How the CDC is distorting covid-19 test results and "the state of the pandemic"

The Atlantic reports (this article shouldn't count against your monthly limit of Atlantic articles if you don't subscribe):

The Centers for Disease Control and Prevention is conflating the results of two different types of coronavirus tests, distorting several important metrics and providing the country with an inaccurate picture of the state of the pandemic.

We’ve learned that the CDC is making … a debilitating mistake: combining test results that diagnose current coronavirus infections with test results that measure whether someone has ever had the virus. The upshot is that the government’s disease-fighting agency is overstating the country’s ability to test people who are sick with COVID-19.

The agency confirmed to The Atlantic on Wednesday that it is mixing the results of viral and antibody tests, even though the two tests reveal different information and are used for different reasons.

This is not merely a technical error. States have set quantitative guidelines for reopening their economies based on these flawed data points. Several states—including Pennsylvania, the site of one of the country’s largest outbreaks, as well as Texas, Georgia, and Vermont—are blending the data in the same way.

Virginia likewise mixed viral and antibody test results until last week, but it reversed course and the governor apologized for the practice after it was covered by the Richmond Times-Dispatch and The Atlantic.… Vermont authorities claimed they didn’t even know they were doing this.

The widespread use of the practice means that it remains difficult to know exactly how much the country’s ability to test people who are actively sick with COVID-19 has improved.…

A negative test result means something different for each test. If somebody tests negative on a viral test, a doctor can be relatively confident that they are not sick right now; if somebody tests negative on an antibody test, they have probably never been infected with or exposed to the coronavirus. (Or they may have been given a false result—antibody tests are notoriously less accurate on an individual level than viral tests.) The problem is that the CDC is clumping negative results from both tests together in its public reporting.

Mixing the two tests makes it much harder to understand the meaning of positive tests, and it clouds important information about the U.S. response to the pandemic.…

Kristen Nordlund, a spokesperson for the CDC, told us that the inclusion of antibody data in Florida is one reason the CDC has reported hundreds of thousands more tests in Florida than the state government has. The agency hopes to separate the viral and antibody test results in the next few weeks, she said in an email.

Wait, the CDC merely "hopes" to fix this "in a few weeks"? Do they plan to leave the mistake online until then? How can they not fix this today?

This is why we shouldn't just "trust the experts."

Are you more likely to be killed by a gun or a car?

The New Republic tells us it's "no longer true" that "you’re more likely to die in a car accident" than by being shot (in the United States). TNR's evidence is that the number of gun-related fatalities has just barely exceeded the number of deaths in car accidents.

But that conclusion is at odds with the evidence. TNR follows the standard practice of those who support stricter gun control in focusing on gun-related deaths in general, without pointing out that many of those deaths are suicides or killings in self-defense.

TNR links to a Washington Post blog post that says that about two-thirds of gun deaths are suicides. So if you, like most people, are confident you're not going to intentionally kill yourself with a gun, then it's safe to say you're more likely to die in a car crash than by being shot.

I don't know how many of the gun deaths were killings in self-defense — I imagine it's difficult to come up with reliable statistics on that. But of course being fatally shot in self-defense is a serious concern only for those who attempt to commit serious crimes.

TNR also says that "gun deaths have inched up" (while fatal car crashes have declined), which gives the impression that you should be alarmed at your increasing likelihood of being fatally shot. After all, the TNR post is largely written in the second person, talking about "you." But the WaPo post clarifies that the increase in "gun deaths" has been driven by increasing gun-related suicides — not gun-related homicides, which have been decreasing.

Is the "affirmative consent" law going to cause rape and sexual assault to be taken less seriously?

There's a good but poorly spelled article in Reason with this headline:

Half of MIT Students Think It's Possible to "Accidently" Rape Someone (Thanks, Affirmative Consent!)

Elizabeth Nolan Brown writes:

Folks from President Obama to swearing 5-year-olds princesses have been citing a statistic that 20 percent of women on college campuses, or one in five, will be sexually assaulted while there—a stat that has also been routinely debunked. However, a new sexual assault survey from the Massachusetts Institute of Technology (MIT)—one of the first schools to release broad data on campus sex crimes—seems to corroborate everyone's favorite sketchy stat. . . .

