Today a headline that caught my eye prompted me to go to the National Snow and Ice Data Center to take a look at what’s been happening with Arctic sea ice lately—as far, anyway, as one can trust the data there. (There are reasons to take it all with a grain of salt.)

Here’s a graph they offer depicting changes in Arctic monthly sea ice extent from February 1979 through February 2010.

Looks pretty alarming, doesn’t it?

But look over at the Y-axis (left). Notice where it starts? At 14.0. That’s an immediate tip-off that this graph is designed to exaggerate the magnitude of change and the rate of trend.

In the early 1990s, I worked with the late Dr. Julian L. Simon, a professor of statistics, economics, and business, editing the monumental book The State of Humanity. The book contains 58 chapters by 60 authors, all world-class experts in their fields, including some Nobel Prize winners in their fields (e.g., Robert W. Fogel, in economics).

One aim of the book was to show trends in human life, health, and mortality; in standard of living, productivity, and poverty; in natural resource supplies and prices; in agriculture, food, land, and water; and in pollution and environmental quality. We sought to show the trends on time scales as long as possible.

While working on that book, Dr. Simon taught me a lesson in how to present statistics that’s stayed with me ever since—and that is the cause of a great deal of frustration on my part as I watch environmentalists (and others) misuse graphic presentation of statistics to manipulate public opinion. The lesson? Whenever possible, in dealing with absolute numbers, use a zero baseline; and when using percentages, use a scale of 100—whether from 0 to 100, or -10 to +90, or -50 to +50. Otherwise, the picture can grossly misrepresent reality, even when it accurately depicts the raw numbers.

Such gross misrepresentation is what the NSIDC achieves with the graph above. By starting the Y-axis at 14 instead of 0, it creates the impression of a much steeper decline in sea ice extent than really occurred.

The next two graphs depict roughly the same raw numbers as those shown in the linear trend (straight) line in the graph above. I eyeballed the start and end points and then created a linear trend between them. The first one uses the same Y-axis as the one above: 14-17:

The second uses a 0 baseline but the same maximum of 17:

The lines in the two graphs depict exactly the same numbers. But which one looks scarier? Obviously, the first.

These comparative graphs illustrate precisely the point nineteenth-century British Prime Minister Benjamin Disraeli once made—as quoted by Mark Twain: “There are three kinds of lies: lies, damned lies, and statistics.”

Keep that in mind when you see trend lines, especially when they’re offered in support of any kind of public policy. If they don’t have a zero baseline, or if they depict percentage changes but not on a 100-point scale, conjure up an image for yourself of what they’d look like if they did. You’ll invariably find that the result are far less drastic—and less alarming (or reassuring) than they appeared when presented improperly.