home

Iraq Update

by TChris

Two stories in the NY Times this morning highlight the administration's post-"mission accomplished" failures in Iraq. Despite claims that insurgent attacks in Iraq are declining, declassified statistics "portray a rebellion whose ability to mount attacks has steadily grown in the nearly three years since the invasion." Meanwhile:

Virtually every measure of the performance of Iraq's oil, electricity, water and sewerage sectors has fallen below preinvasion values even though $16 billion of American taxpayer money has already been disbursed in the Iraq reconstruction program, several government witnesses said at a Senate Foreign Relations Committee hearing on Wednesday.

< The Grammys: Bruce and 'Bring Them Home' | School Suspends 6 Year Old For Sexual Harassment >
  • The Online Magazine with Liberal coverage of crime-related political and injustice news

  • Contribute To TalkLeft


  • Display: Sort:
    Re: Iraq Update (none / 0) (#1)
    by kdog on Thu Feb 09, 2006 at 09:23:36 AM EST
    If memory serves, I believe the USSR also claimed victory in Afghanistan..and we all know how that turned out.

    Re: Iraq Update (none / 0) (#2)
    by Al on Thu Feb 09, 2006 at 09:30:22 AM EST
    It depends on what you mean by "turning out well". Things certainly turned out well for Halliburton and other military contractors. They may yet "turn out well" for American oil companies if they can still persuade the Iraqi "government" to sign over the oil.

    Re: Iraq Update (none / 0) (#3)
    by pigwiggle on Thu Feb 09, 2006 at 09:35:08 AM EST
    Despite claims that insurgent attacks in Iraq are declining, declassified statistics "portray a rebellion whose ability to mount attacks has steadily grown in the nearly three years since the invasion."
    Having looked at the number myself I would say that contrary to the administration and the NY Times, the attacks look to be relatively unchanged over the past three years. Absent any statistical analysis graphs are often no more useful than a Rorschach blot; you'll all see what you want.

    Re: Iraq Update (none / 0) (#4)
    by Punchy on Thu Feb 09, 2006 at 09:54:53 AM EST
    Pig; You missed the word "ability". It didn't say attacks are on the rise; it said the ability to mount attacks has increased. How they measure that is unclear, but you're concluding something that wasn't stated. I also love the word "disbursed"!! How appropriate! Notice it didn't say "spent", or "invested", or "utilized"...it said disbursed!! Literally, the money's been handed out, but not necessarily spent or used on actual construction. Graft Central, is Iraq. Everyone's stealing from the Americans!

    Re: Iraq Update (none / 0) (#5)
    by pigwiggle on Thu Feb 09, 2006 at 10:01:20 AM EST
    Oh I see, they have he ability but aren't using it. Right ...

    Re: Iraq Update (none / 0) (#6)
    by desertswine on Thu Feb 09, 2006 at 10:14:58 AM EST
    Dick "Dick" Cheney on the insurgency:
    VICE PRESIDENT DICK CHENEY: I think that's fair. I don't think anybody thought -- JIM LEHRER: Why didn't we anticipate it? VICE PRESIDENT DICK CHENEY: Well, you can't anticipate everything. You know, we did anticipate a lot of things that didn't happen. We anticipated the possibility of a civil war between Sunni and Shia. That hasn't happened.
    Once again insert favorite Bush disaster: No one could have anticipated ___________. a) Crazies would fly airplanes into buildings. b) The levees would flood. c) The Iraqi insurgency. d) Your choice.

    Re: Iraq Update (none / 0) (#7)
    by Peaches on Thu Feb 09, 2006 at 10:16:53 AM EST
    The point of the article is the reporting of "Joseph A. Christoff, director of international affairs and trade at the Government Accountability Office, who testified before the Senate Foreign Affairs Committee." and his view "It's not going down." This is testimony enogh that shows our presence in Iraq is not benefitting us or the people of Iraq. There have been ebbs and flows in the numbers of attacks, but overall they are not going down--and may even appear to be increasing. Says Christoff, "There are peaks and valleys, but if you look at every peak, it's higher than the peak before." Is he basing this on statistical analysis? Of course not, but as Dylan said, "You don't need a weatherman to tell which way the wind blows."