But students are confused about how alcohol and intoxication affect consent, which perhaps speaks to increasing progressive activism around the idea that drunk people can't give consent. Only about three-quarters of respondents said they feel confident in their own ability to judge whether someone is too intoxicated to consent to sex. And more than half agreed that "rape and sexual assault can happen unintentionally, especially if alcohol is involved."

I just want to repeat that one more time: Half of the MIT students surveyed think it's possible to "accidently" rape someone. When you consider undergraduates alone, this rises to 67 percent.

This is what we get when people push an idea that rape is really often a matter of consent confusion or a drunken misunderstanding and not something that one person (the rapist) intentionally does to another. This is exactly what those of us opposed to affirmative consent standards mean when we worry about it muddying the waters of consent and confusing the definition of rape. About a fifth of female undergraduates and a quarter of male undergraduates surveyed agreed that "when someone is raped or sexually assaulted, it’s often because the way they said 'no' was unclear or there was some miscommunication."

The supposedly progressive new law may actually cause us to retrogress.

As for the spelling: an adverb is based on an adjective, not a noun. Thus, "accidentally," the adverb, is based on "accidental," the adjective, not the noun "accident."

"Your odds of dying in a terrorist attack...

... are still far, far lower than dying from just about anything else." See point #8 at that link, which points out:

In the last five years, the odds of an American being killed in a terrorist attack have been about 1 in 20 million (that’s including both domestic attacks and overseas attacks).

And Megan McArdle explains why deadly terrorist attacks are "so rare":

I can right now think of several terrorist strategies which would kill a lot of people and be nearly impossible to defend against. I'm not going to offer terrorists any ideas, but here's one that's already been tried: the DC sniper. They were caught only because they were too poor to switch the cars they were shooting from; a slightly better bankrolled effort could have effectively gone on forever. And you can't really imagine how you'd re-engineer America to defend against such attacks.

So why don't they happen more? The most convincing answer I've gotten to that question is that fostering terror is only one of the aims of a terrorist attack. These attacks also function as recruiting, and as fundraising promotions for your terrorist organization. There are what you might [call] business considerations, in other words, and those business considerations dictate the kinds of attacks that terrorists want to carry out.

Thank God, randomly shooting one person every week or so does not satisfy the business plan. Terrorists want large, splashy attacks on specific sorts of targets that have high emotional resonance for both the victims and the people on your side who you hope to recruit, or tap for money. This helps explain why Al Qaeda was so obsessed with the Twin Towers, a place that--until they fell--most New Yorkers regarded as a rather ugly landmark containing some so-so office space. To a terrorist group looking for publicity, on the other hand, it had immense symbolic value: the tallest building in America's biggest city, with the hubristic name of "World Trade Center".

That's why we don't get high-frequency, low-intensity attacks on crowded spaces near some Texas town that no one in Abottabad has ever heard of. When attacks on those places happen, they tend to be the provenance of local lunatics for whom the nearby mall, or primary school, has some immense symbolic emotional importance.

And thankfully, there are not nearly as many lunatics as there are schools, or malls, or even marathons. That's why most of our public spaces are safe from mad bombers--and I'm glad to say, why they will remain so.

A slight problem with the "new poverty measure"

It doesn't actually measure poverty.

People and percentages

Paul Krugman points out the fallacy of characterizing a group of people as a 47% bloc: people don't stay in the same situation for their whole lives. So a statistical percentage isn't equivalent to a group of actual human beings.

The exact same fallacy is committed by those who divide the country into "the 1%" and "the 99%."

Why does Obama think ATMs take jobs away from bank tellers?

President Obama said in his recent speech in Kansas:

Over the last few decades, huge advances in technology have allowed businesses to do more with less, and made it easier for them to set up shop and hire workers anywhere in the world. And many of you know firsthand the painful disruptions this has caused for a lot of Americans. . . .

If you were a bank teller or a phone operator or a travel agent, you saw many in your profession replaced by ATMs or the internet. Today, even higher-skilled jobs like accountants and middle management can be outsourced to countries like China and India. And if you’re someone whose job can be done cheaper by a computer or someone in another country, you don’t have a lot of leverage with your employer when it comes to asking for better wages and benefits — especially since fewer Americans today are part of a union.

