Attorneys representing Craigslist told Congress last Sept. 15 that the ubiquitous web classifieds site was closing its adult section.
Under intense scrutiny from the government and crusading advocacy groups, as well as state attorneys general, owner Craig Newmark memorably applied the label "Censored" in his classifieds where adult advertising once appeared.
During the same September hearing of a subcommittee of the House Judiciary, members of Congress listened to vivid and chilling accounts regarding underage prostitution.
EDITOR'S NOTE: Village Voice Media, which owns this newspaper, owns the classified site Backpage.com. In addition to used cars, jobs and couches, readers can also find adult ads on Backpage; for this reason, Women's Funding Network (WFN) and its allies have often called attention to the site, sometimes going so far as to call for its closure.
Certainly we have a stake in this discussion. And we do not object to those who suggest an apparent conflict of interest. We sat quietly and did not respond as WFN held symposiums across Americafrom Seattle to Miamidenouncing Backpage. Indeed, we were never asked for a response.
But then we looked at the "science" and the media's willingness to regurgitate without question these incredible statistics. In the interest of a more informed discussion, we decided to write.
- For background articles go to: www.villagevoice.com/sex-trafficking
- Wall of Shame: CNN, USA Today and other media organizations got fooled
- Super Bowling for Dollars: The great American game is another chance to foist false research on a gullible media
- Beth Schapiro Responds: "We stand fully behind our work"
The congressmen heard testimony from half a dozen nonprofit executives and law-enforcement officials. But the most alarming words of the day came from Deborah Richardson, the chief program officer of the Women's Funding Network (WFN). She told legislators that juvenile prostitution is exploding at an astronomical rate.
"An independent tracking study released today by the Women's Funding Network shows that over the past six months, the number of underage girls trafficked online has risen exponentially in three diverse states," Richardson claimed. "Michigan: a 39.2 percent increase; New York: a 20.7 percent increase; and Minnesota: a staggering 64.7 percent increase."
In the wake of this bombshell revelation, Richardson's disturbing figures found their way into some of the biggest newspapers in the country. USA Today, the Houston Chronicle, the Miami Herald, the Minneapolis Star Tribune and the Detroit Free Press all repeated the dire statistics.
The successful assault on Craigslist was followed by a cross-country tour by Richardson and WFN.
None of the media that published Richardson's astonishing numbers bothered to examine the study at the heart of Richardson's claim. If they had, they would have found what we did after asking independent experts to examine the research: It's junk science.
After all, the numbers are all guesses.
The data are based merely on looking at photos on the Internet. There is no science.
Eric Grodsky, a sociologist at the University of Minnesota who teaches about proper research construction, says the study is fundamentally flawed. "The method's not clean," he says. "You couldn't get this kind of thing into a peer-reviewed journal. There are just too many unanswered questions about the methodology."
Ric Curtis, the chairman of the Anthropology Department at John Jay College of Criminal Justice in New York, led a Justice Department-funded study on juvenile prostitution in New York City in 2008. He's highly skeptical of the claims in WFN's study. "I wouldn't trust those numbers," he says. "This new study seems pretty bogus."
In fact, the group behind the study admits as much. It's now clear it used fake data to deceive the media and lie to Congress. And it was all done to score free publicity and a wealth of public funding.
"We pitch it the way we think you're going to read it and pick up on it," says Kaffie McCullough, the director of Atlanta-based anti-prostitution group A Future Not a Past. "If we give it to you with all the words and the stuff that is actually accurate—I mean, I've tried to do that with our PR firm, and they say, 'They won't read that much.'"
* * *
A Future Not a Past is a product of the Atlanta Women's Foundation, the Juvenile Justice Fund, and Harold and Kayrita Anderson's foundation. To measure the amount of juvenile prostitution in the state, the consortium hired the Schapiro Group, an Atlanta business-consulting operation.
The Schapiro Group's members weren't academic researchers and had no prior experience studying prostitution. In fact, the group was best known for research paid for by the American Chamber of Commerce. The study found—surprise—that membership in the Chamber of Commerce improves a business's image.
