The mass media often misuse statistics when they report about such subjects as health, science, education, and the economy. Media outlets seem to be more concerned with pushing agendas than accurate reporting. Even if there is no agenda, emotionally vivid language in headlines and teasers create and maintain interest in the story.
I’m not writing media criticism, though. This post is as much about research and any of my others. Not many Americans know much about statistics. And journalists have no more training in statistics than most of the rest of us.
Badly understood and misused statistics also influence public policy and legislation. Instead of playing the blame game, the general public ought to learn enough about statistics to view discussions that depend on them with respectful skepticism. Earlier I made a similar point about the necessity of a wait and see attitude toward conflicting scientific studies.
Statistics has a reputation of being an extremely boring discipline to learn. That’s no excuse for those of us who are not statisticians to keep certain basic concepts in mind, including
- correlation vs causation
- relative vs absolute risk
- scales and orders of magnitude
- margin of error
Murders vs mass murders
I was reminded of the importance of statistical literacy by a graph one of my Facebook friends posted. It occurred to me that people could look at that graph and easily draw opposite conclusions.
Should we think, “Mass killings are less than 1% of all homicides and always have been. What’s the big deal?”
Or perhaps, “The rate of mass killings has nearly doubled since 1980! The sky is falling!”
Well, neither, really. For one thing, it’s one graph, one bit of data. All by itself, without some kind of context, it doesn’t mean much at all. The fact that it can so easily lead to emotionally opposite conclusions simply illustrates the truth of an old saying: A text without a context is a pretext.
Here’s an obvious related question: what has happened to the murder rate during the same time period? Have both trends gone in the same direction? Here’s another graph from the same discussion:
Murders rose for a while after 1980, then plunged. Fortunately, the discussion fizzled out before anyone started drawing wild conclusions. Notice that both graphs are based on statistics from different sources. Both graphs summarize the statistics, and anyone using the statistics behind the graphs will have to take notice of differences in their scope and how they were compiled.
The cost of raising children
Ignorance of statistics in social media is far less serious than in the news media. People watch the news or read newspapers and magazines to find out what’s happening in the world.
It’s a serious public disservice when the people who write the stories don’t know or don’t care that they’re completely wrong through misinterpretation of statistics
Statistics from the Department of Agriculture show that the cost of raising a child from birth to age 17 has increased 25% over the past 10 years. CNBC breathlessly announced that the cost had “soared.”
In fact, 25% over 10 years amounts to an annual inflation rate of 2.26%, compounded. Does anyone remember the years of double-digit inflation? Stop to think of the increase of the cost of childrearing during a decade when inflation never went below 10%!
Words like “surged” or “soared” are unjustified in this context. Perhaps journalistic deadlines prevented anyone from taking time to think it through. That makes it all the more important for viewers to recognize when the reporting of the numbers makes no sense.
Hot dogs and cancer
Sometimes groups with an agenda misuse statistics deliberately. A group called Physicians’ Committee for Responsible Medicine took out an ad that claimed that hot dogs cause cancer.
CNN chose to report on the ad. It did mention that the Committee is “an animal rights group that wants us all to be vegans.” That should have been reason enough for the network to be suspicious.
Unfortunately, the writer neglected to note (or perhaps even try to find out) that the American Medical Association, among other organizations, had looked into the links between nitrites and cancer. These organizations concluded that all the studies were flawed and that available evidence did not confirm an association.
This incident is more serious than getting into a lather instead of thinking through the meaning of the numbers. Someone should have checked the facts before running a story based on a phony study.
Misuse of Statistics a National Problem / American Statistical Association
Spinning Heads and Spinning News: The Use and Abuse of Statistics in the Media / Mathematical Association of America