings. I would say it was the quality of the presidency, rather than anything about partisan polarization in general or other such larger phenomena, that best explain this.

  8. I definitely disagree with JB that this approval rating highs and lows aren't directly correlated to polarization.

    I understand the argument of events, but after 9/11 we had a rally around the leader, GWB. I can think of no situation right now, including a large scale attack on the US that could have the same result. How long did it take after UBL's assassination for the country to divide? 45 minutes? Every single event has been followed by further polarization and it's not going to stop.


Note: Only a member of this blog may post a comment.

Who links to my website?