(Via Iron Monkey. Original link recommended)
Step 1. Start with data that shows a positive trend, like this DOT highway fatality data that shows that the traffic fatality rate is at the lowest level ever recorded. This article states the correct conclusion at the top:
The fatality rate on the nation’s highways in 2003 was the lowest since record keeping began 29 years ago, the U.S. Transportation Secretary Norman Y. Mineta announced today. The number of crash-related injuries also dropped to a historic low in 2003.
“America’s roads and highways are safer than ever,” said Secretary Mineta. “The decreasing number of traffic fatalities and record low death rate on our roads shows that we are headed down the right road – one that leads to a safer America.”
But don’t worry, we can find ways to make this data sound scary, as if it means just the opposite.
Step 2. Change the comparison. The fatality rate went down, but we can still make it sound like it went up by comparing absolute numbers (instead of percentages) against some past year.
42,643 people died in traffic accidents in 2003, an increase of 1032 deaths compared to 1999.
(The absolute number can go up even when the rate goes down, because of increasing population and increasing miles driven.)
Step 3. Find some geographic area where things got worse. The fatality rate went down on the whole, but that doesn’t mean it went down everywhere. There are probably some areas where it went up. Find one of those areas and comment on it.
In the District of Columbia, 20 more people died compared to the previous year.
Step 4. Change to percentages if that sounds scarier. Since there were so few traffic fatalities (47) in D.C. in 2002, an increase of 20 is large if stated in percentage terms.
In the District of Columbia, the fatality rate increased by 43% over the previous year.
Step 5. Always round up. Why not make 43% sound even larger?
In the District of Columbia, the fatality rate increased by nearly 50% over the previous year.
As an added bonus, some people who are math-challenged will think that an increase of 50% means that the rate “doubled.” Let them think that.
Step 6. Find some category that got worse. Even though the fatality rate as a whole went down, there is probably some category of accident that increased. Quote that part.
SUV rollover fatalities increased 6.8 percent from 2,471 to 2,639, even as SUV registrations increased 11 percent.
Step 7. Edit to remove context. Oops, that isn’t really worse, it’s better. The second half of that sentence explains what is really going on: there were more SUV rollover fatalities because there were more SUVs on the roads. Since the number of registrations increased more than the fatalities increased, it means there were actually fewer rollovers per SUV on the road. But that doesn’t sound scary, so remove that part, and only leave in the part that makes it sound like things have gotten worse.
SUV rollover fatalities increased 6.8 percent from 2,471 to 2,639.
Step 8. Use “slippery slope” arguments to imply that small changes now are only warnings of much bigger disasters to come.
If these increases continue, eventually 100% of SUVs will roll over and kill their drivers, and everyone in the District of Columbia will die in a car crash. If road speeds keep increasing, in the future it will be common for drivers to speed through school zones at 200mph.
Step 9. Add a scary headline. The headline is the first thing people read, so it will color their interpretation of everything that comes afterwards.
OUR DEADLY HIGHWAYS: HOW SOON WILL YOUR CHILDREN DIE ON THE ROAD?
Conclusion: This example is hypothetical and meant to be a humorous illustration of how to scare people by creatively interpreting statistics. But look for this type of analysis “in the wild” and you just might find it.