    Re: Iraq Update (none / 0) (#8)
    by squeaky on Thu Feb 09, 2006 at 10:20:39 AM EST
    pw-
    Having looked at the number myself I would say that contrary to the administration and the NY Times, the attacks look to be relatively unchanged over the past three years.
    Got anything to back up your WH talking point. Links please.

    Re: Iraq Update (none / 0) (#9)
    by Talkleft Visitor on Thu Feb 09, 2006 at 10:52:42 AM EST
    Insurgency? CHENEY: Well, you can't anticipate everything... for GHWB it was a lack of "the vision thing." for GWB it's "a failure of imagination." Maybe W's apple fell closer to the tree than he'd like to think.

    Re: Iraq Update (none / 0) (#10)
    by jimakaPPJ on Thu Feb 09, 2006 at 11:24:24 AM EST
    et al - This is just usual guerilla stuff. Keep up the small attacks and wait for a political change in other side.. Hmmmm. Now where did we hear of that strategy???

    Re: Iraq Update (none / 0) (#11)
    by pigwiggle on Thu Feb 09, 2006 at 11:24:36 AM EST
    "Got anything to back up your WH talking point. Links please."
    Since when is contradicting the administration a WH talking point? Why don't you RTFA and look at the numbers yourself; your link is in the body of TL's post.
    Is he basing this on statistical analysis? Of course not, but as Dylan said, "You don't need a weatherman to tell which way the wind blows."
    Of course not, statistics are a waste for the true believers. But the rest of us need to get dirty with stuff like reason and critical thought to make sense of the world.

    Re: Iraq Update (none / 0) (#12)
    by squeaky on Thu Feb 09, 2006 at 11:42:04 AM EST
    pw-You are right. The WH is saying insurgent attacks are going down and you said the attacks have been the same for the last three years. Sorry, I realized that I was mistaken just as I pushed the 'Post' button. I was taken aback by your claim that the attacks have stayed the same for three years. Most of what I have read that there has been an increase in attacks. For example:
    According to US military statistics, 34,100 insurgent attacks mostly targeting US and Iraqi troops were recorded last year, up from about 27,000 in 2004, representing an increase of almost 30 percent.
    linkt

    Re: Iraq Update (none / 0) (#13)
    by squeaky on Thu Feb 09, 2006 at 11:46:40 AM EST
    oops..here is the link and this:
    Daily insurgent attacks, Feb 2004 14 Daily insurgent attacks, July 2005 70 Daily insurgent attacks, December 2005 75
    link

    Re: Iraq Update (none / 0) (#14)
    by Talkleft Visitor on Thu Feb 09, 2006 at 11:52:03 AM EST
    squeaky, those numbers need to be meta-analyzed before they can be installed with meaning.

    Re: Iraq Update (none / 0) (#15)
    by Peaches on Thu Feb 09, 2006 at 11:52:53 AM EST
    Of course not, statistics are a waste for the true believers. But the rest of us need to get dirty with stuff like reason and critical thought to make sense of the world.
    True believer in what? In statistics? Look in the mirror Pig, You're the truebeliever.

    Re: Iraq Update (none / 0) (#16)
    by pigwiggle on Thu Feb 09, 2006 at 12:51:40 PM EST
    "True believer in what? In statistics? Look in the mirror Pig, You're the truebeliever."
    No, I'm a consistent skeptic, but a pragmatic one. Before I go out on a limb and say something is one way or another I like to use tools with a substantial record; formal reason, mathematics. These make the trivial to vital possible; simple entertainment to life saving medical treatment. What do we owe your caprice and whim? I'm always surprised by folks' unrestrained anti-intellectual derision of science and mathematics. What's your beef?

    Re: Iraq Update (none / 0) (#17)
    by Talkleft Visitor on Thu Feb 09, 2006 at 12:57:21 PM EST
    the data look pretty straight-forward. But just out of curiosity, what kind of statistical analysis do you think would be helpful?

