Lies, and Damn Lies
With statistics. Fox News (who else?) has repeatedly put up a graphic showing the unemployment rate since December 2007.
Now if you’re like me you’re immediately suspicious. Not just because it’s Fox, but did unemployment really increase at a steady rate for the last two-and-a-half years? What a strange statistical fluke that would be. If it were true, which of course it isn’t. MediaMatters published a couple of these graphs in a story today, graphs I’m reproducing here, and took the news angle. I’m going to try here to explain what techniques they’re using to gull their already gullible viewers into believing something that just isn’t true.
Take a closer look. The horizontal axis has four more-or-less evenly spaced points. But the time frames aren’t really even at all. The first two are 9 months apart; the second two are 6 months apart; and the last two are 15 months apart. So even though the middle two dots represent the shortest time frame, they are the farthest apart. And the opposite is true for the last two dots.
So the horizontal axis wrong; how about the vertical? The first gridline is a little bit above 7M, so let’s say it’s 7.5M. The top line seems to be right at 15M. So there are three steps in between, each at a distance of (15M – 7.5M) / 3 from the one below. So a vertical scale, if there was one, would look like:
15M
12.5M
10M
7.5M
At least, that’s what it would do if you were evaluating the data honestly. Which they’re not.
The first dot we’ll give them: 7M is a bit below 7.5M. The second dot, 9M, is reasonably below the 10M line as well. But the third dot, 13.5M, is right on the line that should be indicating 12.5M.
See what I’m getting at? The March 2009 number should be much higher, thereby steepening the rise between September 2008 and March 2009 (a rise that should be steeper already because the points are just six months apart and so should be closer); and making the rise between March 2009 and June 2010 less steep than it already should have been — if you stretched it out proportionately to 15 months.
Going back to the first alarm, the smoothness of the increase — which is already debunked because the time frames have to be corrected horizontally and the data points are placed vertically wherever the hell they thought would fit their dogma. Are there no data available for any points in between?
Well of course there are. These things are measured all the time. If you go month-by-month proportionately, with a evenly spaced vertical axis, a much different story emerges:
That whole rise between the beginning of the chart and March 2009 (the third point on the Fox chart) was in fact a sickening up-the-roller-coaster hill ride into dangerously bad territory.
Fox would have you believe that since then, that stomach-wrenching trip up the hill has continued unabated. Well, obviously it hasn’t. It’s been a little up, a little down, and the latest trend is that it’s better than it was at its worst, and headed downward again.
“We report, you decide.” Does anybody really believe that any more? “We decide, you stare glassy-eyed at the screen” is perhaps more accurate.