Iowa Caucus: It’s a Marathon AND a Sprint!February 02, 2016
Voters have finally had their first say in the 2016 presidential race. Iowa caucus-goers gathered at locations across the state to listen to speeches and to officially throw their support behind a certain candidate.
Once again, our hypothesis is that social media analytics are better at predicting election outcomes than traditional polling.
Iowa was our first test to this theory.
So, now we come to the most important question of the night:
How did we do?
Let’s start with –
Was Ted’s win a surprise?
At 4:42 PM yesterday, we updated our forecast by including data from 12 AM Sunday, Jan 31st up to 4 PM Monday, Feb 1st. We saw Ted taking a slight lead over Donald then, but Donald was slightly ahead when we expanded the data set to include Friday and Saturday. We knew it was going to be a VERY tight race.
At 6 PM, as broadcasted by CNN, Ted knew it was a tight race and rallied his team to get supporters out to caucus. It has been documented that Ted’s Iowa team is well organized. Props to Ted. He used real-time insights from his ground troops, and turned to action to secure the win.
My wife and in-laws are politically passionate. Two of them are GOP supporters, two are DEM. It makes for lively discussion whenever we get together. As a Permanent U.S. Resident, I can’t vote and never join their arguments debates. But at 5:50 PM, I emailed them updated data, including details that Rubio was doing well yesterday (data I pulled on Monday up until 4 PM). They were very happy about this.
What about Ben & Ted?
Our data showed Ben at 1% versus 9% in the caucus. This is somewhat puzzling to me because Ben was consistently in the top spots (2 through 4) in our national weekly benchmark reports — we started these in August. His numbers started dropping in late November and have never recovered. Likewise, we had Jeb in fourth at 6%, versus sixth place and 2.8%. We need to dig into this further, but in Jeb’s case, I theorize it’s tied to the huge amount of money he spent in Iowa, as reported by The Weekly Standard, and didn’t turn into actual votes.
RealClearPolitics shows that Hillary has consistently polled as the front runner. Using data from Friday to Saturday, we had her leading by 3, in-line with the respected Des Moines Register / Bloomberg Poll. But we saw Bernie surging on Sunday and Monday.
I wanted to understand why our data suggested Bernie would win. I believe there are three logical reasons:
1. Demographics: Age
Social media attracts a younger audience, most notably Millennials. With the youngest being 19 and able to vote now, Millennials are set to overtake Boomers in population this year. But that doesn’t translate to more votes. As reported by a recent Tufts research report, Millennials are less likely to vote than older generations. To win, Bernie will have to get more of them to participate. I pulled in data using the same filters as yesterday and found that Bernie’s audience over-indexes to Millennial when compared to Hillary:
We believe the generational composition differences between the two candidates’ social media supporters contribute to the reason why our prediction differs from final results. We plan to reevaluate this and to determine if we need to adjust our algorithm to replicate real-world generational voting patterns.
Social media conversations about Hillary skew more female.
3. Democratic Procedure
As explained by NPR, Democrats require candidates to secure a 15% viability threshold in each caucus. If they don’t, those supporters have to re-caucus and choose another candidate. As Martin O’Malley never got out of single digit support, we think many of his supporters could have voted for Hillary.
Our Predictions Were More Accurate
With a good methodology and tool set, benchmarking recent history and analyzing real-time data provides extensive view into trends you are unable to get from polling. While nothing can be 100% accurate, these social media insights will arm candidates, journalists and pundits with an ability to see and react to changes as they happen. We will be ready for New Hampshire next week, and will be looking forward to refining our approach throughout the 2016 Presidential Elections.