    Re: Iraq Update (none / 0) (#18)
    by Peaches on Thu Feb 09, 2006 at 01:26:36 PM EST
    I'm always surprised by folks' unrestrained anti-intellectual derision of science and mathematics. What's your beef?
    First of all you make a large sweeping assumption there. There was no unrestrained anti-intellectual derision of science and mathematics. M

    Re: Iraq Update (none / 0) (#19)
    by pigwiggle on Thu Feb 09, 2006 at 02:28:43 PM EST
    The NY Times pulled a fast one in their analysis. They note the insurgency began in April 04, and then analyze the percent change in attacks relative to March 04. Since attacks doubled from March 04 to April 04 their reported percent change in number of attacks is double what is should be. Well, or what it would be from the plain reading of the text. They could have normalized to June 03 for an effective quadrupling of the real percent change or August 04 and shown a relatively stable insurgency. Fast and loose.
    "the data look pretty straight-forward. ... But just out of curiosity, what kind of statistical analysis do you think would be helpful?"
    It think it looked like a noisy but essentially flat line; that is, from their projected insurgency beginning to now. But a very simple test that can be used to describe the relationship between two quantitative variables is regression. At first blush I would fit it with a straight line and check the quality of fit against the likelihood of accepting a positive slope (increasing insurgency) against some level of confidence. I think you'll find the data is far too noisy to make any prediction. Increasing, decreasing, stable, whatever; that's why I called it a Rorschach test, folks are going to see what validates their preconceived notion. Just like the NY Times.

    Re: Iraq Update (none / 0) (#20)
    by Sailor on Thu Feb 09, 2006 at 03:03:27 PM EST
    pw, if I get time tomorrow, (I intended to do it today but didn't have time), I'll plot the data and link to the resulting graph. I'm not sure about 'noisy' data, I was first going to plot coalition deaths per month. At any rate, I tihnk we can agree that we 'haven't turned a corner', the insurgency 'is in its last throes', or that there is an improvemnt in QoL for iraqis. Just today, from the BBC:
    The head of Israel's domestic security agency, Shin Bet, has said his country may come to regret the overthrow of former Iraqi President Saddam Hussein.

    Yuval Diskin said a strong dictatorship would be preferable to the present "chaos" in Iraq, in a speech to teenage Jewish settlers in the West Bank.


    Re: Iraq Update (none / 0) (#21)
    by Edger on Thu Feb 09, 2006 at 03:05:38 PM EST
    Sailor, vidoitspeak - 404 page not found

    Re: Iraq Update (none / 0) (#22)
    by Sailor on Thu Feb 09, 2006 at 03:08:17 PM EST
    Oh, and folks; PW and I rarely agree, but in my experience, he speaks with his own mind and doesn't follow the rnc talking points. We might disagree, but I think it to be an honest disagreement. BTW, pw, what would you suggest I use as my null hypo?

    Re: Iraq Update (none / 0) (#23)
    by Edger on Thu Feb 09, 2006 at 03:09:57 PM EST
    The head of Israel's domestic security agency, Shin Bet, has said his country may come to regret the overthrow of former Iraqi President Saddam Hussein. I imagine George is throwing temper tantrums and scaring the crap out of WH staffers today... hope he's got a bottle in the bottom drawer

    Re: Iraq Update (none / 0) (#24)
    by Edger on Thu Feb 09, 2006 at 03:11:28 PM EST
    Sailor: he speaks with his own mind and doesn't follow the rnc talking points Yes. I agree.

    Re: Iraq Update (none / 0) (#25)
    by Sailor on Thu Feb 09, 2006 at 03:13:20 PM EST
    Sorry edger, I'm not dyslexic, but my fingers are;-) The above link is now correct, and I would humbly recommend this post.

    Re: Iraq Update (none / 0) (#26)
    by Talkleft Visitor on Thu Feb 09, 2006 at 03:17:50 PM EST
    pw: I think you'll find the data is far too noisy to make any prediction. I see. You want to use the data to make a prediction. Given the degree of flux in this system, that seems unlikely. But as an indicator of progress and the status of our project, it's pretty clear and bad news. Said Joseph A. Christoff,
    director of international affairs and trade at the Government Accountability Office, who testified before the Senate Foreign Affairs Committee during a hearing on Iraq stabilization and reconstruction...
    "There are peaks and valleys, but if you look at every peak, it's higher than the peak before."


