We are inundated with warnings, advice and studies that deeply affect how we live. Yet, many are poorly understood, if not misleading.

It's commonplace to note, for instance, that two-pack-a-day smokers face 20 times the lung cancer risk of non-smokers. What does that really mean to someone contemplating taking up the habit?

On average, around 10 nonsmokers per 100,000 are expected to get lung cancer. At 20 times the risk for heavy smokers, about 200 per 100,000 people will contract the disease. That's certainly a big statistical difference. But it still means that 99.8 percent of two-pack-a-day smokers will avoid lung cancer.

No one ever reports the actual incidence of smoking and lung cancer in this way. It's a safe bet that most people think it's a virtual certainty smoking will generate the horrible disease. In a 1995 federal survey, most high schoolers said that smoking was far more dangerous than using barbiturates or downing up to five beers at a time.

The truth is that smoking kills almost no one under age 35. On average, even the 400,000 people said to die of "smoking related" deaths every year live to be 72 years of age, compared with the national average of 76. Drugs, alcohol, driving and even high-risk sports slaughter incomparably more young Americans than smoking. But diverted by constant media messages funded by public and private health scare advocates, our attention is tragically diverted to far less significant problems.

I do not smoke, nor do I allow my children to smoke. So why should I care about antismoking hyperbole?

The reason is that gross factual distortion, even in the name of laudable causes, profoundly devalues public debate. Policy priorities are shaped by ad campaigns, buzz words and media bias, not careful reflection. I cringe when my son, after being fed an unrelenting diet of smoking horror stories at school, has his sensibilities warped to the point where he thinks cigarettes are an evil commensurate with violent crime. Our perceptions of real risk, our ability to choose to live in ways that best achieve our goals, can be dangerously obscured.

All of this went through my mind last week when the national media dutifully gave top billing to the latest version of the American Lung Association's annual "State of the Air" report. The study purports to evaluate America's air quality region by region. It warns that "dirty air" puts millions of people nationwide -- especially the elderly and children -- "at risk." Radio, newspaper and TV outlets breathlessly pronounced that Southern California, as usual, had the nation's dirtiest air. A local PBS station seized the moment to contemplate the profound "contradiction" of a Los Angeles sunset: something "so beautiful, yet toxic."

Few, of course, bothered to look at the numbers.

The report is based on three years of raw data collected from air monitoring stations in about 600 U.S. counties. It supposedly tallies the number of days each county exceeds certain federal ozone standards. Points are assigned based on the severity of each such exceedance. Areas with perfect air quality -- no exceedances -- are given an "A" rank. The rest receive grades of "B" to "F."

One problem with the study is that it doesn't measure "days" of bad air at all. A bad "day" means that ozone counts rose, no matter how briefly, at least once above the federal standards during an eight-hour measurement period. The greatest exposure to bad air that can possibly be measured by the data is just eight of 24 hours. In almost every case, moreover, ozone will rise above acceptable levels for just a fraction of the eight-hour measurement period. Bad air exposure times are dramatically overstated.

The report's rankings also correspond with extremely minute air quality variations. Orange County, for example, is assessed an "F" grade because it supposedly exceeded the ozone standard four days per year. Rural Sonoma County earns a "D" ranking with 2.5 exceedances. Mountainous, isolated Inyo County is graded a "B" region with an average of 0.7 exceedances per year.

These grades reflect absurdly small differences. Assuming each exceedance lasted for the full eight-hour monitoring period -- which is highly unlikely -- at the very worst ozone counts were actually within applicable standards 99.6 percent of the time in Orange County, 99.8 percent of the time in Sonoma County, and 99.9 percent of the year in Inyo.

There is no data whatsoever that even remotely suggests that measurable health risks correspond with such infinitesimally different ozone exposures. Most environmental health studies assume truly massive exposure to identify health outcomes, like the infamously large doses of saccharine that were required to produce cancer in rats. Even then, many are hard-pressed to identify a corresponding danger.

But what about the truly bad areas? Aren't they hotbeds of lung disease?

According to the report, Los Angeles, Charlotte, Atlanta and Knoxville each had about 40 bad ozone days per year. Houston had about 60, San Bernardino 115, and many San Joaquin Valley towns over 70 bad ozone days. These are certainly unwelcome results. Even so, ozone was below federal limits in areas like Los Angeles at least 96.3 percent of the time. Fresno, Houston and San Bernardino were in compliance at least 90 percent of the time.

Given such relatively small adverse exposure times, it's not surprising that the report doesn't attempt to link specific health risks with a corresponding ozone exposure. The fact is that California boasts a lung disease death rate nearly 20 percent lower than the national average, virtually identical with that of Minnesota, a state that ranks among the least ozone-impaired regions of the country.

Unquestionably, factors other than ozone counts affect lung disease. But such nuance is precisely what's absent from the air quality report. It is very important to understand and consider the costs and benefits of various life choices. To do that, we need to demand far more honesty and integrity from the health advocacy groups that flood our unthinking media with frightening, but strikingly incomplete information.