Now, . . . there’s been a certain crowd in Washington for the last few decades who respond to this economic challenge with the same old tune. “The market will take care of everything,” they tell us. If only we cut more regulations and cut more taxes — especially for the wealthy – our economy will grow stronger. Sure, there will be winners and losers. But if the winners do really well, jobs and prosperity will eventually trickle down to everyone else. . . .

It’s a simple theory – one that speaks to our rugged individualism and healthy skepticism of too much government. It fits well on a bumper sticker. Here’s the problem: It doesn’t work. It’s never worked.

If you're going to make such a stark claim and phrase it in pragmatic terms of what has "worked," you should at least make sure the facts are on your side with respect to your own specific examples. But Matthew Yglesias disproves Obama's claim about ATMs replacing bank tellers with one simple chart (from this site):

Yglesias explains:

[I]t's true that since the recession started, we've seen fewer people on the job as tellers. But that's not a decades-long technologically induced trend. It's a recession. The [U.S. Bureau of Labor Statistics] continues to project long-term growth in the number of bank tellers presumably because banks like to open branches to attract deposits. It's true that bank tellers as a share of the labor force should shrink but that's different. Banks are doing more with a little bit more, not doing more with less.

Of course, one question is: why didn't Obama have anyone fact-check his speech? But more importantly, this shows Obama's attachment to left-wing ideology. He has faith that technological progress destroys jobs. If you're strongly inclined to see advances in technology as setbacks to society, then it's easy to believe that the idea of an efficient free market doesn't make sense.

Mickey Kaus debunks the New York Times article on the "near poor."

The New York Times ran a prominent article last week on "near poverty":

They drive cars, but seldom new ones. They earn paychecks, but not big ones. Many own homes. Most pay taxes. Half are married, and nearly half live in the suburbs. None are poor, but many describe themselves as barely scraping by.

Down but not quite out, these Americans form a diverse group sometimes called “near poor” and sometimes simply overlooked — and a new count suggests they are far more numerous than previously understood.

As a commenter on Kaus's blog points out, it's amazing that driving a non-new car is now considered a sign of poverty or anything close to it.

Here are my favorite points by Kaus (with all emphasis in the original):

– “Perhaps the most surprising finding is that 28 percent work full-time, year round.” The Times thinks this 28 percent figure is surprisingly high. (“These estimates defy the stereotypes of low-income families,” says the Census official). Does 28% seem high to you? To me it seems low.

– “Bruce Meyer, an economist at the University of Chicago, warned that the numbers are likely to mask considerable diversity. Some households, especially the elderly, may have considerable savings.” The “near poor” category also includes unemployed 23 year old college graduates from wealthy families, stockbrokers who had a really bad year, moderately paid workers who live in Silicon Valley (where, thanks to the cost-of-living correction, you can make $51,000 and still be “near poor”). Indeed, the vast diversity of the “near poor” category makes it virtually useless. It is a granfalloon, Kurt Vonnegut’s term for a false class of people.

– . . . [A]s society grows richer, you’d think more people would be able to take a year off and live off their assets. Yet they show up as poor—because the poverty numbers measure income, not wealth. (I qualified for the low-income Earned Income Tax Credit once when I owned a house in Georgetown. There was no asset test.) I’m not saying these people are a significant portion of the statistically poor. But they’re probably a growing portion (maybe even the “fastest growing portion,” to use the standard journalistic con that makes the growth of a small population seem significant). . . .

– I’m also suspicious of the way the fancy new poverty measure takes into account regional variations in the cost of living and medical expenses. I live in an expensive part of an expensive region because it is worth it to me. I could live in North Dakota. Does that make me “poor” or have I chosen to consume in one way (nice town) rather than another?

IN THE COMMENTS: Remembering how American "poverty" looked in the Soviet Union.

The problems with poverty statistics

The U.S. Census Bureau's September report saying that poverty is soaring may have been "off the mark."

The New York Times explains:

Concocted on the fly a half-century ago, the official poverty measure ignores ever more of what is happening to the poor person’s wallet — good and bad. . . .

[S]afety-net programs have played a large and mostly overlooked role in restraining hardship: as much as half of the reported rise in poverty since 2006 disappears.

The Times also says the Census Bureau's report ignored the fact that "rents are higher in places like Manhattan than they are in Mississippi."

There are "fewer people in abject destitution, but a great many more crowding the hard-luck ranks of the near poor, who do not qualify for many benefit programs."