    Re: Iraq Update (none / 0) (#27)
    by Edger on Thu Feb 09, 2006 at 03:24:14 PM EST
    Very well said Requiem, Sailor... they would be proud of you. Fortunately I still have both of mine. Speaking of Coretta... there is much I miss from those years...

    Re: Iraq Update (none / 0) (#28)
    by Peaches on Thu Feb 09, 2006 at 03:30:27 PM EST
    I see. You want to use the data to make a prediction. Given the degree of flux in this system, that seems unlikely.
    What Pig subsribes to is that if you have a data set and it doesn't show a trend using statistical analysis (regression) than the data is not useful and you shouldn't say anything about it. This is as absurd as saying that if the data set did show a trend you would be able to make an accurate prediction.

    Re: Iraq Update (none / 0) (#29)
    by Talkleft Visitor on Thu Feb 09, 2006 at 03:34:16 PM EST
    This is as absurd as saying that if the data set did show a trend you would be able to make an accurate prediction. That's what I was getting at as well. Thanks for expressing it more clearly.

    Re: Iraq Update (none / 0) (#30)
    by Talkleft Visitor on Thu Feb 09, 2006 at 03:53:02 PM EST
    What Pig subsribes to is...
    Strange. All I thought Pig was saying was that you couldn't tell whether this particular graph showed an increase without doing a sound analysis of the data.
    "There are peaks and valleys, but if you look at every peak, it's higher than the peak before."
    That's not quite the way it looks to me. I onced browsed through a book on guidelines in the use of statistics. In the section on the use of bar graphs and pie charts it said "don't". The raw numbers are always more useful. How does the number of attacks relate to the strength of the insurgency? If that's the way to measure their strength all we have to do to defeat them is withdraw.

    Re: Iraq Update (none / 0) (#31)
    by Talkleft Visitor on Thu Feb 09, 2006 at 03:57:26 PM EST
    If that's the way to measure their strength all we have to do to defeat them is withdraw. Now you're starting to understand.

    Re: Iraq Update (none / 0) (#32)
    by pigwiggle on Thu Feb 09, 2006 at 04:30:13 PM EST
    Sailor-
    "BTW, pw, what would you suggest I use as my null hypo?"
    A slope less than or equal to zero, or alternately, a positive slope. I tried to find the raw data in the GAO report, but they just have the same graph the NY times reprinted. To be clear, my assertion is there is no significant change past April 04. I'm sure there would be a trend before. Peaches-
    "What Pig subsribes to is that if you have a data set and it doesn't show a trend using statistical analysis (regression) than the data is not useful and you shouldn't say anything about it."
    No, what I'm saying is that if you have a data set for which there is no statistically significant trend you should refrain from claiming a trend. Of course some (accurate) information is always more useful than none.
    "This is as absurd as saying that if the data set did show a trend you would be able to make an accurate prediction."
    Prediction was a poor choice of words. What I meant was that the data is likely too noisy to produce a statistically meaningful trend. The debate is if or not the insurgency has escalated, not if the trend will continue (the more interesting question).

    Re: Iraq Update (none / 0) (#33)
    by Talkleft Visitor on Thu Feb 09, 2006 at 05:34:56 PM EST
    pw: the data is likely too noisy to produce a statistically meaningful trend. Even relative to trend, without prediction, Christoff's comments were a reserved but reasonable read:
    "It's not going down," he said. "There are peaks and valleys, but if you look at every peak, it's higher than the peak before."


