Saturday, August 6, 2005

Newspaper lies, damn lies or statistics? Seems like just yesterday I was citing the Atlanta Journal-Constitution as a good example of something. Today it's serving as a bad example. It's that versatility that makes me love newspapers. Thanks to bloggers Say Uncle and No Quarters for pointing out this editorial, particularly its use of statistics.* I needed a good "bad example" to use in a "numeracy" and fact-checking lecture to  journalism students this fall. Here's the key passage from the editorial: "Like many other employers, ConocoPhillips bars its employees from bringing firearms onto the work site, even if the weapon is kept locked in the employee's car in the workplace parking lot. It's a wise move, given that 487 people were shot to death at work in 2003, the latest year for which numbers are available. Those gunfire fatalities accounted for more than three-quarters of all the homicides on work sites that year." NRA members aren't the only ones who should be annoyed at this use of statistics, which quickly  prompted skeptical bloggers to find smaller "workplace shooting" numbers elsewhere and run them under headlines like "Why do newspapers lie?" I looked at it from the other angle: I trust professional journalists not to make stuff up and to double-check impressive statistics that arrive in partisan reports or press releases. The AJC is a substantial paper, not some supermarket throwaway, so where did that 487 come from? Did someone check it? If I were the paper's fact-checker, could I find it? Rather than Google my way through a maze of blogs and partisan websites, I headed for major  sources of federal statistics. Within a few minutes I confirmed that the number came straight from the U.S. Bureau of Labor Statistics (see the PDF report linked here). Accuracy is one thing, but the statistic is still misleading as used in that editorial. Why? The number is a correct answer, but to what question? If you read carefully, you don't even need to check the stats to see that the number was never relevant to an editorial about gun-toting employees. Notice the wording, "487 people were shot to death at work." As the BLS report shows, the 487 total apparently includes every police officer, cab driver, bank teller, pawnbroker and quicky-mart employee shot and killed "at work," whether the shooter was a fleeing criminal, disgruntled customer, armed robber -- or a well-armed angry employee. That last (smaller) category is the number that would fit the Atlanta editorial. Maybe somewhere in the BLS Injuries, Illnesses and Fatalities website there's a report broken down by  "shooter" categories. I didn't see it during a quick Saturday morning visit; and I didn't call the bureau's PR folks for help sifting through the stats -- which is something a working professional reporter or editorial writer should have done instead of just throwing readers the biggest number in sight. (Research tip: If I hadn't heard of the BLS, I could have reached its site through the Department of Labor at www.dol.gov or the more general U.S. government gateways,  www.FirstGov.gov and www.FedStats.gov) The writer's alternative: "When in doubt, leave it out." In this case I doubt that bigger or smaller numbers would change readers' minds about issues like gun ownership, workers' rights or employer rights. Badly-chosen or misleadlingly-applied numbers certainly don't help... especially when there are scores of conscientous, skeptical and interlinked bloggers out there willing to keep professional journalists on their toes. Moral for journalism students: Read carefully. Make sure of the questions numbers answer, as well as the answers. Check your facts. Check again. * About the headline: The often-quoted phrase is usually "There are lies, damn lies -- and statistics." 2:50:35 PM    comment []

© Copyright 2008 Bob Stepno. Last update: 7/19/08; 1:07:48 PM.