How to Chart Untruths

In his New York Times column today, Paul Krugman published a damning, evidentiary indictment of the Bush administration’s wantonly optimistic — and highly inaccurate — jobs forecasting. It’s wonderfully concise, to the point, and heavily reliant on a powerful graphic that charts predictions that the White House has made for “nonfarm payroll employment” in 2002, 2003 and 2004 against the actual data provided in a joint report from the Economic Policy Institute and the Center on Budget and Policy Priorities.

Below: Chartjunk. This illustration from today’s Times undermines a valid point with poor formatting.

On closer inspection though, I was disappointed to find that the chart is actually somewhat misleading. The y-axis, which charts nonpayroll employment in millions — with a tick mark every five million — begins at 125 million. The Bush administration’s estimates are overzealous by as much as 6.9% (though, to be fair, this year’s figure is off by a more reasonable 1.9%).

Krugman Chart

Even two to seven percent is still a pretty drastic miscalculation, but the way the chart is drawn, the error looks more on the order of a forty percent miscalculation. In order to modify this chart to be somewhat more accurate, one would need to add fourteen more tick marks along the y-axis.

Truncating data in this way, especially to save column inches, is a common practice, but in this case it feels as if Krugman pulled his punches a little bit. His point is a good one, but it unfortunately falls prey to the economics of page layout and forces a distortion that could be easily misinterpreted. I’ve always felt that the best ammunition against Republican obfuscation is a clear, objective reading of the facts, and this is almost, but not quite, such a reading. I’ve also long felt that a case can be made against the Bush administration using the tools of information design — a clear-cut presentation of the data that shows the fallacies in their logic. Again, close but not quite there.



  1. You think the chart is “somewhat misleading” because of a lack of physical space on the printed page? I think that’s crap. Krugman is a well known Bush-hater who’ll do whatever he can to bring down the Bush Administration. The data is misleading on purpose, as it helps to further serve Krugman’s own “obfuscation”. It’s a common practice to warp statistical data this way, and it’s very easy to do.

    The NYTimes has a long way to go before I’ll once again think of them as a credible news organization. Krugman’s sanctioned bullshit doesn’t help me get there.

  2. The op-ed column in question is very short – it’s all about the chart.

    A picture is worth a thousand obfuscated words.

Thank you! Your remarks have been sent to Khoi.