The Census Bureau will release a new report on Monday, hopefully correcting some of these errors.

This other New York Times piece looks at other measures of poverty from the past few years. Some of the findings:

Poverty is declining among women ages 25-39.

Five million children in America have risen out of poverty in the past 10 years.

Poverty is declining among Hispanics, and declining dramatically among blacks. Poverty is increasing among Asians.

Poverty is decreasing in rural areas and increasing in urban areas.

Poverty appears to be increasing among the elderly, but this "could be a statistical anomaly. Many elderly people use retirement savings but do not report that money as income."

Have educated women stopped getting married?

Nope.

A silver lining to budget cuts for police departments

Budget cuts are forcing the police to come up with innovative ways to police more efficiently, possibly spurring police departments to achieve the same or even better results than in the past, at lower costs.

That article focuses on this example (though it also lists several others):

Partnering with criminologists from George Mason University, a team led by Sacramento Police Sergeant Renée Mitchell identified 42 “hotspots”—street corners that attracted the highest percentages of violent crime in California’s second most violent city.

As part of a 90-day study conducted between February and May this year, Mitchell and her team assigned officers to visit a randomized rotation of three or four of these hotspots for 12 to 16 minutes apiece during shifts. That meant police would inhabit Sacramento’s most dangerous corners about every two hours. The officers were told to be “highly visible” during these visits—to step outside patrol cars, to talk with people.

This was a change for Sacramento police. It focused on places to target rather than specific crimes, and relied on data rather than police instinct. The results, Mitchell says, were striking.

“Part I” crimes—which include violent offenses such as murder, rape and robbery, as well as property crimes such as burglary and vehicle theft—decreased by 25 percent in these hotspots. Calls for service decreased by nearly 8 percent.

The police were able to achieve those huge successes in the most high-crime areas of the city for just $75,000, which the article tell us is "less than one percent" of the police department's 2011 budget of $116 million. If my math is right, that's technically true but a dramatic understatement: 75,000 divided by 116 million isn't just less than one-hundredth; it's less than one-thousandth (about 0.065%). If you multiply the cost by 4 to estimate what this practice would cost year-round (instead of the study's 3-month period), it's only about a quarter of 1% (about 0.259%). The media expect us to think of "one percent" as "the smallest possible percentage," so they don't bother to make even smaller divisions than that.

Mitchell, the police sergeant who led the team that implemented the study, says:

“Arrests are glamorous. . . . People want to see that guns and drugs are being taken off the streets. But that’s reactive. We should be working to prevent. Our job is to reduce the opportunity for crime, not necessarily to patrol every street corner and make high-profile arrests. Sooner or later we’re going to have to face that.”

I've tagged this post with "unintended consequences." That phrase is normally considered negative, but it can be positive too.

(The article is from a brand-new site by the Atlantic called the Atlantic Cities.)

Is rape during war undercounted or overcounted?

Both.

Mitt Romney's Republican opponents charge that under his health care reform, there are long wait times to see doctors.

They also say primary care physicians are less likely to see new patients than they were before.

And they're right — according to Jonathan Cohn, a left-of-center supporter of President Obama's health care reform.

There just doesn't seem to be any evidence that Romneycare caused those problems. Cohn explains why at the link.

But I suspect that Cohn's approach — rigorously scrutinizing the relevant data — is not exactly going to dominate the discussion in the 2012 Republican presidential primaries.

A commenter on Cohn's post adds that even if Romney's state health insurance law does lead to negative consequences, we shouldn't assume that a national version of Romneycare (which is more or less what Obamacare is) will lead to similar consequences:

I know a (poor) woman who had moved from MA to Florida a few years ago. When universal health care was introduced in MA - and she subsequently got pregnant - she moved back. Though I haven't seen any studies it seems likely that such "market forces" are drawing a number of indigent people in which would tend to increase the load on the system. . . .

[This] problem is solved immediately by a national plan.

The theory of federalism says that states act as "laboratories of democracy," allowing us to see the consequences of policies on a relatively small scale before deciding whether to enact them nationwide. But individual states are so different from the entire United States that the results of a state-level policy experiment might be meaningless or even worse.

Are men declining? If so, why, and what's the solution?

Among men, median wages have "declined by 28 percent, or almost $13,000 (in constant dollars)" since 1969. (And no, 1969 was not cherry-picked as their peak; they peaked in 1973.)

