Suspicious Statistics

Paul Krugman has a chart from the U.S. Department of Labor, Bureau of Labor Statistics showing the ratio of civilian employment to population (EMRATIO) on his Conscience of a Liberal blog today.

"The chart above shows the employment-population ratio,  the ratio of employed Americans to the adult population. By this measure it’s been a weak economy all along — and now it’s falling off a cliff."

One of the first things that strikes me about this graph is the range of employment percentage (Y-axis). If the chart showed the percentage fluctuation in the context of zero to one hundred percent employment, the variation would look pretty minimal. And why are we looking at only ten years?

So I went out to the Federal Reserve Bank of St. Louis website to see what else I could find.  Sure enough, here's the same chart, but showing employment since the 1940's.

This picture seems to tell a different story! But now you have to take into account that women were joining the workforce in greater numbers during this period.  (In fact, given the relatively narrow percentage range, I surprised the increase isn't greater.)

Here's another period to look at—the last two administrations.

What do you think? Do Clinton and the Democrats get credit for the first half of the period? The Republican congress? Do Bush and the Republicans get the blame for the last half? Whose fault were the two burst bubbles?

So, I started this suspicious of Dr. Krugman for cherry-picking data to support his position. After thinking about it, I'm not so suspicious of him. I'm suspicious of statistics! It's not trivial to avoid making them say what you want them to say.