As the Republican candidates and the media gather in Las Vegas for a CNN debate tonight — the fifth GOP debate of this election season — terrorism and foreign policy promise to be a big new focus.

But that issue matrix has changed early and often this election cycle, depending on current events. We've worked with MIT Media Lab's Laboratory for Social Machines to see what topics Twitter users have been interested in this year. The lab, with an assist from Twitter HQ, can analyze all the tweets in the United States. A computer program detects tweets that mention election issues or candidates — ranging from 150,000 to 250,000 a day— and then categorizes them based on the issue and candidate they're about. In that way, we can see which issues or candidates have had the biggest share of the conversation.

We looked closely at the past four weeks. A big spike in the election-related conversation around guns came after the Dec. 2 shooting in San Bernardino, Calif., that left 14 people dead, for example. (The data does not classify whether tweets are pro- or anti-gun-control, just that the topic became a bigger share of the election-related conversation on Twitter.)


Foreign policy and national security, which includes terrorism-related tweets, has taken up a large share of the conversation since Nov. 1 — continuing the trend evident for most of the year.


Immigration, the issue that has been one of Trump's signatures throughout his campaign, was a major topic on campaign-related Twitter even before he entered the race. But the volume went even higher in July and August, after Trump said Mexico was sending rapists and other criminals across the border.


The biggest spikes in election-related conversations around guns have been around mass shootings. The conversation really ticked up after the Dec. 2 San Bernardino shooting, the shooting in Oregon at Umpqua Community College on Oct. 1 that killed nine people, and the June 17 shooting in Charleston at a historic African American church that also killed nine.


Racial issues also spiked in mid-June around the Charleston shooting, in which the alleged shooter was later reported to be seeking a "race war." But the largest spike in racial issues as a share of the election conversation came in August, around the one-year anniversary of the fatal shooting of Michael Brown by a police officer in Ferguson, Mo.


Increases in the conversation around abortion and the election came after undercover videos from an antiabortion group showed a top Planned Parenthood official talking in cavalier tones about the procurement of fetal tissue, and then when the debate over defunding Planned Parenthood picked up in the House.


And it should surprise no one that Donald Trump is the candidate who's dominated the election conversation on Twitter since he entered the race in June.

All the candidates have spiked when they enter the race; there was also an increase in interest in Ben Carson as his poll numbers rose in October. But none could hold a candle to Trump when it comes to attention on Twitter — a trend that mirrors Trump's domination of coverage in the mainstream media, which some Trump critics ascribe to his continued strong poll numbers.


About this analysis

This is the first result of a collaboration between The Washington Post and the Laboratory for Social Machines (LSM) at the MIT Media Lab. The goal of this analysis is to measure what election issues Twitter users are talking about most and how they’re talking about them. The larger goal of the project is to move beyond the traditional focus on the “horse race” competition between candidates and give more attention to the ideas and policy questions at stake.

Thanks to a gift from Twitter, LSM has access to all tweets since the founding of the social network, a database that grows by about 500 million tweets each day. A computer program detects tweets that mention U.S. election issues or candidates, assigning each one to the best-fitting issue category. Human coders check the assignment of random samplings of tweets, and their assessments are then used by the program to make the classification more accurate. The volume of tweets about each election issue is divided by the total volume of tweets about all election issues to estimate its “share” of the conversation.

The conversation among Twitter users is not representative of the public at large. Twenty percent of Americans use the platform, and their demographic makeup and levels of political interest differ from the public overall. The analysis should be viewed as a readout on the views of the platform’s users.

LSM was founded last year to explore new ways technology can be used to solve social problems. LSM’s director is Deb Roy, an associate professor at MIT as well as chief media scientist with Twitter. Also at the lab are experts on journalism, law, politics and other aspects of the public sphere, as well as 11 graduate students and one postdoctoral associate. The project is supported by a grant from the Knight Foundation.