Rooting for the Home Team: Debates and Elections

Today will mark the second presidential debate for the 2018 Election in Brazil, and it promises to be memorable. There will be even-handed back-and-forths, deft moderation, and high-minded discussions…for an approximate total of five minutes. The other 115 minutes will likely involve conspiracy theories, slander, and unintelligible shouting.

Debates, of course, are supposed to be a cornerstone of democracy. Voters tune in to be swayed by the best arguments, and then make their choices based off of that. Lots of conventional wisdom surrounding debates has to do with this potential to persuade. Kennedy-Nixon. Lula-Collor. Get your (wo)man up there, looking good, with no five-o’-clock shadow, and write your name in political lore with some snappy one-liners.

To test this ability to persuade, I (with André Bello and Lucio Rennó from the University of Brasília) took data from the wonderful, aptly-named 2014 Brazilian Electoral Panel Study (BEPS). The BEPS has seven waves of panel data and questions about watching, talking about, and hearing comments about three separate debates before the first round of the election.

In a paper that will be out for submission soon, we used a difference-in-differences model to measure both the direct effects (watching them) and indirect effects (talking about them or hearing comments about them) of debates on vote intentions. In other words, were respondents more likely to change who they were going to vote for in the first round after watching debates?

In a word, no. In two words, hell no. In fact, those who paid attention to the debates were actually less likely, on the whole, to change their minds.

table-4129918790-1534555598504.jpg

This was probably because the audience of debates is far from open-minded. Of those in the sample who watched the debate on Record, less than 10% were undecided (as opposed to those who didn’t, which added up to almost 30%). Of those who watched the debate on Globo, less than 5% were undecided (as opposed to 14% of those who didn’t).

So do debates themselves crystallize opinions? Or is it simply a selection effect? It could be both, in all honesty, and it will be difficult to tell without an experiment in which the selection is randomized and controlled.

Will this be the same in 2018? Hard to tell because the field is so wide open. There are probably many more genuinely undecided voters tuning in now (or decided Lula supporters looking to see if anyone is up to snuff) so it could be different this time around.

What about the effects of second-round debates? I’ll tell you that, too…eventually.

In the meantime, sit back and enjoy the memes.