The consultants came up with a novel, if not very scientific, method for tabulating juvenile prostitutes: They counted pictures of young-looking women on online classified sites.
"That's one of the first problems right there," Grodsky says. "These advertisers are in the business of making sales, and there's a market for young-looking women. Why would you trust that the photographs are accurate?"
In other words, the ads, like the covers of women's magazines, are relentlessly promoting fantasy. Anyone who has tried online dating understands the inherent trouble with trusting photographs.
Even if the person placing the advertisement is the one in the picture, there's no telling how old the photo is, says David Finkelhor, the director of the Crimes Against Children Research Center at the University of New Hampshire. "How do you know when the pictures were taken?" he asks. "It's not illegal for an 18-year-old who's selling sex to put up a picture of herself from when she was 16."
And if, for the sake of argument, the photos were an accurate portrayal, how do you train those viewing the photographs to guess the correct age?
In fact, you don't.
Before conducting its full study, the Schapiro Group tested the accuracy of its method in a sample of 100 observers. At one point, the 100 observers are described as a "random sample." Elsewhere, they are described as "balanced by race and gender."
These 100 adults were shown pictures of teenagers and young adults whose ages were known. They were asked to guess whether the subjects of the photos were younger than 18.
"The study showed that any given 'young'-looking girl who is selling sex has a 38 percent likelihood of being under age 18," reads a crucial passage in the explanation of methodology. "Put another way, for every 100 'young'-looking girls selling sex, 38 are under 18 years of age. We would compute this by assigning a value of 0.38 to each of the 100 'young' girls we encounter, then summing the values together to achieve a reliable count."
This is dense gibberish posing as statistical analysis.
When the team went on to conduct its full statewide study, it simply treated this 38 percent success rate as a constant. Six new observers were then turned loose to count "young-looking" sex ads on online classifieds sites such as Craigslist and Backpage.
That total count was then multiplied by 0.38 to come up with a guesstimate of how many children were being trafficked.
"This is a logical fallacy," says Steve Doig, the Knight Chair in Journalism at Arizona State University, who reviewed the study at our request. "Consider this analogy: Imagine that 100 people were shown pictures of various automobiles and asked to identify the make, and that 38 percent of the time, people misidentified Fords as Chevrolets. Using the Schapiro logic, this would mean that 38 percent of Fords on the street actually are Chevys."
But the Georgia sponsors were happy with the results—after all, the scary-sounding study agreed with what they were saying all along. So WFN paid Schapiro to dramatically expand the study to include Michigan, Minnesota, New York and Texas. (Georgia's Kayrita Anderson sits on the board ofWFN.) WFN says it would like to have the study ultimately running in all 50 states.
The count of online classifieds featuring "young women" is repeated every three months to track how the numbers change over time. That's the source of the claim of a 64 percent increase in child prostitution in Minnesota in a matter of months.
But that's not how a scientific study is supposed to work, says Finkelhor. "They don't tell you what the confidence intervals are, so these changes could just be noise," he says. "When the Minnesota count goes from 102 to 112, that's probably just random fluctuations."
There's a more fundamental issue, of course.
"The trend analysis is simply a function of the number of images on these sites," Finkelhor says. "It's not necessarily an indication that there's an increase in the number of juveniles involved."
Despite these flaws, WFN, which held rallies across the nation, has been flogging the results relentlessly through national press releases and local member organizations. In press releases, the group goes so far as to compare its conjured-up data to actual hard numbers for other social ills. "Monthly domestic sex trafficking in Minnesota is more pervasive than the state's annually reported incidents of teen girls who died by suicide, homicide and car accidents (29 instances combined); infants who died from SIDS (six instances); or women of all ages murdered in one year (37 instances)," reads the study.
Of course, those other figures are rigorously compiled medical and law-enforcement records of actual documented incidents, so it's not exactly an apples-to-apples comparison.
The police who tally many of those actual statistics—as well as records of real face-to-face encounters with juvenile prostitutes—don't seem to be very impressed by the statistics put forward by WFN. "The methodology that they used doesn't really show the numbers that back it up," says Sergeant John Bandemer, who heads the Vick Human Trafficking Task Force in St. Paul. "We take it with a grain of salt."
