Misleading with statistics

The headline on the AP story is breathless. "Army suicides highest in 26 years!"

That basic fact is true; Army suicides are up sharply, just like they spiked during the first Gulf War. The 2006 rate was 17.3 suicides per 100,000, a near doubling of the low of 9.1 per 100,000 in 2001.

But a closer look at the numbers is in order before we start jumping to conclusions.

The 17.3 rate translates into 99 suicides out of a population of about 500,000 soldiers. So it's hardly an epidemic.

And if you compare it to civilian suicide rates, it's even less of an issue. A pair of pdfs here produce the following table:

2004 CIVILIAN SUICIDE RATES (per 100,000 population)
Overall: 11.1
Ages 15-24: 10.4
Males: 17.7

Wait a second, you say. Other than that "males" category, the military suicide rate is clearly much higher than the civilian rates.

But look what happens when we break down the "age" category even further and combine it with gender:

Males, age 15-19: 12.65
Males, age 20-24: 20.84

You can see where I'm going here. Soldiers are mostly males in their early 20s. So a proper comparison of apples to apples shows that the military suicide rate, despite being at a 26-year high, is still lower than the comparable civilian rate. All that in spite of combat stress, the stress of being part of a "stretched" military, and access to all sorts of military-grade weaponry.

People are right to be concerned. The rate has doubled, after all. It's clearly a symptom of strain and each one is a personal tragedy besides. The military should do what it can to reduce those numbers.

But let's not overreact. The problem is small, and soldiers are still less likely to kill themselves than civilians are. This is more an example of shallow and innumerate reporting than it is a sign of serious problems in the military.

media, suicide, politics, midtopia

Redistricting update

In Wednesday's post about a new redistricting algorithm, I focused on the technical specifics of the proposed method, and the pros and cons that made it different from previous proposals.

I deliberately avoided delving into all the standing arguments about how best to draw districts, largely because I've discussed them in tedious detail before. But judging by the comments and e-mails I've received, a quick overview would be useful.

The complicating factor is that there are situations where gerrymandering produces a better result than a purely nonpartisan approach. That's because redistricting involves several legitimate but competing principles:

1. District boundaries should make geographic sense.

2. District boundaries should be nonpartisan.

3. The makeup of Congress should reflect the makeup of the citizenry. If a given group makes up 15 percent of the citizens, it should probably have about 15 percent of the Congressional seats.

4. Districts should be socially coherent, so that their representative can truly represent them. A suburban neighborhood on the edge of the city, for instance, is better grouped with a suburban district with similar demographics than an urban district with which it has nothing in common.

The problem is that #4 is highly subjective, and it's hard to get #3 if you want both #2 and #1. For instance, assuming minorities are somewhat evenly spread through the population, a totally nonpartisan approach would create zero districts where blacks, say, are a majority -- greatly reducing the political power of black voters.

So one consequence of a completely objective method for drawing districts would probably be a steep drop in the number of minority members of Congress.

That doesn't feel right. That's why increasing minority representation is one of the few legal exceptions to the "no excessive gerrymandering" rules.

Besides leading to travesties like District 12 on the above map of North Carolina, such efforts created a whole new set of problems. Republicans, for instance, found that if you draw those minority districts right you not only get more minorities but you also get more safe Republican seats. Republicans gained 10 House seats in the 1992 elections -- 12 of them in states where minority districts had been created. Gains in those states, in other words, offset losses elsewhere.

This, in turn, has contributed to careerism and the alarming polarization of national politics, because someone with a safe seat is free to demagogue as much as they like, and it's harder to find common ground with other legislators. Why would a representative from poor inner-city Detroit care about the issues most dear to people in suburban Orange County, Calif.? Or vice versa? Their constituents have almost nothing in common.

If more districts were politically and racially mixed, you might find more legislators with direct experience and interest in a range of issues, making sane policy and pragmatic compromises more likely.

The problems don't end there.

Once you allow gerrymandering for one purpose, it opens the door for a whole host of questions, like: How much gerrymandering is too much? If it's okay to gerrymander for race, how about gender or religion or other demographic features? You end up having to engage in a lot more arbitrary, complicated and difficult-to-defend line-drawing than if you simply ban the practice altogether.

And while we must recognize race as a political force, why should we encourage it? Being willfully color-blind often disguises residual racism, but that doesn't mean we should build racial assumptions into the very structure of our political system. Perhaps if we stop reinforcing the idea that race should be a factor in politics, it will stop being as much of a factor.

In the end, while #3 and #4 are commendable ideals, in my book they come in second to #1 and #2. As an extension of adjusting the algorithm to account for existing political boundaries I'm willing to accept very minor adjustments to a district's boundaries in order to nudge it over into "minority" status. That will result in fewer minority districts, but more than if no adjusting were done at all.

To minorities who say that such a move destroys their political power, I'd say "join the club." Speaking as an agnostic political moderate, I can confidently say my views aren't proportionately represented in Congress, either.

The solution to that is to organize politically to create a voice out of proportion to your numbers. It worked for the religious right; it works for unions; it can work for racial groups, too. And it has the added advantage of making a given minority's concerns part of the political calculus of a far larger number of Congressmembers. Sufficiently organized, that could result in far greater political influence than could every be achieved by packing minorities into their own districts.