    Re: Iraq Update (none / 0) (#34)
    by Sailor on Thu Feb 09, 2006 at 05:56:12 PM EST
    pw, thanks for your reasoned and accurate responses. BTW, I wasn't asking for a stats lesson, I was asking what hypothesis you would use as a mean:mean. Before our invasion is unfair, as (type2) = 1. The first 3 months after is unfair, because neither type 1 nor type 2 have any meaning. So if I were to suggest 3 months of American deaths after bush pranced under the mission accomplished banner, would that make a fair null hypothesis? I really don't see it as multivariant, death is pretty much a zero (for) some game.*

    * I never knew it was possible to be proud and ashamed of a pun at the same time;-)

    Re: Iraq Update (none / 0) (#35)
    by Sailor on Thu Feb 09, 2006 at 06:17:20 PM EST
    OT, sorry all: edger, thanks, and what an elegant post you had. I keep reading and posting here because it shows me other viewpoints, helps define my response and whets my appetite;-) Now back to you regularly scheduled programming.

    Re: Iraq Update (none / 0) (#36)
    by Talkleft Visitor on Thu Feb 09, 2006 at 07:57:43 PM EST
    From the BBC news:
    The head of Israel's domestic security agency, Shin Bet, has said his country may come to regret the overthrow of former Iraqi President Saddam Hussein. Yuval Diskin said a strong dictatorship would be preferable to the present "chaos" in Iraq...


    Re: Iraq Update (none / 0) (#37)
    by pigwiggle on Fri Feb 10, 2006 at 06:33:24 AM EST
    "I was asking what hypothesis you would use as a mean:mean."
    You don't need to do it like that: just calculate a regression line using the method of least squares which will give some slope; calculate the standard error for this slope; using the assumption the random error is normally distributed you would then use the t-test; your general confidence interval is then [slope+/- t*(standard error)]. Or so my stats book says. It's hardly my subject so others may have more valuable advice. Anyway, I think you are wasting your time. I've looked at a lot of noisy data, some stuff I desperately wanted to be able to say something about. I'll be very surprised if there is enough there. The more interesting thing is the large dips around important events. Like the near halving of attacks following the January 30 national assembly election or the drop following the constitutional referendum. I would be interested in the January numbers following the December election of their first full term government. Maybe the insurgents shot their wad in the lead up to these events and needed time to restock, maybe local sentiment following these events made it more difficult for insurgents to operate, maybe the prior escalation is due to the US military seeking insurgent contact before the events and the subsequent calm is due to their putative cleanup. I see all of these possibilities as a good sign.

    Re: Iraq Update (none / 0) (#38)
    by Peaches on Fri Feb 10, 2006 at 07:19:06 AM EST
    Lets get back to
    What's your beef?
    My beef is that Pig is attempting to throw up a lot of smoke and dust so we don't get the message of this article while promoting his own agenda of painting the NYTimes as a liberal source while he is nuetral hiding behind the veil of statistical analysis and pretending he is doing science and mathemaitics when nothing of the sort is called for here. The question is pure and simple--objective journalism and has nothing to do with doing science or statistics. From his previous posts:
    Having looked at the number myself I would say that contrary to the administration and the NY Times, the attacks look to be relatively unchanged over the past three years. Absent any statistical analysis graphs are often no more useful than a Rorschach blot; you'll all see what you want.
    The NY Times pulled a fast one in their analysis.
    The substance of the article was not the graphs. As was pointed out earlier it was a reporting on a report given to the senate intelligence agency by Joseph Christoff. The Times then provided a link to the graphs with the data for individuals to draw from it what they wish. If one desired--and why anyone would is beyond me--one could even do some statistical analysis and run a regression. The more important conclusion from the article is the administrations claim that they are winning the war and that they have the insurgents on the run despite all reports to the contrary including reports from their officers in Iraq. The article never claimed that there was statisical analysis indicating a trend towards an increasing number of attacks. again: Joseph Christoff on insurgent attacks
    It's not going down.
    There are peaks and valleys, but if you look at every peak, it's higher than the peak before
    He is not claiming any trend. He is describing data on the number of attacks in Iraq by insurgents and the NYtimes does credible journalism by providing us with the numbers so we can draw our own conclusions.