* * *
The experts we consulted agreed the Schapiro Group's published methodology raises more questions than it answers. So we went to the Schapiro Group to ask them.
Beth Schapiro founded the Schapiro Group in 1984, starting out mostly with political consulting. The bulk of the group's work, Schapiro says, consists of public-opinion research. In 2007, the group installed its own phone-bank center, and the group's website advertises services ranging from customer satisfaction surveys to "voter persuasion calls."
Counting hard-to-find exploitation victims wasn't exactly in the company's repertoire when it was asked by A Future Not a Past to devise a study on juvenile prostitution in 2007, but Schapiro jumped at the opportunity.
The Georgia studies included efforts to count juvenile prostitutes on the street, at hotels and in escort services, but they also marked the debut of the problematic online-classifieds study that would later be reproduced in other states.
In a phone call this month, Schapiro insisted her study was the first effort to try to scientifically determine the number of juvenile prostitutes—a claim that would likely surprise the authors of dozens of previous studies, several of which are footnoted in her own report.
When we asked Schapiro and Rusty Parker, the leader of the classifieds study, to fill in some of the missing pieces in their methodology, they had a hard time coming up with straight answers. In fact, Parker couldn't remember key information about how he constructed the study. When asked where he got the sample pictures used to calibrate the all-important 38 percent error rate, he wasn't sure.
"It was awhile back," he says. "I forget exactly where we got them from."
Parker was equally fuzzy on how the researchers knew the ages of the people pictured in the control group. "Um . . . I'm afraid I do not remember," he says.
You might say that this is important information. The Schapiro Group has been telling the world it cracked the alchemical code that transforms dumb guesses into hard statistics and that the magic number is 0.38. But the leader of the study can't remember the procedure he followed to get that number.
Neither Schapiro nor Parker had any answers when asked if there was any empirical reason to believe their two critical assumptions: that online photos always represent what the prostitutes actually look like, and that the six handpicked observers conducting the state studies have exactly the same error rate as the initial test batch of 100 random citizens.
Instead, Schapiro beat a hasty retreat, saying the study results shouldn't be read as actual incidents of prostitution.
"We're the first to tell you this is not a precise count of the number of girls being prostituted," Schapiro says. "We make no bones about that."
Of course, a precise count of the number of girls being prostituted is exactly what the statistics are being presented as in the media, in press releases and in Schapiro's own study. When this is pointed out, Schapiro reverses herself.
"Well, yes, these are specific numbers," she backpedals. "And yes, they are hard numbers, and they are numbers that we stand completely behind."
This is the kind of cognitive whiplash you have to endure if you try to follow Schapiro down the rabbit hole. The numbers have the weight of fact and can properly be cited as actual incidents of juvenile prostitution, she insists. But when pressed to justify the broad and unsupported assumptions of her study, she says it is just a work in progress and the numbers are only approximations.
Schapiro's grasp on empirical rigor is such that when asked point-blank to choose between her two contradictory interpretations—estimates or facts—she opts for "all of the above."
"I would square the circle by saying that you can look at them both ways," she says.
Any reporter who had read the methodology of the Schapiro report would have been left with doubts, and any reporter who followed up would probably have been treated to the same baffling circuit of non-answers. The fact that the study's findings continue to be rebroadcast in news outlets across the country suggests that not one reporter has bothered to read the study about which he or she is writing.
"You see this kind of thing a lot, unfortunately," says Rick Edmonds, a media-business analyst for the Poynter Institute who writes frequently about statistics. "The kind of skepticism that reporters apply to a statement by a politician just doesn't get applied to studies."
The Crimes Against Children Research Center's Finkelhor says he understands the pressure on reporters to cite figures when they're writing about juvenile prostitution. But, he says, it's something they need to resist because despite what groups such as WFN would have you believe, there simply are no good statistics.
"You have to say, 'We don't know. Estimates have been made, but none of them have a real scientific basis to them,'" Finkelhor says. "All you can say is, 'This is the number the police know about, and we think there are more than that, but we don't know how many more.'"
