Apparently the Washington Post and CBS have their difficulties with mathematics these days. Earlier today, the Post reported that the death toll in Iraq from reprisals following the destruction of the Askariya shrine in Samarra had topped 1,300. Later today, most news organizations agree with Iraqi and American officials that the Post’s numbers were greatly exaggerated, as Editor and Publisher reports:
Sectarian violence that followed last week’s bombing of a Shiite shrine have killed more than 1,300 Iraqis in the past few days, many times the figure previously reported by the U.S. media and the military, The Washington Post reported early Tuesday.
Later, however, Iraq Prime Minister Ibrahim al-Jaafari called such high death totals “inaccurate and exaggerated,” without mentioning the Post.
In comparison, The New York Times reported Monday that the recent violence “brought the country to the brink of civil war and left at least 200 dead.” Others had produced similar figures.
On Tuesday, the Times increased that number to “379 dead and 458 wounded, the nation’s Council of Ministers said today. At least 246 people in Baghdad alone were killed, the top two city morgue officials said.” …
The Associated Press carried this on Tuesday: “The Post cited figures from the Baghdad central morgue, but an official there told The Associated Press that as of Sunday night they had received only 249 bodies tied to the violence. The Post figure appeared high based on police and hospital reports from the major population centers at the time of the attacks.”
The Los Angeles Times, after noting the different figures today, added another, from Haidar Safar, a Ministry of Health official in charge of compiling data from hospitals and morgues across the country. He said 519 Iraqis have died from violence across the country since the blast occurred.
A Knight Ridder report from Baghdad late Tuesday stated that an American military official in Baghdad said U.S.-led coalition forces had been able to confirm only 220 such deaths since last Wednesday’s bombing.
No one argues that any of these numbers represent good news, but the report of 1300 deaths (which I used in a previous post) makes all of these counts pale in comparison. The methodology used by the Post appears somewhat suspect; their reporters counted dead bodies in a Baghdad morgue and assumed all of the deaths that appeared violent came from sectarian vendettas following the bombing. However, the morgue itself says that it has seen nowhere near the number of bodies claimed by the Post.
How many people have died in the violence this week? Too many, of course, and until today’s 68 bombing deaths, it had appeared that the violence had burned itself out. The cycle of retribution has limited itself to Baghdad, where the most radical of both sects concentrate in a relatively small area. It looks as though the Post simply got their astronomical number wrong, which detracts from the reporting and undermines its credibility. They need to demand better methodology from their reporters and, frankly, better editing in their offices.
The Post isn’t alone today, either. CBS released a poll showing the George Bush has tanked in public opinion, dropping to a miniscule 34% coming into the midterm primary season. This would worry most politicians, but the CBS poll has a major sampling problem, as reported at The Corner and just about everywhere else in the blogosphere.
First, the poll samples adults in general, not voters or likely voters. That’s not fatal, but it does tend to skew the data and make it less reliable as a predictor of voter action. However, what makes it completely unacceptable is the wide disparity between Republicans and Democrats in the sample. Even when weighted by CBS to correct for a 13-point Democratic advantage in the sample, the gap remains at nine points, and Republicans still wind up with less representation than independents. That nine-point gap skews the end results and makes this poll representative of … New York, Massachussetts, and California, but not the rest of the nation.
In contrast, the more reliable tracking poll at Rasmussen shows that Bush’s numbers have held steady at the mediocre level of the mid-40s. Today’s result shows a 43% approval rating, down six points from its two-week peak. That seems a bit more realistic than CBS’ numbers.
CBS has an explanation of its polling process at its blog, Public Eye, and the Anchoress has a long list of blog links debating the topic. Vaughn Ververs writes a calm and rational defense of the methodology, but in the end cannot explain two aspects of their sampling — the huge disparity in the raw numbers between Republicans and everyone else, and the weighting that winds up with almost the same disparity as before.
If this is the best math that the Exempt Media can muster, our educational system needs a lot more focus on basics.