    Re: Iraq Update (none / 0) (#39)
    by Talkleft Visitor on Fri Feb 10, 2006 at 07:39:18 AM EST
    Wait, did the guy who wrote this:
    the attacks look to be relatively unchanged over the past three years.
    and this:
    ...Increasing, decreasing, stable, whatever; that's why I called it a Rorschach test, folks are going to see what validates their preconceived notion.
    just write this?:
    The more interesting thing is the large dips around important events... I would be interested in the January numbers following the December election of their first full term government. Maybe the insurgents shot their wad... maybe local sentiment... maybe the prior escalation... I see all of these possibilities as a good sign.
    reading tea-leaves for a sign that things are groovy.

    Re: Iraq Update (none / 0) (#40)
    by pigwiggle on Fri Feb 10, 2006 at 08:09:11 AM EST
    Peaches-
    "My beef is that Pig is attempting to throw up a lot of smoke and dust so we don't get the message of this article while promoting his own agenda ... "
    No smoke and mirrors. I made one straightforward and verifiable statement ... but do you seriously think folks around here are so simple that my critique of this single point will somehow wipe clean any conclusions they may have gained from the entire text? Please.
    "The substance of the article was not the graphs. As was pointed out earlier it was a reporting on a report given to the senate intelligence agency by Joseph Christoff."
    How you or I interpret the 'substance' of the article isn't in question. I made a single comment about the kicker of the article. Mr. Glanz opens with "... portray a rebellion whose ability to mount attacks has steadily grown in the nearly three years since the invasion." I admit the truncation further exacerbates the already dubious interpretation; but it is T.Chris' doing, exactly what I was responding to in my original post. Don't try and shift this to other features of the article that I did not comment on, smoke and mirrors indeed.
    "The article never claimed that there was statisical analysis indicating a trend towards an increasing number of attacks."
    Ehhem ... yeah, that was my criticism. The article claimed, "Sweeping statistics ... appear to portray a rebellion whose ability to mount attacks has steadily grown ... ", but failed to support this bold, albeit qualified (appear) assertion. I suppose the qualification saves Mr. Glanz from an otherwise unsupported claim, but then why would we care about his impressions? That's hardly objective journalism, is it?
    "He is not claiming any trend. He is describing data on the number of attacks in Iraq by insurgents and the NYtimes does credible journalism by providing us with the numbers so we can draw our own conclusions."
    He, as in Mr. Christoff, did not claim a trend and quite aptly described the variation in attacks, as quoted. However, He, Mr. Glanz reporting for the NY Times, provided the numbers and characterized them as "... portray a rebellion whose ability to mount attacks has steadily grown ...". I didn't comment on Mr. Christoff's testimony.
    "... veil of statistical analysis and pretending he is doing science and mathemaitics when nothing of the sort is called for here."
    I think it is called for. If you want to claim a trend, show me a meaningful trend. If you want to illustrate my point that the raw numbers are more of a Rorschach exam, then knock yourself out with the Dylan quotes.

    Re: Iraq Update (none / 0) (#41)
    by pigwiggle on Fri Feb 10, 2006 at 08:12:35 AM EST
    Punisher- Right, it's open speculation. The difference is I'm not trying to pass it off as anything else. Nice try though.

    Re: Iraq Update (none / 0) (#42)
    by Peaches on Fri Feb 10, 2006 at 08:35:23 AM EST
    The article claimed, "Sweeping statistics ... appear to portray a rebellion whose ability to mount attacks has steadily grown ... ", but failed to support this bold, albeit qualified (appear) assertion. I suppose the qualification saves Mr. Glanz from an otherwise unsupported claim, but then why would we care about his impressions? That's hardly objective journalism, is it?
    It is precisely as objective as your own take on the data--and this is my beef. You are pretending to be more objective because of a reliance on regression when it is not needed here. In fact, the data does appear to portray a rebellion whose ability to mount attacks has steadily grown. No one who is paying attention to the War over there can deny that fact. Then we can get into the nitty gritty of the numbers and define the period of the term and draw whatever conclusions we want. However, the really important point, which you continue to dodge, is that the numbers do not indicate in any way that the insurgency is weakening no matter what time period you choose unless it is so short as to be insignificant--thus the description of peaks and valleys. But, thats it for me--I'm done. Wait
    Its a hard, Its a hard, Its a hard Its a hard rain--gonna fall.


