STATS ARTICLES 2004
Tampa Tribune Attacks Scientific Evidence
with Unscientific Study
June 22, 2004
S. Robert Lichter
Paper fans local health fears with 119th story on local industry.
Some media health scares come from misinterpreting the scientific evidence — think of cancer from high-tension power lines or alar on apples. Others come from replacing science with speculation — think of brain tumors from cell phones, or auto-immune disorders from breast implants. But rarely does a newspaper actually try to undo the existing scientific evidence with its own unscientific study.
Yet that’s what the Tampa Tribune just did in a major feature on health problems allegedly associated with a phosphate rock mining and processing plant. In a front-page feature on April 18 that included multiple stories, sidebars, and a large graphic display of its own health survey findings, the paper portrayed the Coronet Junction area of Plant City, Fla. as a sick community.
This was hardly news to readers of the paper: Since July 2003, the Tribune had published 118 stories about Coronet Industries, the owners of the phosphate plant. (By contrast, The St. Petersburg Times published only 30 stories during the same time period.) Coronet had been accused of poisoning local residents in a lawsuit by California-based law firm Masry and Vitito (who have employed the controversial publicity magnet Erin Brokovich as a researcher on the case). And, in an equally controversial move, the lead reporter on the first wave of stories on the plant, Deborah Alberto, left the paper in the Fall of 2003 to work for two additional law firms collaborating in Masry and Vitito’s lawsuit.
The Tribune’s managing editor, Donna Reed, admitted that the timing of Alberto’s re-employment was “regrettable,” but she insisted that it did not diminish the accuracy and fairness of the paper’s reporting on Coronet. (Reed also came in for criticism in the St. Petersburg Times for failing to disclose that her husband was a former employee of Coronet.)
While both issues provoke typical ethical questions about perceived conflicts of interest, transparency, and credibility, the Tribune’s major journalistic failing was an act of commission rather than omission.
The Tribune’s survey of 250 households near the Coronet Industries plant found that at least one resident in 25% of households had cancer, 46% reported joint pain or spinal problems, 24% had observed “unusual health problems” with pets or livestock, 9% had children with “delayed mental or physical development,” and so on through nine categories of ailments. The lead article didn’t leave much doubt about the paper’s purpose in conducting the study, or its sympathies. It began:
The old gray buildings on the edge of town are a grim specter, heavy and forbidding, like the fear that has enveloped this rural community... To some the fear is as tangible as the neighbors and relatives who have been diagnosed with cancer or the pets that have died in peculiar circumstances.... Yes, many residents believe that there’s something wrong here, and they suspect the culprit is down the road at Coronet Industries.
So, for all the lurid language (which goes on at length), is this report a public service, pointing out a health hazard that will force officials to protect local residents from the toxic byproducts left by a callous corporation? Actually, no. The county, state, and federal governments have all looked into this, and no one has found any evidence of major environmental health problems, much less problems that can be traced to this particular plant.
The inquiry began in earnest last year, when citizen complaints about the health hazards of pollution from the plant led to the discovery of either arsenic, cadmium or radium in levels exceeding the state-mandated maximums in 28 of 146 residential wells that were tested. All these substances are byproducts of phosphate processing. This triggered an investigation into whether there was an increase in the incidence of eight cancers associated with long term exposure to these substances – cancer of the lung and bronchus, liver, bladder, kidney, prostate, skin, breast, and bones.
Local, state, and federal officials examined the air, soil and water to measure contamination levels; they analyzed the urine of residents to measure any residue of contaminants in their bodies, and they conducted an epidemiological survey to compare the incidence of these cancers in the immediate area with the rest of the county and the state.
The result? No elevated health hazards. Nada. A clean bill of health so far, although the investigation is continuing. Or, in more scientific language: “there was no statistically significant increase in the number of observed cancer cases in the community;” “air in the vicinity of Coronet Junction appears to be generally the same as other areas;” “all values [of the soil samples] are well within standards considered safe for play and daily activities for private residences;” “the measured exposures to these chemicals [from the urinalysis] pose no apparent public health hazard.” As for the contaminated wells, “identification of a contaminant ... does not necessarily mean that exposure will cause illness.... the [agency] categorizes the risk to public health as a no apparent public health hazard.” (emphasis in original)
So what’s the story — literally? Where’s the news, when careful government investigations turn up nothing? Well, for the Tribune, the story is that residents have ailments, and somebody needs to document them. Hence the survey, whose bold graphics visually reinforce the impression of very high rates of disease and unidentified maladies. Unfortunately, none of this can prove anything that might help clear up what one headline describes as a “Medical Mystery.”
Oddly, the Tribune recognizes what it calls the study’s “shortcomings: [it] wasn’t designed by an epidemiologist, the responses are anecdotal, the ailments unconfirmed.” An American Cancer Society official is quoted as saying “the data aren’t going to be comparable in any way.” Right on all counts. So why do it? It was “an exercise in journalism, not science. We wanted to know what ailed people, not what caused it [sic].... It provided a glimpse at the health problems experienced by people in the 250 households...”
It also provided a glimpse of how not to conduct a survey. It was administered by reporters who told residents they were “investigating reports of possible health problems that state and federal agencies say may be linked to contaminated air and water in neighborhoods bordering the Coronet phosphate processing plant.... [we would] appreciate your assistance in helping us assess the potential threat to the community.”
Any reader who has ever studied survey research is now cringing over what social scientists call the “contamination” of this survey by prompting or leading respondents. Residents are primed to think in terms of pollution from the Coronet plant, which is sufficiently threatening to the community as to worry the government and the local media. This can dramatically affect responses, especially since the categories of illness are so vague, e.g., “delayed mental or physical development,” “fertility problems,” “unusual health problems.”
Given an opportunity to search for ailments, people tend to find them, particularly against the backdrop of an ongoing controversy, which already has them worried. (Indeed, some residents have filed lawsuits.) The literature is filled with instances in which self-reports of illness increase due to media coverage of alleged health risks, even when they prove unfounded. That is why it is not the ailments but their causes are crucial to this debate.
It’s also puzzling that a survey dealing with the Coronet plant should ask about so many ailments that have nothing to do with the contaminants at issue. For example, it’s not clear what proportion of the “unusual” health problems of pets that residents report could potentially be linked to pollution.
Similarly, it’s pretty scary that one out of four households have been touched by cancer. But there are over 100 types of cancer, only eight of which are relevant to the controversy. The inevitable result of such survey questions is to produce an inflated impression of rampant illnesses, which are somehow associated with the plant. (Otherwise, why would you ask about them?)
The Tribune thinks it has produced “a good baseline, a snapshot of the community...” But it’s a baseline that can’t be usefully compared to anything, because the snapshot was taken through a distorted lens. The entire project betrays a misunderstanding of how science informs health reporting. It is not enough to say that you are doing “journalism, not science,” and then proceed as if the lessons of science are irrelevant to reporting. Sometimes good journalism requires you to understand and properly apply scientific research methods. Otherwise, you end up with both bad science and bad journalism.
Dr. Robert Lichter is President of the Statistical Assessment Service