In what has been a Presidential campaign filled with surprises, Friday's letter from FBI Director James Comey has sparked the latest wave of political hysteria. Putting the politics and merits of the latest email controversy aside, an interesting polling question arises--how does such news affect the polls, and how will pollsters actually tie this controversy to polling numbers?
As we get closer to the election, the rapidity of polling results is increasing. Pollsters almost constantly have polls in the field. The problem, of course, is that in the race to get out the latest results, polls lag over several days. So, although there will be a rush to assign any movement in the race towards Trump to this latest email controversy, we need to be careful in assigning causality to one news story, especially so close to the race.
In fact, this is exactly the time of a race where both sides tend to consolidate their bases. The reported "tightening" in the polls over the last week has mainly been driven by Republicans consolidating their support for Trump. It is possible these latest allegations could consolidate that support even more, but it is a part of an expected consolidation trend that occurs close to Election Day.
As we are closer to the end of the race, the subset of voters for whom news will now change their vote is getting smaller and smaller. As much as I hate the overused pundit expression "baked in the cake," many voters who are pro-Clinton have already accounted for her prior email issues, as have many voters who are pro-Trump. The real effect of such news may be to just reinforce prior opinions.
So, when you look at the polls this week--and you hear news outlets trying to ascribe this email story to any changes in candidate support--ask yourself:
1. What dates were the polls taken? Given the news was announced on Friday, has the voting public really had enough time that poll results today can incorporate this news?
2. Could there be effects of such news that aren't reflected in the simplistic metrics "vote for Trump" or "vote for Hillary"? Are there other questions in the polls besides the "horse race" that specifically ask about voters' views on emails that might provide better insight?
3. Finally, remember, turnout matters. And, predicting the effects of such news on turnout can be very tricky--and is something pollsters are far less adept at figuring out. For example, would negative news due to emails galvanize Republicans to vote for Trump, or could it galvanize a liberal base that feels it is unfair to Clinton to vote in greater numbers? Both are possible!
One of the great parts of polling is that it helps provide some concrete data on the state of the electorate. The downside is it can lead us to oversimplify complex narratives. Voting behavior is challenging to predict, especially in light of a campaign where so many negatives already exist about both candidates.