    Re: Iraq Update (none / 0) (#43)
    by Talkleft Visitor on Fri Feb 10, 2006 at 09:08:34 AM EST
    pw: I see all of these possibilities as a good sign. I'm actually happy for you that you can keep your rosy outlook. pw: it's open speculation. The difference is I'm not trying to pass it off as anything else. You don't dispute Christoff's
    "There are peaks and valleys, but if you look at every peak, it's higher than the peak before."
    but you take issue with Glanz's
    "portray a rebellion whose ability to mount attacks has steadily grown"
    I guess you see some major discrepency between the two. If you're taking issue with the use of the word "steadily" then (IMO) you're splitting hairs, and I'm sorry we've spent time discussing this. Whatever, it's cute that this thread has turned into a mini-symposium on stats, an important field, as you've noted. But I think I'll drop it here.

    Re: Iraq Update (none / 0) (#44)
    by pigwiggle on Fri Feb 10, 2006 at 09:21:56 AM EST
    Peaches-
    "It is precisely as objective as your own take on the data--and this is my beef. You are pretending to be more objective ..."
    I'm not pretending to be anything other than some guy blathering on a blog. And just why would you expect a standard of objectivity that even your Mr. Glanz failed to rise to, anyway? How's this; when I start getting NY Times salary for my posts I'll take them a bit more seriously.
    "... he really important point, which you continue to dodge, is that the numbers do not indicate in any way that the insurgency is weakening"
    I'm not 'dodging' the point; I agree (I'll refer you to yesterday's 10:34 post, 3:28 post, and 5:30 post, today's 7:33 and 9:09 posts, oh, and I guess this one too). Should I say it a seventh time?
    "and this is my beef."
    Whatever. You just wanted to pick a fight because I questioned these numbers as proof of your conviction.

    Re: Iraq Update (none / 0) (#45)
    by pigwiggle on Fri Feb 10, 2006 at 09:30:57 AM EST
    Punisher- It is appropriate for Mr. Christoff, GAO director of international affairs and trade, to opine in his congressional testimony; indeed, he was asked to do so. Although I will point out that he was obviously wrong, each peak is not higher than the next. It is inappropriate for Mr. Glanz, NY Times reporter, to offer his opinion about the appearance of some data, or what he imagines some data to portray. This wasn't an editorial piece.

    Re: Iraq Update (none / 0) (#46)
    by Talkleft Visitor on Fri Feb 10, 2006 at 09:59:34 AM EST
    pw: inappropriate... This wasn't an editorial piece. fair enough. Maybe, as Atrios likes to say, time for a blogger ethics panel.

    Re: Iraq Update (none / 0) (#47)
    by Peaches on Fri Feb 10, 2006 at 10:22:16 AM EST
    It's Friday and I'm bored at work. Plus it is always hard to tear oneself away from a debate. The constant shifting of ground from me and my interlocuter is always entertaining enough in itself.
    It is inappropriate for Mr. Glanz, NY Times reporter, to offer his opinion about the appearance of some data, or what he imagines some data to portray. This wasn't an editorial piece.
    But Mr. Glanz was simply reporting on the testimony of Mr. Christoff. Journalist should not interject their opinion, but they can certainly summarize the contents of a testimony. The graph provided by the Times were the same graphs Mr. Christoff used in his testimony, as you indicated
    I tried to find the raw data in the GAO report, but they just have the same graph the NY times reprinted
    So, certainly the Times wasn't showing any bias. I have no reason to disbelieve Glanz that the testimony gave a general viewpoint that
    Sweeping statistics ... appear to portray a rebellion whose ability to mount attacks has steadily grown
    Which he supports with some quotes from Christoff. That, my friend, is objective journalism.

    Re: Iraq Update (none / 0) (#48)
    by Sailor on Fri Feb 10, 2006 at 11:23:32 AM EST
    As promised here is a link to a chart of coalition military deaths, by month, since May 2003 when Bush spoke under the banner 'Mission Accomplished.' The numbers are from the DoD, and a simple linear regression shows a rather pronounced trend upwards.