* * *
In her own online photos, the woman who commissioned the Schapiro Group study looks to be in her 50s, with blue eyes, graying hair and a taste for dangly earrings.
Kaffie McCullough first approached the Schapiro Group about conducting a study of juvenile prostitution in Georgia when, as director of A Future Not a Past, she realized that having scientific-sounding numbers makes all the difference in the world.
In early 2007, McCullough approached the Georgia Legislature to ask for money for a regional assessment center to track juvenile prostitution.
"We had no research, no nothing. The legislators didn't even know about it," she recalls. "We got a little bit. We got about 20 percent of what we asked for."
Later that year, the first Schapiro Group counts were made, and when McCullough returned to the Legislature the following session, she had the study's statistics in hand.
"When we went to the Legislature with those counts, it gave us traction—night and day," she says. "That year, we got all the rest of that money, plus we got a study commission."
McCullough touts the fund-raising benefits of the study whenever she can. Since the Schapiro study was picked up for replication nationwide by WFN, McCullough has acted as a sort of technical consultant for state groups as they debate whether to invest money in the project. Whenever she's asked, McCullough tells the local groups the money they spend will come back to them with hefty dividends.
"I would say, 'The research costs money, but we've been able to broker—I don't know what it is now, I think it's more than $1.3, $1.6 million in funding that we never would have gotten,'" McCullough says.
She initially maintained that she stands by the Schapiro Group study, in part because she has been told that "it is the same scientific methodology that science has been using for a long time to measure endangered species."
But when pressed on whether she really believes that counting Internet photos is reliable, she grants the sex-work industry isn't exactly the gold standard of truth in advertising.
"That's absolutely correct," she says. "That's part of how that business operates: It's a bait-and-switch."
And given the tricky nature of the photographs, she admits that counting pictures isn't exactly a precise way to measure juvenile prostitutes.
"I can't guarantee that any picture that four of those six people said looked young—that may not be the girl you'd get if you called up," she concedes.
Asked if she has any reason to believe the six observers in the study have the identical 38 percent error rate as the 100 random citizens who were the initial test subjects, she allows that it might be worth revisiting that question.
The basic truth is that the study exists in service of the advocacy, and if news outlets present the Schapiro Group's numbers as gospel, it certainly doesn't hurt the advocates' cause.
Admitting that there isn't any authoritative scientific count of juvenile prostitution, as Finkelhor recommends, isn't an option in McCullough's book. She recalls an early presentation she made in Nebraska, when a politician gave her a piece of advice that stuck.
"He said, 'If you-all as a movement don't start having numbers, you are going to lose the money,'" McCullough recalls. "'How can you justify millions of dollars when there are only hundreds of victims that you're actually serving?'"
* * *
Editor's Conclusion: Last week, on March 16, the drumbeat continued in the U.S. Senate with a briefing on domestic minor sex trafficking that featured Hollywood actress Mira Sorvino and the startling statistic that 100,000 children are trafficked for sex annually in America.
Trafficking, in labor and sex, became a defining issue in the administration of President George W. Bush. But as an investigation by the Washington Post in 2007 revealed, victims in the sex trade were difficult to come by.
Today, advocates have shifted media attention to allegations of trafficking in children.
But facts to suggest a plague of underage perversion simply do not exist, despite claims to the contrary.
In a deficit-obsessed Congress, there is a long line of those seeking tax dollars to raise awareness of trafficking: government agencies, nonprofits, religious groups, the well-intentioned, as well as abolitionists opposed to everything from pornography to adult services.
It is no surprise that some seek to use children as a wedge.
If you like this story, consider signing up for our email newsletters.
SHOW ME HOW
You have successfully signed up for your selected newsletter(s) - please keep an eye on your mailbox, we're movin' in!
Responsible parties prosecute predators and rescue victims. Not everyone with a microphone is responsible.
The challenge of keeping children out of the hands of exploiters is real, but solutions are not clear in an atmosphere of hyped hysteria.
This article appeared in print as "Weird Science: How a bogus child-sex-trafficking study fooled some of the most respected media outlets in the country."