Mark Treinen, 52, editor, Green Bay Press-Gazette, Green Bay, Wis.

Treinen has served as the editor since April 2019. He is also the news director for USA TODAY Network in Central and Eastern Wisconsin. He has been an editor with Gannett newspapers in Wisconsin for 25 years.

Just one day after voters selected Donald J. Trump as president, the Pew Research Center published an article aiming to answer this question: “How could the polls have been so wrong about the state of the election?”

The authors acknowledged it was too soon to tell exactly why election polls showed Democrat Hillary Clinton with a commanding lead headed into November. But their article shared the thoughts of many pundits at the time—that pollsters asked the wrong voters, or that people who responded were afraid to say they’d vote for Trump, or that analysts made faulty predictions about which candidate would generate more enthusiasm on Election Day.

In the nearly four years since that election, the narrative about disastrous polling has evolved. Nate Silver of FiveThirtyEight observed in 2018 that in retrospect, the 2016 polls were only slightly farther off the mark than for the four previous presidential elections and actually average in accuracy in a longer-term historical comparison.

All of which is to say that newsrooms should rely less on polls—survey responses are not the same as completed ballots—but the results still provide information to help identify shifts in mood, attention and attitudes about elections, candidates and issues. As one source in a broader strategy for covering elections, polls are not inherently bad.

Our experience in local politics reporting in Wisconsin, including during the 2016 election, demonstrates where we should shift any time and attention we take away from poll-obsessing: Journalists covering major elections should speak to and observe voters themselves, a lot of voters over a broad geography and period of time.

We’re doing more of that in our news organization, and we are positioning ourselves to catch trends we might have missed if we were overly focused on survey results. Those subtle changes in tone and rhetoric, the more prominent and prolific campaign signs in our communities, the political donations pouring in from individuals, even—gasp!—Facebook comments are all indicators that should feed into news coverage of the election.

Then, maybe, the question in November will be: “How did news organizations get this  election so right?”