Comments on: President Bush, methodology expert? Not! http://blog.psaonline.org/2006/10/17/president-bush-methodology-expert-not/ Thu, 18 Mar 2010 05:33:23 -0700 http://wordpress.org/?v=2.9.1 hourly 1 By: Joe Roeber http://blog.psaonline.org/2006/10/17/president-bush-methodology-expert-not/comment-page-1/#comment-1562 Joe Roeber Thu, 19 Oct 2006 06:00:45 +0000 http://blog.psaonline.org/2006/10/17/president-bush-methodology-expert-not/#comment-1562 What you don't say, David, apropos the Lancet estimates, is that statistical estimates of this sort produce distributions with associated probabilities. It is natural enough to take the middle of the distribution as a rough indication but it is a shorthand that should be set in the context of its measures of confidence. From the data I've seen, these are quite wide. Which is to say that the central number is not very meaningful. Your half the estimate is safer. Methodology apart - and everyone shd know the limits of statistical estimation - to treat it as a hard number is to mislead. Given the sophistication of the people at the White House, I assume this is intentional. Another lie. The other thing you don't t quite say is that any death is unacceptable, let alone the smaller numbers, LET ALONE even half the horrendous Lancet estimate. Is this how we improve the lives of the Iraqis? You do touch on it in your concluding point abt breaching the Geneva Convention... as if those tough, straight-shooting Texans care about such pansyish concerns. (Real men don't worry about other people.) Of course, every such revelation exposes all the more cruelly the lies told by the US and UK about the reasons for the invasion. Bring democracy to the Iraqis? Gimme a break! Re-colonise the Middle East and subordinate local concerns to American priorities, more like. Joe Roeber What you don’t say, David, apropos the Lancet estimates, is that statistical estimates of this sort produce distributions with associated probabilities. It is natural enough to take the middle of the distribution as a rough indication but it is a shorthand that should be set in the context of its measures of confidence. From the data I’ve seen, these are quite wide. Which is to say that the central number is not very meaningful.
Your half the estimate is safer. Methodology apart – and everyone shd know the limits of statistical estimation – to treat it as a hard number is to mislead. Given the sophistication of the people at the White House, I assume this is intentional. Another lie.
The other thing you don’t t quite say is that any death is unacceptable, let alone the smaller numbers, LET ALONE even half the horrendous Lancet estimate. Is this how we improve the lives of the Iraqis? You do touch on it in your concluding point abt breaching the Geneva Convention… as if those tough, straight-shooting Texans care about such pansyish concerns. (Real men don’t worry about other people.)
Of course, every such revelation exposes all the more cruelly the lies told by the US and UK about the reasons for the invasion. Bring democracy to the Iraqis? Gimme a break! Re-colonise the Middle East and subordinate local concerns to American priorities, more like.
Joe Roeber

]]> By: Emerson Tan http://blog.psaonline.org/2006/10/17/president-bush-methodology-expert-not/comment-page-1/#comment-1548 Emerson Tan Wed, 18 Oct 2006 06:53:10 +0000 http://blog.psaonline.org/2006/10/17/president-bush-methodology-expert-not/#comment-1548 Ok, so David knows all that stuff, I think he left it out for reasons of brevity. I also forgot to say I thought it was a good post. Ok, so David knows all that stuff, I think he left it out for reasons of brevity. I also forgot to say I thought it was a good post.

]]> By: Emerson Tan http://blog.psaonline.org/2006/10/17/president-bush-methodology-expert-not/comment-page-1/#comment-1539 Emerson Tan Tue, 17 Oct 2006 21:28:21 +0000 http://blog.psaonline.org/2006/10/17/president-bush-methodology-expert-not/#comment-1539 I think David has missed something key here. Essentially, that the administration doesn't care about Iraqi civillian deaths. Everything is keyed towards minimising US service deaths, which are reported. The US force protection strategy really doesn't help in many cases resulting in excessive civilian deaths and prevents US troops from acting effectively to supress the sectarian violence. Muslim or otherwise, Iraqi deaths get reported simply as numbers and there are no yellow ribbons or empty plots in Arlington for the US electorate to see and grieve over. I think David has missed something key here. Essentially, that the administration doesn’t care about Iraqi civillian deaths. Everything is keyed towards minimising US service deaths, which are reported.

