(Seth Wenig/AP)

As we approach the likely end of Bernie Sanders’s bid for the presidency, what from the volumes of campaign trail data might have foretold the end of his campaign?

Drawing on the Internet Archive’s Television News Archive, the graph below shows each Democratic candidate’s percentage of mentions of all the Democratic candidates across the major U.S. national television networks from Jan. 1, 2015, through this past weekend.

Almost from his entrance into the race, Sanders captured 20 to 40 percent of the mentions of Democratic candidates and pulled even with Clinton in early January. The two remained even, despite a few ups and downs for one or the other, until April 20, when Clinton began pulling ahead in terms of news coverage.


Google Trends data shows a similar timeline, with Sanders actually beating Clinton in Web search interest through the week of April 24, when they pulled even. YouTube searches show an even more dramatic fall for Sanders, from nearly doubling Clinton searches over the past year to nearly equal interest starting this same week.

To capture how the two candidates were portrayed visually, more than 70 million worldwide news images have been processed through Google’s deep learning algorithms over the past six months (the same algorithms used to watch political television ads). In mid-May, the images appearing in articles about Sanders were compared with those appearing in articles about Clinton over a 72-hour period.

Photographs appearing in Sanders coverage tended to depict large arenas filled to capacity with cheering crowds, while photographs appearing in coverage of Clinton depicted many topics focused on current events. In short, the visual narrative surrounding Sanders focused on the size of the energetic crowds he was drawing, while the narrative surrounding Clinton was on her reaction and connection to global events. One plausible implication: Clinton was being covered as a potential head of state, while Sanders was being covered as a cultural phenomenon.

Perhaps one of the most powerful insights comes from a talk that Jordan Tigani and Felipe Hoffa gave recently at the Google I/O 2016 conference. In their presentation of how Google’s BigQuery database can be used to analyze very large election datasets, they used Reddit to examine who each candidate’s supporters really are.

For this demonstration, they took all of the users posting in the SandersForPresident, The_Donald and hillaryclinton subreddits and looked at which groups those users were posting in four years ago. Perhaps unsurprisingly, Trump’s commentators were active in groups such as “guns,” “MensRights,” “conspiracy” and “ronpaul.”

But the most interesting comparison is between Sanders and Clinton. Clinton’s commentators posted across topics ranging from “lgbt” to “PoliticalDiscussion” to “baseball” and “soccer.” Sanders supporters, on the other hand, were more tightly clustered in topics such as “occupywallstreet” and (outer) “space.”

Put another way, the users commenting on Reddit about Clinton’s bid for the presidency had previously been posting on Reddit about many different topics. Users commenting about Sanders, however, had been posting on a narrow range of topics, and especially left-leaning topics, such as Occupy Wall Street.

In short, Clinton’s commentators on Reddit seemed to span a broader spectrum than Sanders’s supporters. Of course, Reddit is just one of many social media sites, but it offers at least some insight into the backgrounds of each candidate’s supporters.

Based on these numbers, Sanders was always going to be vulnerable as the election moved beyond the specific issues important to his supporters. After after the New York primary, Sanders’s media mojo waned alongside any chance of winning the nomination.

Kalev Leetaru is a senior fellow at the George Washington University Center for Cyber and Homeland Security. He thanks the Internet Archive’s television news archive for the use of its data in this analysis.