NORTH CAROLINA — The day after Mitt Romney announced Paul Ryan as his vice presidential pick, the two traveled here for a couple of campaign stops, drawing decent crowds. How decent? That’s a good question.

John Frank of the Raleigh News & Observer estimated 1,700 people came to see the pair at the NASCAR Technical Institute in Mooresville, with 1,000 outside. The Washington Post’s story had the same inside estimate of 1,700, but also reported that 4,000 people were outside the venue. NBC News said the two candidates were taking advantage of “enormous” crowds. and then followed with a crowd estimate of 2,000.

Later, at the next stop in High Point, Zeke Miller of Buzzfeed tweeted that the Romney campaign said the crowd numbered 10,000 to 15,000, an estimate that was frequently retweeted. The Greensboro News & Record used the number in its story without attribution after initially posting that no crowd estimate was available.*

Miller, who seems to routinely shares crowd size estimates from campaign sources on the trail, followed up by agreeing with and retweeting Garrett Haake of NBC News, who said that there was no way the traveling press could verify the Romney campaign’s estimates since they couldn’t see overflow areas. Haake estimated the outside crowd at 2,000 or 3,000. The Associated Press and Reuters both stayed away from crowd estimates in stories about the NC visits, as did The New York Times in its blog, The Caucus.

Meanwhile, throughout the day all the way to Ryan’s homecoming in Waukesha, Wisc., the Romney campaign supplied its own crowd estimates—starting with 4,700 in Mooresville, 10,000 to 15,000 in High Point, and then 13,200 in Waukesha—numbers that fit a nice narrative for advocacy media and bloggers but sounded over the top. The Washington Examiner headline—“New Obama panic: Romney crowd sizes”—made the Drudge Report this morning. But the traveling media began to stop taking the numbers at face value. NBC’s Haake shared a fire marshal’s estimate in High Point of 1,200 inside the main venue, 2,000 in an overflow space and no real estimate outside. And in Waukesha, Miller hedged his bets after tweeting the campaign’s official estimate of 13,200: “Crowd doesn’t appear quite that high. More than 5K for sure. Hard to tell since people are standing.”

Indeed, crowd size is hard to estimate accurately, though it often seems to be part of a story. And often enough, estimates can provoke partisan outrage—as when early estimates of 2 million at the big Tea Party rally in Washington in September 2009 were deemed dubious , or when the Million Man March in Washington back in 1995 was said to be less than that.

The high end of crowd estimates in North Carolina even made it into a fundraising email, but not from the side you would expect. The Democratic Congressional Campaign Committee sent an email Aug. 13 using the number to try to raise funds. The email said: “I just got this disturbing report: Yesterday’s Romney-Ryan rally in North Carolina pulled in an overflow crowd of 15,000 people. There’s no spinning that number.”
Or perhaps there is.

What’s disturbing, or perhaps enlightening: The DCCC email cited The Wall Street Journal’s coverage of the NC Romney visit, summoning media authority using a paywalled source, but the Journal story only used “thousands” to describe the crowds. And what’s heartening: The Under the Dome political blog from the Raleigh News & Observer noted the DCCC fundraising letter and called out the 15,000 number as an exaggeration, “grossly overblown.”*

Now, with many more candidate visits in the works, along with possible crowds of supporters and protesters at the national conventions—the GOP’s in Tampa at the end of August and the Democrats here in Charlotte the first week of September—How can journalists measure crowds? Is it even important to give a crowd estimates when they are so hard to nail down?

Luckily, journalists collectively have lots of experience and advice.

• In 2008, Jane Kim wrote here at CJR about the Obama campaign’s tendency to provide crowd counts along with validators of those counts, noting that sometimes those validators weren’t necessarily the best judges. The short version: “Let’s use caution in employing unverifiable numbers in the service of colorful crowd descriptions.”

• In 2009, Steve Doig, a data journalist and journalism professor at the Cronkite School at Arizona State University, wrote about crowd counts ahead of the Obama inauguration. He spelled out a mathematical method and also suggested photos as ways to help estimate massive crowds.

• In 2010, the editors here at CJR asked, uh, the crowd about whether journalists should even try to do crowd counts, with a post noting that the National Park Service no longer gave crowd estimates for protests in Washington, DC. One tongue-in-cheek comment, from James Lynch: “An AP reporter on the presidential campaign trail told me once to get an accurate crowd estimate ask the organizer, divide that number by 2 and subtract a third.”

• In 2011, on the World blog for what was then MSNBC and is now NBC News, journalist Alex Johnson examined the difficulty of measuring crowds in Tahrir Square, where estimates ranged from 100,000 to 2 million, and he referred readers to Doig’s post. The best part about this post: It wasn’t just aimed at journalists but also at the readers of news at MSNBC’s site.

Crowd estimates do matter. For one thing, growing crowds set a narrative of increased interest and support for campaigns or causes, and organizers often cite the crowd numbers used by media as proof of that squishy term, “momentum.” That can translate into donations, so the motivation for potential money can encourage organizers to spin numbers ever higher. Being as accurate as possible when making crowd estimates, then, matters because the numbers can produce effects. So the watchwords are care and caution.

For example: take reports from protests in May in Charlotte during the Bank of America’s annual shareholders meeting. Initial reports by media sympathetic to protesters estimated “thousands” protesting. The Charlotte Observer appropriately exercised caution and said 500 to 750 people protested, in a story that gave the police estimate (400 or 500) and the organizers’ own estimate after the event was over (750).

For the Romney and Ryan visit to Mooresville on Sunday, the Statesville Record & Landmark took a smart approach amid a dearth of immediately available official crowd estimates. Reporter Jim McNally interviewed a USA Today reporter, Jackie Kucinich who had been traveling with the Romney campaign to get some context on the crowd. (The reporter, by the way, is the daughter of Dennis Kucinich, a Democratic US congressman from Ohio.) McNally quoted Kucinich: “I’m not good at estimating crowds but this and all the people outside, this looks like the biggest one I’ve seen,” she said. “The crowds were maybe 1,000 to 1,500 people most of the time but yesterday [Saturday] they were bigger, and this is certainly bigger.”

While there’s some danger in reporters interviewing reporters, in this context, way down in the story as a method of trying to nail down the relative size of the crowd, the technique worked nicely. Providing context, being open about how difficult crowd estimates can be, and finding “validators” or knowledgeable sources for the numbers, are sound principles for journalists as the crowds gather this fall.

Clarification: The News & Observer’s use of the high GOP crowd-size estimate in High Point was added to this story after it was initially published. Also, the two paragraphs starting with “The high end of crowd estimates” were also added shortly after this story was initially published.

Andria Krewson is an independent journalist in Charlotte and a student in the University of North Carolina's master of arts in technology and digital communication. She worked at The Charlotte Observer for many years. Find her on Twitter at http://twitter.com/underoak or http://twitter.com/akrewson.