A video clip of my appearance last weekend on MSNBC is here. It is brief, but I did state the main point: the shutdown leveled the House playing field in a rather unexpected manner. Republican-gerrymandered districts (FL, MI, OH, NC, PA, VA, WI) have swung away from Republicans about twice as hard as the rest of the nation. Why? When you pack your opponents into a few districts, your own districts end up with a lot of independents, who are swayable.
My own analysis of current surveys indicates that in a Congressional election today, Democrats would retake the House with >90% probability and a 50-seat margin. Last week I pointed out that assuming historical patterns, by next November that probability for the actual election diminishes to 50% or less. Another estimate gives a comparable probability of 25-35%.
However, the last two weeks have been quite strange – as evidenced by the fact that you are visiting the Princeton Election Consortium, reading about polls an entire year before a midterm election! So something might be different. I’ve never seen such a rapid and large shift in the generic Congressional preference. Will this “shutdown bounce” last? Or will it dissipate quickly?
I estimate that in the absence of any further political crisis events, the current bounce will affect opinion for between 2 and 6 months. Here’s the argument.
First, let’s think about it from the standpoint of my own field, neuroscience. The brain doesn’t have one memory system – it has multiple memory systems. Our memory for factual events requires the hippocampus, and those memories can fade over time. In contrast, strong emotional memories (like post-traumatic stress disorder) require just one experience, use the amygdala, and can be very sticky, i.e. long-lasting. On top of this complexity is whether we convert our memories into actions. In this case, the question seems to be whether independent voters will retain their change in preference for a long time.
When suitably aggregated, the generic Congressional ballot is a good predictor of the national vote. Between now and November 2014, I would normally expect the popular vote to be shifted by 3.4+/-4.0 points toward the Republicans compared with current measured conditions. This is based on the fact that most years, opinion seems pretty stable:
Even when a shift does occur, it would be expected to take several months and come as the election draws near. In the wave election of 2010, opinion moved slowly toward Republicans, ending with a big swing starting in October:
But what we’ve seen in the last two weeks doesn’t look like that at all. More than anything else, it looks like a single big event.
To see anything comparably fast, we have to look at Presidential ratings. Here is President George W. Bush’s approval rating after two memorable and emotional events, the attacks of September 11th, 2001, and the invasion of Iraq:
A sharp jump in opinion can also happen during Presidential races, when voters are paying attention. Here are some bounces from 2008 and 2012 associated with Sarah Palin’s VP acceptance speech, Paul Ryan’s addition to the ticket, and the first Obama-Romney debate:
Although all these bounces come from different types of political polls, it might be possible use them to get a sense for how lasting the current bounce will be.
The reason they are useful is that not all bounces are created equal. The Palin and Ryan bounces were smaller (2-5 point swings) and furthermore might have been simply a rallying of party faithful. The largest events, 9/11 (a 60-point swing) and the Iraq invasion (a 30-point swing), are more likely to include Democrats, Republicans, and independents.
What’s notable is that the larger the swing, the longer-lasting it is. It’s not just that there is a longer way for opinion to fall. On average, voters take longer to shift when the swing is large. Why is that? One possibility is that the size of the swing is a measure of how powerful the experience was. Obviously, 9/11 was an unmatched event in public memory. (Measured in units of opinion, the Congressional shutdown was about one-eighth as large in its effect as the 9/11 attacks.)
Now let’s convert the duration of each bounce to a “half-life,” defined as how long it takes opinion to get halfway back to where it was headed before the bounce. I estimate the above bounces as follows:
|Event||Swing (%)||Half-life (months)|
|Iraq invasion 2003||30%||3|
This suggests that a shift of the size we have just seen should have a half-life of 1 to 3 months. Since declines seem to be steady over time, that would suggest that opinion will get back to where it was in 2 to 6 months, well before the election – assuming that no further politically disruptive events occur.
Given the current political climate, it is entirely conceivable that more events will happen. The government is only reopened until January, and President Obama asked yesterday if we’re going to be doing this all again in 3 months. If so, the shutdown bounce would start looking more like a long-lasting shift.