In response to my earlier post, Stephen Nicholson sends me this piece (gated), in which Nicholson, Gary Segura, and Nathan Woods find that divided government boosts presidential approval by about 2 points, controlling for several other factrors, The piece was published in 2002 and their data cover the period from 1949-1996, so later periods of divided government are not included.
I read this evidence as suggesting a small effect, so I don’t think it should provide much consolation to Obama if Republicans take one or more chambers of Congress on Tuesday.
I was then inspired to investigate an oft-repeated claim that’s been subject to little thorough testing: divided government helped Clinton by boosting his approval ratings during the government shutdown. in 1995 and 1996. Here, for example, is Talking Points Memo back in April when Newt Gingrich was suggesting that Congress simply refuse to fund various federal agencies, which TPM read as his endorsing a shutdown:
This sounds a lot like Gingrich’s actual playbook during the GOP’s first two years in the majority in 1995 and 1996, when he shut down the government twice during political battles with Bill Clinton. The end result was that the public blamed the GOP for the impasse, and Clinton’s approval ratings went back up and he was re-elected easily.
Now, to be clear, public polling about the 1995-96 shutdown itself suggests that the more of the public supported Clinton than Gingrich. Steve Kornacki at Salon provides some of that data in this post. But did his approval ratings go up?
Below is a graph of Clinton approval in 1995-96, with a smoothed trend line in blue and vertical lines that demarcate the two periods of shutdown. I have made the trend line to do less “smoothing” and therefore be more sensitive to changes in poll numbers—essentially stacking the deck in favor of finding some apparent effect.
Despite this deck-stacking, Clinton’s approval rating appears largely unaffected by the first shutdown. Some apples-to-apples comparisons of polls make this even more apparent. For example, consider Gallup. Kornacki writes:
By Nov. 20, another Gallup poll pegged Clinton’s approval rating at 53 percent, the highest it had been in nearly two years. Suddenly, the ‘94 midterm debacle seemed like ancient history.
That’s true, but it overlooks a key fact: a Gallup poll conducted November 6-8, before the first shutdown, pegged Clinton’s approval at 52%. Or consider CBS/NY Times polls from October 22-25, 1995, three weeks before the first shutdown and, from November 19, 1995, the last day of the first shutdown. In both polls, 48% approved of Clinton.
The second shutdown actually coincides with a decline in Clinton’s approval. I’m not going to make too much of that because this graph does not allow for strong causal claims. I’ll just note that at the end of the second shutdown, Clinton’s approval rating was really no higher than it was before the first shutdown.
There are perhaps other ways we can score the government shutdown. As I noted, other public opinion data suggested that more people faulted Gingrich than Clinton. And clearly Gingrich blinked first. Clinton “won” in that sense.
But the public did not come to feel more favorably toward Clinton during this period, which suggests again that life under divided government isn’t easy, even when presidents fight the opposite party and win.