Return to index page

December 17, 2002

Damn Lies

Warning: rant about graphing data to follow!
Check out today's entry in about Bush's plan to reduce that oh-so-odious tax burden on the rich, as reported in the Washington Post. The Seattle Times also picked up this one, and attempted to liven the debate with an illustrative graphic. Of course, they screwed it up totally. The 1% bar is already represented by the 5% bar, and they missed out the entire middle class (60-95% quantile of income)! (Hint to graphic artists: if you're going to draw a bar chart of percentages, make them add to 100%.) I re-did the chart in Excel, and you can see that a real chart tells a very different picture: actually, it's the middle classes that carry the greatest share of the tax burden, however you look at it. Poor Tufte must be rolling in his grave (I heard he was killed when one of his sculptures toppled on him.)

Posted by jay at December 17, 2002 03:57 PM | TrackBack
Comment spammers: see our Unauthorized Advertising Policy and rates

Check out the last quote on that "sculptures" link-- you were quoted before you were even born! Now that is a trick.

Posted by: jay on December 17, 2002 05:46 PM

Taling about damned lies and Tufte rolling over in his grave! You claim that "...actually, it's the middle classes that carry the greatest share of the tax burden, however you look at it. " Your "corrected" chart splits the top 5% into two groups to minimize their contribution. A cursory examination of your chart shows that if you combine your final two groups and display the top 5% as a single unit, that 5% pays 56.1% of the total taxes. So tell me again who pays the greatest share?

Posted by: Dan on November 5, 2003 09:23 AM

The division into society into 1%, 5% and 60% quantiles was done in the original article, so we have to stick with that. My objections to the original graphic are:

1. The 60%-95% quantiles are left out of the chart. Why?

2. The purpose of the chart is to compare contributions of predefined sectors. Why compare the SUM of two groups with one other, and place it in the context of a comparison of two groups measured individually? That just doesn't make sense.

Posted by: david on November 5, 2003 11:31 AM

Dans point was more that you split the top 5% between 1 and 2-5, Is this an arbitrary split. You say predefined but by whom was it predefined and for what reason.

There are only three class in common parlance. Lower, middle and upper or in economic terms the poor, the middle class and the rich.
As a middle class person in the 40% (or the Australian equivalent) the top 5% is all the same to me so it would make an even more stark display if the top 5% were shown to pay at least as much as the whole middle class.

Your graph shows the rich split between the rich and the very rich. Why have you not split the middle class between the upper middle class and the lower. Its just as valid.

Just because the newspapers used that split, you don't have to.

Posted by: Brendan Eales on November 5, 2003 02:46 PM

I am NO statistician, but I do know a few things about sociology and political economics. The point of splitting the "rich" from the "very rich" is to show the tremendous consolidation of wealth among that tiny sliver of the ultra-rich. Forget two splits, Dan and Brandon, split that last 5% into 20 segments and see how extreme it gets. The US now has greater income inequality than even in the 1920s-- we're headed back to the Gilded Age with its robber barons. Read anything recent by Paul Krugman and he can break this down for you. Quite simply, break "the rich" down into however many segments you wish, but each rung up the latter represents a similar rise in political sway. These facts are so plain and so plainly alarming, the ultra-rich and their political water-carriers need charts like the one that started this post to muddy the waters.

Posted by: jay on November 5, 2003 07:26 PM

Short comment, I know, but my general response to this is that the best chart would be one that did it percentile by percentile, and lined them up to show some sort of curve. It's late so I may not be as clear as I'd like but I think I got my idea across.

Posted by: Nyoka on November 5, 2003 09:00 PM

However, even the correction is misleading. Basically the analysis is a two-dimensionel one: one dimension is the income, the other is the share of federal taxes paid by income percentiles. Such an analysis is called pareto analysis (concentration analysis9 and an appropriate graphical representation is available for this, too.

Posted by: Ralph on November 6, 2003 02:19 AM

That's absolutely correct: pareto analysis is appropriate here (in fact, any kind of distributional analysis would be an improvement).

However, it's unlikely that a Seattle Times reader would understand a Pareto chart. One of the challenges of graphics for the media is to make them illustrative, representative and easily comprehensible. That's why I decided to stick with a bar chart, since that was the format in the original article.

Posted by: david on November 6, 2003 08:21 AM
Post a comment