With two debates behind us, the first presidential debate and the vice presidential debate, public discussion has swirled around the question of who won, and what impact did it have? Since these are not the rigorously judged debates of the high school/college debate circuit, winners may be determined by pundits, by polls, or even by the candidates themselves.

We decided to take a slightly different approach. Since we’re data fanatics, we have been tracking the candidates and election issues for quite some time, feeding the results of Google Trends searches of various keywords into — of course — our own business incident detection system, to see if and when anomalies occur in the data.

Web Searches for “Donate” = Support for Candidate

Early on, we decided we needed a way to measure support for the main candidates, Hillary Clinton and Donald Trump. We determined that the word “donate” together with each candidate’s last name was a decent proxy for this. In other words, if users increased the number of times they searched for “clinton donate” or “trump donate,” this indicated increasing support for the candidate in question.  (BTW we checked if it made any difference if we used the candidates’ full names or just the first name, and we checked various forms of the word “donate”, and the most representative results came from using the last name together with the word donate.)

We also tracked certain hot button issues, like abortion and taxes, to see if there were any correlations between changes in the search data for these issues and support for the candidates. You can see some of the results of our data tracking in the screenshot below.

Data Reveals: Increased Support for Clinton Post Debates, and After NYTimes Tax Article

Searches for both candidates + the word “donate” increased immediately after both debates, however the increase in Hillary Clinton’s “donate” searches surged upward more significantly than Donald Trump’s in both cases, indicating that she generated a groundswell of public support after the event.

Clinton had another anomalous data event with regard to people searching to donate right after the New York Times broke the story about Trump’s tax returns, indicating that he may not have paid taxes for 18 years. The story went live on the publication’s web site on Saturday night, Oct. 1, and then in the Sunday print edition October 2, and there is a clear jump in Clinton’s “donate” searches on the 2nd, which correlates to searches for “Trump and taxes.”

The interest in “Trump + taxes” tapered off rather quickly — in a matter of days — and by the time of the VP debate it had declined significantly.

election-search-trendsSometimes All You Need is a 3am Tweet Storm

Interestingly, Trump’s donation searches jumped soon after his 3am tweet storm, on Sept. 30, to a level nearly as high as his post-debate peak in donation interest. Trump has another few peaks, on Oct. 1 and 4 (before the VP debate) for which we were not able to find any rhyme or reason – we welcome your ideas in the comments.

Will the 2016 election be decided between the Twitterverse, staged TV events and the New York Times? By tracking time series data it’s possible to to see trends, identify anomalies and correlate between the various pre-election events and the mood of the electorate.

featured image credit: CNN

Written by Anodot

Anodot leads in Autonomous Business Monitoring, offering real-time incident detection and innovative cloud cost management solutions with a primary focus on partnerships and MSP collaboration. Our machine learning platform not only identifies business incidents promptly but also optimizes cloud resources, reducing waste. By reducing alert noise by up to 95 percent and slashing time to detection by as much as 80 percent, Anodot has helped customers recover millions in time and revenue.

You'll believe it when you see it