Even better would be to stop viewing representation through the prism of race and start organizing around specific issues instead. Symbolically important as minorities in Congress can be, would black voters, for instance, really prefer a black representative with whom they totally disagree, or a white representative whose positions they support?

Here in Minnesota, one of the main political divides is rural vs. urban. I find it difficult to believe that a rural black has more in common politically with an urban black than with a fellow rural dweller of any color.

In sum, then, the collective good of removing politics from the redistricting process generally outweighs the collective good of proportional representation. The first should be the priority, while the second should be a bonus to be added where possible -- but only if it doesn't derail the whole shebang.

redistricting, politics, midtopia

Nonpartisan redistricting

Last week, Simon over at Stubborn Facts linked to a group that has developed a mathematical algorithm for drawing Congressional districts -- one that would make gerrymandering a thing of the past.

The so-called splitline algorithm follows a few simple rules to divide up a state using the fewest number of straight lines as possible.

For example, here's how Tennessee looks now:


And here's how it would look if the algorithm were used:


Obviously, the computer-generated map seems far more in keeping with the spirit of geographical representation.

But just as obviously, the upside of total nonpartisanship is gained at the expense of ignoring all existing natural and political boundaries. The district lines would arbitrarily split cities, neighborhoods, even streets. It would be technically simple to determine what district you were in using a GPS device, but it would be hard to do so simply looking at a map.

That said, gerrymandering often produces the same result, and for far less defensible reasons.

Such a problem seems solvable, however. The algorithm could be linked to a database of geographical and political boundaries, and modified to draw the simplest districts while giving maximum deference to those boundaries. The key point -- automated, nonpartisan district drawing -- would be retained. All that would change is that the district borders would get a little more complicated in order to be easier to understand.

Like Simon says, it's a start, not a finish. But it's a promising one.

BTW, here's how Minnesota might look.



redistricting, politics, midtopia

Supply-side math

An economist weighs in with a trashing of a Wall Street Journal editorial that uses atrociously bad math in defense of supply-side economics -- in particular, an attempt to at last locate the elusive Laffer Curve. There's a more detailed discussion over at Kevin Drum's blog.

The WSJ's purpose was to show a "sweet spot" for taxation, where if you raised rates beyond that tax revenue actually went down. But there are several problems with their chart.

One is that they're using corporate taxes as their yardstick. But corporate taxes make up a much higher percentage of revenue in a tiny tax haven like Luxembourg than they do in large, diversified economy like the United States. And the main outlier, Norway, has huge corporate tax revenues because of its state oil monopoly. Those factors should have set off alarm bells for whomever was preparing the chart.

They compound the problem by drawing their line through Norway instead of either throwing out that data point (as an obvious outlier) or at least averaging it with the other data points. Then, as Drum points out, if the curve is to be believed, tax revenue crashes to zero at around 33 percent -- even though the chart itself shows companies with rates higher than that having significant tax revenue.

Third, we have no idea how or why they chose the countries in the sample -- indeed, some of the data points are unlabeled. There's no way to tell if the sample is representative, complete, or meaningful.

But mostly, the graph doesn't show any significant conclusions. For example, Australia has one of the highest revenue figures with a corporate tax rate of about 31 percent. But three unnamed countries have significantly lower revenue with the same tax rate. The only possible conclusion is that corporate tax rate is simply not the major influence at play.

As one of the commenters at the first link points out:

This (the chart) is the dumbest thing I've ever seen, and I've seen a lot. You don't have to be an economist. All you have to be is somebody who knows what a scatter diagram is.

Silly people.

The basic idea of the Laffer Curve is reasonable -- that there's a sweet spot where taxation is high enough to generate substantial revenue, but not so high that it discourages work and investment. It's simply an expression of the law of diminishing returns. The devil is in the details: where exactly is that sweet spot? Nobody has yet shown it.

It doesn't help that one of the main assumptions behind the Laffer Curve probably is false: that a 100% tax rate will generate no tax revenue because nobody will work if all their earnings are confiscated.

It's false because many people work for reasons other than money. If you love the work, you'll do it for free. Work provides a sense of accomplishment, a chance to get out of the house, a sense of worth.

Further, the Laffer argument assumes that the confiscated money is poured down a hole in the ground. In reality, it's fed back into the economy -- where it may, for instance, provide a worker at a state-owned factory with food and shelter in return for work. In other words, there are ways to incentivize people to work that don't involve money.

I'm certainly not arguing in favor of state socialism. While revenue at a 100% tax rate wouldn't be zero, it wouldn't be very high, either. Money also is the best incentive for a free society, letting people make choices and reap the benefits or drawbacks of those choices. It also seems to work the best for unlocking creativity and hard work, or persuading people to go to the trouble of pursuing specialty training or performing dangerous or unpleasant jobs. I'm just noting the practical reasons why nobody knows whether the tax sweet spot is 25 percent or 80 percent.

The WSJ folks are still idjits.

Laffer Curve, economics, politics, midtopia