The US force protection strategy really doesn’t help in many cases resulting in excessive civilian deaths and prevents US troops from acting effectively to supress the sectarian violence. Muslim or otherwise, Iraqi deaths get reported simply as numbers and there are no yellow ribbons or empty plots in Arlington for the US electorate to see and grieve over.

]]> By: R http://blog.psaonline.org/2006/10/17/president-bush-methodology-expert-not/comment-page-1/#comment-1531 R Tue, 17 Oct 2006 16:01:51 +0000 http://blog.psaonline.org/2006/10/17/president-bush-methodology-expert-not/#comment-1531 | Iraq Body Count ... “authors have drawn conclusions from unrepresentative data.” Without further detail, this claim seems meaningless. The point of statistical inference is to draw inferences about an unknown population by means of sampling. Thus, without knowing the population characteristics, one cannot tell how representative a particular sample is. (If one already knew the population characteristics, there'd be no point in sampling.) This is where statistical theory becomes useful. The drawing of a random sample is the best that one can do without good population knowledge. (Population knowledge does allow one to correct a random sample; for example, if it is known that the response rate to a survey decreases in the respondants' income, then a random sample will be more representative of poor people.) | the Lancet numbers ... mean about 500 unexpected violent deaths per day | throughout the country, which seems unlikely. This may seem unlikely if our standard is the number of deaths reported in the media. However, the Lancet authors note that: "Aside from Bosnia, we can find no conflict situation where passive surveillance [such as relying on media reports] recorded more than 20% of the deaths measured by population-based methods." | the United States should ... promote more accurate and transparent | monitoring of the true effects of violence on Iraqi civilian health. Can we upgrade this to "must demand" from "should ... promote"? A credible study by established researchers, supported by top research universities, and published in a top medical journal, has claimed that our project of liberation has killed two thirds of a million people - placing this among the most murderous periods of Iraqi history. If we wish to claim that we care for Iraqis, we cannot simply dismiss this off-handedly. Why has the independent assessment called for by the researchers in 2004 still not been conducted? | Iraq Body Count … “authors have drawn conclusions from unrepresentative data.”

Without further detail, this claim seems meaningless. The point of statistical inference is to draw inferences about an unknown population by means of sampling. Thus, without knowing the population characteristics, one cannot tell how representative a particular sample is. (If one already knew the population characteristics, there’d be no point in sampling.) This is where statistical theory becomes useful. The drawing of a random sample is the best that one can do without good population knowledge. (Population knowledge does allow one to correct a random sample; for example, if it is known that the response rate to a survey decreases in the respondants’ income, then a random sample will be more representative of poor people.)

| the Lancet numbers … mean about 500 unexpected violent deaths per day
| throughout the country, which seems unlikely.

This may seem unlikely if our standard is the number of deaths reported in the media. However, the Lancet authors note that: “Aside from Bosnia, we can find no conflict situation where passive surveillance [such as relying on media reports] recorded more than 20% of the deaths measured by population-based methods.”

| the United States should … promote more accurate and transparent
| monitoring of the true effects of violence on Iraqi civilian health.

Can we upgrade this to “must demand” from “should … promote”? A credible study by established researchers, supported by top research universities, and published in a top medical journal, has claimed that our project of liberation has killed two thirds of a million people – placing this among the most murderous periods of Iraqi history. If we wish to claim that we care for Iraqis, we cannot simply dismiss this off-handedly. Why has the independent assessment called for by the researchers in 2004 still not been conducted?

]]> By: Chuck Pena http://blog.psaonline.org/2006/10/17/president-bush-methodology-expert-not/comment-page-1/#comment-1529 Chuck Pena Tue, 17 Oct 2006 15:48:46 +0000 http://blog.psaonline.org/2006/10/17/president-bush-methodology-expert-not/#comment-1529 It is interesting that while the Bush administration casually dismisses the Lancet study as "discredited," it always insists that everyone killed as a result of U.S. forces bombing a target in Iraq are either insurgents or terrorists. Yet they can offer no proof or methodology for making such claims, but expect us to accept their assertions. It is interesting that while the Bush administration casually dismisses the Lancet study as “discredited,” it always insists that everyone killed as a result of U.S. forces bombing a target in Iraq are either insurgents or terrorists. Yet they can offer no proof or methodology for making such claims, but expect us to accept their assertions.

]]>