Some results are counterintuitive, though. Twitter users in some of Utah's reddest counties were more positive toward Clinton than Trump, including Uintah (in which Mitt Romney bested Barack Obama 90 percent to 9 percent in 2012), Morgan (89 percent to 9 percent), Duchesne (89 percent to 9 percent), Sevier (89 percent to 9 percent), Millard (89 percent to 9 percent) and Box Elder (88 percent to 10 percent).
Meanwhile, among the 18 counties skewing red are the traditionally bluer-hued Salt Lake, San Juan and Grand counties.
The U.'s College of Engineering team started with 250 million geo-tagged tweets, of which 1.6 million related to the presidential election. The software designated each tweet as positive or negative and assigned it a score.
"With sentiment analysis, it will try to predict the emotions behind every human being when he or she is talking or writing something," Debjyoti Paul, a U. doctoral student and the project's co-leader along with associate professor Feifei Li, said in a news release. "With that in mind, we are not just trying to look at the information in the tweets. We are trying to incorporate the emotion with the information."
Uses for the team's software may include more accurate measures of product and restaurant satisfaction and better-informed voice-enabled assistants, like Apple's Siri.
Among the software's other election-related findings:
• Republicans were more frequent political tweeters than Democrats by 17 percent.
• Delaware was the only uniform state, with all its counties leaning blue.
• Surges in pro-GOP tweets were observed during the Republican National Convention in June and after a video showed Donald Trump describing sexually aggressive conduct toward women in October, while surges in pro-Democrat tweets came after the past two debates and after an Oct. 1 Times story on Trump's avoidance of federal taxes.