That's from a post called "The Struggles of Men" on the New York Times' Economix blog.

If you go to that post and look at the chart, you can see how much difference it makes whether we include all men, or just men who are working full time. Not surprisingly, the latter measure (represented by the red line on the chart) paints a rosier picture.

The blue line includes men who have dropped out of the labor force. If you go by that measure, men's wages seem to be taking a firm nosedive.

An Economix commenter says:

After about 1970 there was a big expansion of the labor force, for three reasons. The first was the Hart-Cellar act of 1965, which produced a big increase in immigration. The second was a surge of women into the labor force after 1970. The third was the entry of the baby boom kids in the late '70s.

It really shouldn't be a surprise that an expansion of the labor supply leads to stagnant or falling prices (wages).

(That commenter has expanded on those points in a blog post, which seems to be totally anonymous.)

Glenn Reynolds posted the blog post to Facebook, and Brittany Gardner had this take (she gave me permission to quote from our discussion using her name):

As female strengths (empathy, social networks, collaboration) become increasingly marketable in a workforce, we will have to be sure that male strengths (competition, experimentation, systems thinking) can find a new niche and balance to match. With many traditionally "male-dominated" fields being automated or outsourced, I think that one logical entry field will be one women have dominated for a while.... child-rearing. Early education needs a complete overhaul to meet the needs of the next generation, and who better to design and implement new systems and experiment bravely and rationally, than men?

Men may also find they are uniquely equipped to get kids off the couches, net, and video games for a bit, and into the physical world, working on hands-on projects, games, experiments, and exploration.

I responded:

I agree about child-rearing, but the norms against men taking a lead role in raising children seem to be a lot more entrenched than the norms against women in the workforce. It's no longer socially acceptable to question whether women can do any job they want, but boys will still be teased if they pretend to be daddies; we'd feel more comfortable to see them pretending to be mass murderers. Many people might accept the idea of a stay-at-home dad in theory, but how many men would truly feel comfortable in that role in practice? Alas, men aren't seen as one of the kinds of people we're supposed to be concerned with, so anyone trying to draw attention to these issues will be fighting an uphill battle.

Glenn Reynolds said:

I think John's right, especially as any man who actually does want to work with children is immediately suspect as a child abuser.

(See "Eek! A Male!")

Here's Brittany Gardner's response to my comment:

There is a growing number of stay-at-home dads at the park and at playgroups I've attended. I was impressed at how much "better" they seemed to be at effortlessly managing behavior and teaching/interacting with their children . . . .

If it's true that unemployment is becoming a reality for more men, talking up roles with the next generation makes a lot of sense, and other projects can be worked on simultaneously. (All the stay-at-home fathers I know also work from home, or have productive hobbies they can work on simultaneously and involve their children in, to both parties' benefit and enrichment.)

The idea that steer-heading the next generation of humans is somehow emasculating, needs to be abandoned.

To paraphrase a quote I read recently: Sense of masculinity has always flowed from men's utility in society. Not the other way around.

And here's her response to Glenn Reynolds:

I think the suspicion directed at men who want to work with children mirrors the suspicion of motives we direct at women who take leadership positions . . . I think the more we normalize both, the better.

How to lie with statistics: Unemployment edition

The US Department of Labor's new unemployment report -- which seems to show a major drop in the unemployment rate yet a tiny increase in actual jobs -- is explained here. (via)

A new study that says gay people are coming out earlier than in the past . . .

. . . is wrong in a "new and interesting" way.

That blog post (1) explains how the study went wrong and (2) asks whether it could have possibly been right.

As to the second point, the blogger (Ben Goldacre) explains:

It’s a difficult analysis to design, because in each age band, there is no information on gay people who are not yet out, but may come out later, and also it’s hard to compare each age band with the others.

(The comments section on that post also has a lot of relevant insights.)

This reminds me of the oft-repeated factoid that "50% of marriages end in divorce." How could you ever determine whether this is true? You can observe divorces that have actually happened, but you can't possibly know whether existing marriages will end in divorce.

Even questions that seem to be about concrete, observable facts can't necessarily be answered by empirical research.

Why are red states' workforces more dominated by public workers than blue states?

Here's a chart of state and local public employees as a percentage of the population in each state.

