Posting On Social Media About The Elections? Milwaukee Data Scientists Are Keeping Count

Feb 13, 2020

If you post about this year's elections on a blog or on social media like Facebook, you may wind up being a tiny part of new Milwaukee-based study. Researchers at UWM and Marquette University hope to use data science to learn more about voter priorities.

Political polls, usually done by telephone, already tell us some things about potential voters. But UW-Milwaukee marketing professor Purush Papatla says polls have shortcomings.

"Polling can only done at certain times. And, polling takes time to get the data back, and analyze and find the results," Papatla said.

READ: Breaking Down Election Polls

Papatla told a Democratic National Convention host committee community forum Wednesday that he supports polling. But he says a new project involving the Northwestern Mutual Data Science Institute, Marquette University, and UWM can complement polling data by providing more information practically every second. 

The researchers, including students, are using computers to search the internet and social media for when people write publicly about this year's elections. Amber Wichowsky, associate professor of political science at Marquette, says just in the last month there were 148-million mentions, or conversations, not including about the impeachment of President Donald Trump.

(From left) UW-Milwaukee's Purush Papatla and Marquette University's Amber Wichowsky at the Milwaukee 2020 Community Growth Summit, which was held Wednesday at Northwestern Mututal.
Credit Chuck Quirmbach

Wichowsky says besides quantity, the data scientists want to know about what she calls alignment. 

"We know that Twitter users, in some dimensions, aren't your typical American voters. So, are the issues that are being prioritized online, are these the issues that resonate with the typical American voter?” Wichowsky said.

Wichowsky also says the research will look into whether voters and candidates are on the same mental page. So, the data team is also tracking the presidential contenders advertising, debate transcripts, campaign speeches, and tweets. She says there are some early trends.

"When it came to Bernie Sanders' tweets, he really focuses on just a few core items. Billionaires came up quite a bit in that analysis," Wichowsky said, adding that another Democratic candidate, Pete Buttigieg, often references South Bend, Ind., where he was mayor. 

The researchers say they're aware of internet trolls (people who deliberately post inflammatory or off-topic messages) and bots (automated social media accounts). But the Data Science Institute says such fakery is relatively small, and experts are developing tools to better detect it. 

Still, UWM's Papatla says the local research will track trolls, and bots, too because they're having an effect. 

"They are influencing what people think and say. So, we shouldn't throw them out," he said.

More findings from the data science research into elections are expected later this year.     

Do you have a question about innovation in Wisconsin that you'd like WUWM's Chuck Quirmbach to explore? Submit it below.

_