What are the characteristics of junk science? Any one of the following should raise a red flag, says Steven J. Milloy, executive director of The Advancement of Sound Science Coalition:
- One-study wonders. "Study shows laxatives cause cancer." "Selenium prevents cancer, new study says." Can these headlines be right? Unlikely. No one study proves anything. Science is not a quick-and-dirty one-study endeavor. Scientific knowledge is gained slowly and incrementally over time.- Science by press conference. Science occurs in labs, not press conferences or newsrooms. Simply put, headlines aren't science. Dan Rather, Peter Jennings and Tom Brokaw read script off teleprompters; they aren't Nobel laureates. The Washington Post and New York Times are newspapers, not peer-review science journals; "60 Minutes" is television entertainment, not a science-on-video series.
- Causation. Junk scientists love to say that one thing "causes" another. But causation is usually more complex than a junk scientist would have you believe.
- Body counts. "Smoking kills 400,000 people every year." "Obesity kills 300,000." "Air pollution kills 60,000." But those figures are not a simple as they seem, often ignoring other complex issues.
- Weak association epidemiology. Studies reporting relative risks on the order of 3.0 and less are generally not reliable. Epidemiology is most useful is identifying large risks of rare diseases. For example, studies of heavy smoking and lung cancer report a relative risk of about 20; those of aspirin and Reye's syndrome in children report a relative risk of 35.
- Modeling. Many try to use computer-based mathematical modeling to predict how systems will react under various conditions. Depending on the complexity of the system, modeling may work to some degree. The problem, however, is that some modelers think they can model anything regardless of complexity. Systems like global climate, ecosystems and higher-order living organisms (like humans) are probably too complex to be represented by mathematical