It's not surprising that Washington, DC, as the seat of the federal government, has a lower percentage than any of the 50 states, since we're not counting federal employees.

What is surprising is that the states with the largest such percentages are overwhelmingly Republican-leaning. The 10 states whose workforces are most dominated by state/local public employees have 8 solidly Republican states, with only 1 swing state (New Mexico) and 1 Democratic state (New York, whose percentage is still far from the highest).

The 10 states with the proportionately smallest state/local public workforces are a mixed bag of red, blue, and purple states, including the deep-blue Massachusetts.

Catherine Rampell, the author of that New York Times blog post, crunches the numbers and finds that, indeed, "the more dominated a state is by [state/local] public-sector workers, the less likely that state was to vote for the Democratic presidential candidate."

The state with the highest percentage is Wyoming. You might think this is an anomaly because it has the smallest population; therefore, basic services and infrastructure require a higher percentage of its workers. Similarly, a commenter on the Times post says:

The results are largely explained by geography and demographics. With respect to geography, for example, most state and local employees are employed in the educational sector. States that are not densely populated, such as Wyoming, tend to have a lower teacher-to-student ratio (Wyoming has an average ratio of 1 teacher to 13.1 students and the average national ratio is about 1 to 24 as reported here on Economix). Also, low population states such as Wyoming and Alaska require relatively higher ratio of employee to general population to deal with basic governmental functions (each state has the same number of governors, etc)[.] Econom[ies] of scale play[] a role here.

That wouldn't seem to gibe with Rhode Island, which has one of the smallest populations and the 6th-lowest percentage of state/local public workers. But that commenter has an explanation: the states with the highest percentages have a large area, so they need lots of government for things like highway maintenance.

Rampell observes:

[I]t is not such a big contradiction for states to be hostile to candidates perceived to be expanding the size of the federal government, and to still employ lots of workers at the state and local levels.

Maybe we should stop equating "red states" with opposition to "government." They merely tend to be against the federal government having a major role in their lives. And isn't that natural? They have relatively little need for it -- they already enjoy so much help from state and local government.

War photography and violence

This New York Times piece on war photographers taking photos of dead or mortally wounded American soldiers ends with a "wholly unexpected" comment from photographer Don McCullin:

“I feel I totally wasted a large part of my life following war. I get more pleasure photographing the landscape around my house in my twilight years.

"Have we learned any lessons from the countless pictures of pain and suffering? I don't think we’ve learned anything. Every year, there’s more war and suffering."

But is that last statement true? Steven Pinker wrote this essay saying it's actually the opposite:

Violence has been in decline over long stretches of history, and today we are probably living in the most peaceful moment of our species' time on earth. ...

Conventional history has long shown that, in many ways, we have been getting kinder and gentler. Cruelty as entertainment, human sacrifice to indulge superstition, slavery as a labor-saving device, conquest as the mission statement of government, genocide as a means of acquiring real estate, torture and mutilation as routine punishment, the death penalty for misdemeanors and differences of opinion, assassination as the mechanism of political succession, rape as the spoils of war, pogroms as outlets for frustration, homicide as the major form of conflict resolution -- all were unexceptionable features of life for most of human history. But, today, they are rare to nonexistent in the West, far less common elsewhere than they used to be, concealed when they do occur, and widely condemned when they are brought to light. ...

The decline of violence is a fractal phenomenon, visible at the scale of millennia, centuries, decades, and years. It applies over several orders of magnitude of violence, from genocide to war to rioting to homicide to the treatment of children and animals. And it appears to be a worldwide trend, though not a homogeneous one. The leading edge has been in Western societies, especially England and Holland, and there seems to have been a tipping point at the onset of the Age of Reason in the early seventeenth century. ...

On the scale of decades, comprehensive data again paint a shockingly happy picture: Global violence has fallen steadily since the middle of the twentieth century. According to the Human Security Brief 2006, the number of battle deaths in interstate wars has declined from more than 65,000 per year in the 1950s to less than 2,000 per year in this decade. In Western Europe and the Americas, the second half of the century saw a steep decline in the number of wars, military coups, and deadly ethnic riots.

Zooming in by a further power of ten exposes yet another reduction. After the cold war, every part of the world saw a steep drop-off in state-based conflicts, and those that do occur are more likely to end in negotiated settlements rather than being fought to the bitter end. Meanwhile, according to political scientist Barbara Harff, between 1989 and 2005 the number of campaigns of mass killing of civilians decreased by 90 percent.

