Rose Parade: Big crowd, but who's counting?*
Peter Apanel of Portland, Ore., was counting noses, not smelling the roses, when he read the day-after story on the Rose Parade. His e-mail said: "The Times reported that an estimated 700,000 people attended this year's Rose Parade. That figure is obviously false, so I'm wondering why The Times continues to report false attendance figures year after year."
Apanel's comments led California Editor David Lauter to send his own message to the staff (see below) about the problems in providing crowd counts and offering some guidelines for avoiding the sort of errors Apanel had noted.
Here's the passage from the Jan. 2 story that drew Apanel's attention: "More than 700,000 people lined the parade route, a fire official said, the biggest crowd count in three years -- thanks in part to the warmer weather, which climbed into the 70s. It was the kind of bright, clear day that parade boosters have long used as a floral-scented advertisement for sunny Southern California."
Readers so often raise questions about crowd estimates in stories that there's an entry in the "frequently asked questions" part of this journal.
As for the Rose Parade, as a Times story in 2003 reported, "it's almost impossible to come up with accurate estimates of the crowds that pack the route each year. Tournament officials long claimed that 1 million people attended the parade -- a number that researchers at Caltech have said is statistically impossible. Last year, the Tournament estimated parade attendance at 800,000, attributing the lower turnout in part to fears brought on by the 9/11 attacks." (The Jan. 2, 2002 story said, "While broadcasters trumpeted the traditional crowd estimate of 1 million, police and veteran parade-goers said the crowd seemed slightly smaller this year.")
The Jan. 2, 2008, story was more cautious, reporting that "tens of thousands of people lined the streets of Pasadena and millions more watched on television from around the world."
The Times wasn't as careful in 2007: "With an estimated 1 million spectators, finding a spot to view the festivities was as tough as ever."
Pasadena parade aficionados probably already know this, but for those who don't recognize the reader's name, Apanel is, as a 2001 Times story says, "founder and self-appointed Doo Dah czar," the Doo Dah Parade being a spoof of the Rose Parade. His e-mail explained, "I was the organizer of the Pasadena Doo Dah Parade for its first 18 years, and in 1983 I organized a team of photographers to photograph all of the spectators along the entire length of the parade route. Afterwards, I pieced together the prints, counted everyone in attendance, and came up with a figure of 31,957 for the 1.25 mile long parade route. We didn't have bleachers, but the density of spectators was comparable to that of Rose Parade spectators."
Apanel says his own research since then has shown what The Times' 2003 story reported: "In the microfilms of old newspapers in the public library I found stories going back to the early 1930s which contained crowd estimates of 1 million-plus spectators at a time when the Rose Parade was just over half its current length. This is a sacred cow with very old roots, which probably explains why so many people would rather do the myth, than do the math."
For those ready to do the math, Apanel's proposed method of counting is at the end of this post. Below is Lauter's note to staff on the difficulty, if not folly, of tossing around crowd counts.
There are things in the journalism business which we get so consistently wrong that one begins to wonder whether we labor under some error-dooming curse. Consider the endlessly repeated problem of unreliable crowd counts. Political rallies, anti-war demonstrations, pro-war demonstrations, concerts, tractor pulls -- at virtually any large gathering, one question which naturally occurs to a reporter is "I wonder how many people are here?" The question is logical. So, up to a point, is the reporter's usual method of getting an answer, which is to find someone in authority and write down what he or she says, then put it in the paper, with proper attribution, of course. "Police said ... protest organizers said ... the fire department said ...." The problem, sadly, is that while the quotes may be accurate, what those people say on the subject is almost never true.
Counting any large number of things is hard to do even when the objects are inert and the counter has no deadline. The job is vastly more difficult when one is trying to count people moving around in a space with unclear boundaries and the question needs to be answered right away. There is a standard method for doing the job. It involves taking pictures of representative parts of the crowd, superimposing a grid on top of the pictures, counting the number of people in several squares of the grid and then multiplying to get the full count. Needless to say, the police, protest organizers or fire officials to whom reporters address their questions about crowd sizes almost never do any of that -- they have jobs to do. Instead, they use what George Berklacy, chief of public affairs for the National Park Service in Washington, used to call the SWAG system--scientific wild-ass guess. And we write it down. We really should stop.
My reason for going on about all this is that a reader wrote to chide us for including in our most recent Rose Parade story a crowd estimate of "more than 700,000," attributed to an official of the Pasadena Fire Department. And meaning absolutely no disrespect to the reporters involved in the story, who are excellent, the reader was absolutely right.
The unreliability of the Rose Parade crowd count has become a legend of its own. For years, Pasadena officials used to annually repeat a crowd count of one million. In the 1980s, a gentleman named Christopher Lee used to write Letters to the Editor taking us to task for repeating that count. In that same era, a reporter from the Pasadena Star News worked with the late Al Hibbs of the Jet Propulsion Laboratory to estimate the maximum possible crowd along the parade route: 512,000 Hibbs reported, "giving every benefit of the doubt to the upper number." In 1991, our former colleague Jesse Katz wrote a page-one story on the folly of crowd counts. Jesse's story even included a handy formula for calculating a maximum-possible crowd size for parades (it works out to about 130,000 people per mile of parade route if they're all standing up, packed tight as sardines the whole way with no room for coolers, chairs or empty spots at the back).
For a time, all that attention gets results. In 1997, our Rose Parade story (by Frank Clifford) quoted a police spokesperson as estimating that "hundreds of thousands of people" lined the parade route and noted that "parade officials said they abandoned crowd counts after previous attendance figures were criticized as too high." That was the right solution. But memory is fallible, newspaper staffs come and go, and error creeps back in.
So, please, in this new year, let's resolve to do better: Let's not use numbers that are simply someone's guess. Unless there is good reason to believe that we have a real, scientific crowd count, it's best to simply give an order of magnitude: tens, hundreds, thousands, hundreds of thousands. What we give up in fake precision, we'll gain in actual accuracy.
Here is reader Apanel's methodology based on his Doo Dah experience: "The route is 5.5 miles long, or 29,040 feet. So, if you allowed two feet per person, for spectators standing shoulder-to-shoulder, it would take 29,040 people to form one row of spectators along both sides of the entire parade route. Therefore, to have 700,000 people in attendance, you'd need to have 24 rows of people, packed together like sardines, on both sides of the entire parade route. And even when you factor in bleacher seating, which accounts for perhaps 100,000 people, at best, that would still require 21 rows of people standing along both sides of the parade route. Obviously, that's physically impossible, and, just as obviously, the density of the crowd, with all the chairs, coolers, blankets, and other gaps, comes nowhere near maximum density.
"To come up with an accurate crowd estimate is equally simple. The solution is to use the same basic techniques that scientists use to count flocks of birds in the wild, or blood cells in the human body. Take photographs of several representative sections of viewing area along the parade route, and extrapolate the numbers from there."
*Yet more: In an e-mail Tuesday night, Lauter refers to Patt Morrison's Jan. 1 post on this subject, which said that a Times science writer once went out and measured the sidewalk space available for parade viewers; Lauter noted that the writer in question was Robert Gillette. Gillette's story appeared a couple of days after the 1979 parade. He measured the sidewalk (23 feet from the blue line on the pavement to the front of the buildings) and wrote a story in which he also cited a letter to the editor in the Pasadena Star News which reported that three Caltech PhDs had calculated a maximum possible crowd of 500,000.
Photo: Pre-parade viewers along Colorado Boulevard in Pasadena Jan. 1 at 5 a.m. Credit: Jay L. Clendenin / Los Angeles Times