Shellshocked, that they lost the election. Dumbfounded, that their polls were wrong. Meltdowns, on national TV and in tweets. Why were these smart people so completely surprised?
On election night, Mitt Romney and the Republicans were shellshocked. I know that several people have argued that this is problem with how Republicans think. Republicans supposedly have a world view that makes them less open to scientific evidence and more likely to dismiss reality in favor of their beliefs (for examples, see a recent column by Paul Krugman and a book on the Republican brain by Chris Mooney). But this is not a Republican problem. With a different election outcome, I would be writing about shellshocked Democrats. The surprise that Mitt Romney and the Republicans experienced reflects fundamental failures of thinking. Everybody, Republicans and Democrats, sees what they want to see. We all bias our data analyses to get the answers we want. And by imagining a possible outcome, we make ourselves believe that the outcome is more likely to actually happen. Of course, reality doesn’t always conform to our beliefs, biases, and imaginings.
We are all victims of the confirmation bias – the tendency to see what we expect to see and ignore things that don’t fit with our beliefs. In the weeks leading up to the election, hundreds of polls were released. You could find polls with Obama clearly ahead and polls with Romney winning. You could see poll trends showing Romney gaining in critical battleground states and others showing Obama’s lead holding. But if you really wanted to see clear Romney trends, then you had to accentuate some polls and ignore others. If you wanted to be confident of an Obama victory, then you had to watch state polls and ignore national ones. Constructing your biased version of the election wasn’t a problem for most of us because people noticed and remembered the polls that fit with their beliefs. There were even some individuals who “unskewed” the polls that weren’t consistent with their beliefs to make every bit of evidence fit with their world view.
Since many people worry about getting obsessed by only the numbers they liked, several groups and individuals constructed poll averages. Of course, the averages of the polls were only as reliable as the actual polls themselves and this leads to the second cognitive problem that led to shellshocked Republicans.
The second mental mishap that afflicted Republicans this year (and that will surely catch Democrats in another election) was biased data manipulation. The results of polls are not the original poll responses. Few people respond to telephone polls and some groups of individuals are more likely to answer phone surveys than others. The people who computed the percentage of Obama and Romney voters constructed their estimates after making several assumptions and decisions about the raw poll numbers. The pollsters had to decide how many people to call in which locations, whether to call cell phones as well as land-lines, whether to use robocalls or live interviewers, who was a likely voter, what percentage of different age and racial groups would actually vote, and how to figure in the impact of early voting. At each decision point, pollsters’ cognitive biases can sneak into the process. Do you believe young voters and minority voters won’t show up on election day? Do you think conservatives are more enthused this year? Do you think cell phone only individuals are very similar to people who still have landline phones? Romney and the Republicans had their own private pollsters (as did Obama and the Democrats). Both sides made assumptions about how to best conduct their polls and about who would vote. Some of these very smart people assumed an incorrect mix of who would vote. For several pollsters, their biased assumptions led to election night reality shocks.
Of course, this is why Nate Silver (of the New York Times 538Blog) is one of the heroes of this election – his assumptions in computing an average of the state and national polls were clearly supported by the election outcome. The bias in selecting polls you like and the decision errors that many pollsters made have been widely discussed following the election.
But one other mental mishap may also have led to shellshocked Republicans. This other cognitive bias has not been widely discussed but nonetheless may have contributed strongly to the surprise Romney and the Republicans experienced. When you imagine potential future events, you start to believe that the event is more likely to actually happen than if you don’t imagine the event. Szpuner and Schacter (2012) reported that repeatedly imagining events additively contributes to your belief that the events will happen. Importantly, their mental simulation effect only worked for emotional events – both positive and negative events. Romney’s polls numbers began to rise after the first debate this year. When Republicans looked at those poll numbers, many began to imagine winning the election. They began to imagine Mitt Romney as president. I’m also sure that this was a positive event for Republicans. I also suspect that they repeatedly imagined a President Romney. I think Mitt Romney fell for this cognitive bias as well. He saw his campaign’s poll numbers (which were, as we now know, a bit biased). He began to imagine his future as president. He regularly described what he would do on his first day. I read reports that on election day he believed he would win. According to reports, Romney only prepared a victory speech for election night. Clearly Romney and many Republicans imagined winning the election. But imagining doesn’t make something real. They were shellshocked when their beliefs and imaginings crashed into the wall of reality.
Sometimes our cognitive biases run into the wall of reality. I actually think reality doesn’t crash our beliefs as often as it needs to. In many situations, we can hold onto our biases by selecting and noticing only some information, by biasing how we collect information, and by imagining our desired future outcomes. In many situations, reality exists, but we can ignore it. But in an election, eventually the votes get counted and someone is shellshocked by reality.