Since McCullin purported to base his view that war photography is futile or counterproductive on empirical evidence, I hope he'd change his view if presented with this contrary evidence. But it wouldn't be surprising if he didn't. People reflexively refer to any kind of social problem as an "increasing" problem, and this tendency seems to be more powerful than statistics. Pinker lists a few factors that cause people to make this mistake:

Partly, it's because of a cognitive illusion: We estimate the probability of an event from how easy it is to recall examples. Scenes of carnage are more likely to be relayed to our living rooms and burned into our memories than footage of people dying of old age.

So, ironically, war photography itself feeds into the belief that war photography is ineffectual.

More from Pinker:

Partly, it's an intellectual culture that is loath to admit that there could be anything good about the institutions of civilization and Western society. Partly, it's the incentive structure of the activism and opinion markets: No one ever attracted followers and donations by announcing that things keep getting better. And part of the explanation lies in the phenomenon itself. The decline of violent behavior has been paralleled by a decline in attitudes that tolerate or glorify violence, and often the attitudes are in the lead. As deplorable as they are, the abuses at Abu Ghraib and the lethal injections of a few murderers in Texas are mild by the standards of atrocities in human history. But, from a contemporary vantage point, we see them as signs of how low our behavior can sink, not of how high our standards have risen.

Pinker's essay only observes that there has been a decline; he doesn't try to explain it. He ends by saying:

With the knowledge that something has driven [violence] dramatically down, we can also treat it as a matter of cause and effect. Instead of asking, "Why is there war?" we might ask, "Why is there peace?" From the likelihood that states will commit genocide to the way that people treat cats, we must have been doing something right. And it would be nice to know what, exactly, it is.

There's no way war photography could be the main answer to Pinker's question, since he's talking about a trend that's been underway since long before photography existed. But the fact that people are willing to look at the reality of war in vivid detail might have played a small role in the progress we've made.

Why is it so hard to put a number on Americans without health insurance?

George Will and Megan McArdle take issue with the Census Bureau's figure from 2008 that 46 million Americans are in need of health insurance. They also disagree with President Obama's statement in his speech to Congress last week that there are "more than 30 million American citizens who cannot get coverage." Will estimates that there are "perhaps 20 million persons who are protractedly uninsured." McArdle suggests that a better statistic would be "more than 15 million" Americans.

Anthony Wright has a good response on TNR. I pretty much agree with him and frankly don't have much to add to what he says.

The debate over these statistics can be a red herring. There's a reason why conservatives and libertarians are taking a different stance on how to interpret the statistics than liberals. You can't interpret the data without expressing your ideology. If you're a liberal, you're likely to view the number of Americans without insurance as a crucial fact, maybe the most important fact in the whole health-care debate. If you're a conservative, you're likely to see that statistic as a smokescreen covering up the reality of the situation.

For instance, Will thinks the statistics are misleading because they gloss over the transitory nature of being insured. He says the 46 million figure is just

a "snapshot" of a nation in which more than 20 million working Americans change jobs every year. Many of them are briefly uninsured between jobs. If all the uninsured were assembled for a group photograph, and six months later the then-uninsured were assembled for another photograph, about half the people in the photos would be different.

But Will bizarrely implies that his observation only mitigates the problem. It's arguably worse than the statistics suggest, since a "snapshot" fails to convey the fact that people who are insured right now need to worry about whether they'll be insured in the future. Wright points out: 

When just looking at a two-year period, far more people--nearly 1 in 3--find themselves uninsured.

It's understandable that opponents of health-care reform want to make the number of uninsured seem as insignificant as possible by quibbling over which ones we should really be counting. (Illegal immigrants? Legal immigrants? Poor people who aren't on Medicaid? Households making $75,000? $50,000?) It might be comforting, as Wright puts it, "to "think of the uninsured as a discrete population, one that can be marginalized." But the real problem is much broader: our health care is contingent on something as unreliable as employment, with only an incomplete safety net for Americans in need. Those of us who feel strongly that this is a grave problem aren't going to have our minds changed by the pseudo-statistical debate over how many uninsured Americans actually matter.

Unemployment in the double digits

Already.