    Re: Iraq Update (none / 0) (#49)
    by Sailor on Fri Feb 10, 2006 at 11:58:25 AM EST
    Oops, damn blogger, here's the link.

    Re: Iraq Update (none / 0) (#50)
    by pigwiggle on Fri Feb 10, 2006 at 01:10:22 PM EST
    Interesting; I used DOD data and found a completely different trend. As the GAO noted the data is incomplete before February 04, so I've excluded the incomplete points. Take a look. I found a slope of -0.2 +/- 0.5 deaths/week using a 90% confidence interval. Even with this relatively loose interval we can't say if deaths are increasing or decreasing. I could go back and try and work with the incomplete set but the sparse data will only increase the error. I put the xls sheet in the same directory. Knock yourself out.

    Re: Iraq Update (none / 0) (#51)
    by pigwiggle on Fri Feb 10, 2006 at 01:24:36 PM EST
    Well, I went back and reworked the sheet simply dropping the missing points. The slope is then 0.7 +/- 0.6 deaths/week. Using a 95% confidence interval the number is 0.7 +/- 0.7 deaths/week. Anyway, I'm starting to feel rather stupid beating this like we have. I'm done.

    Re: Iraq Update (none / 0) (#52)
    by pigwiggle on Fri Feb 10, 2006 at 01:47:18 PM EST
    Man, I botched that. The real numbers are 0.04 +/- 0.04 at 95% CI for all data, 0.04 +/- 0.03 at 90% CI for all data, and -0.03 +/- 0.06 at 90% CI for only the complete portion.Oh, and I used the Excel intrinsic functions to check these. They are correct this time. Sorry.

    Re: Iraq Update (none / 0) (#53)
    by Talkleft Visitor on Fri Feb 10, 2006 at 03:20:24 PM EST
    That, my friend, is objective journalism.
    Christoff:
    It's not going down
    Glanz:
    steadily grown
    Those two statements aren't consistent. Even the graph provided by the New York Times is inconsistent with the term "steadily", let alone "grown". Glanz:
    nearly three years
    Yet the graph, and presumably Mr. Christoff's report, only discuss the last two years (almost). "How to Lie with Statistics" should be required reading. This article would make, I think, an excellent example, particularly the last four paragraphs. I consider myself a liberal, but propaganda from the left is still propaganda. On the other hand, the discussion between pigwiggle and Sailor has been very interesting.

    Re: Iraq Update (none / 0) (#54)
    by Sailor on Fri Feb 10, 2006 at 07:29:43 PM EST
    PW, I don't see any basis for your numbers. And I really can't understand how you can say that the nhumber of American and other coalition forces was 'incomplete' before February 04. BTW, there were 132 American casualities in Nov 04! You numbers are way off. Here is a link to them and another link. The best one is here. Because you can download the xml and grep it for import into your stats package.

    Re: Iraq Update (none / 0) (#55)
    by pigwiggle on Sat Feb 11, 2006 at 07:21:41 AM EST
    Sailor- The numbers are directly from the DoD, and are only for US deaths due to hostilities in Iraq.
    "BTW, there were 132 American causalities in Nov 04! You numbers are way off."
    You are confused by the format. They are plotted by the week, not month. If you add the four weeks in Nov. 04 the deaths total 130. The links you provided use data from the DoD site I sourced.
    "And I really can't understand how you can say that the nhumber of American and other coalition forces was 'incomplete' before February 04."
    For whatever reason the DoD doesn't have numbers for several weeks prior to February 04. Perhaps they know when these soldiers died; certainly they know the number of dead. But if we use data prior to these dates we run the risk of using non-sequential deaths, which will ruin the analysis. Anyway, I did it both ways and am satisfied we can't say there is a meaningful trend, either way. My excel sheet is in the parent directory of the graph jpeg. Careful with it, it's the incorrect one where I transposed the x and y columns in the slope and error analysis.

    Re: Iraq Update (none / 0) (#56)
    by squeaky on Sat Feb 11, 2006 at 07:28:03 AM EST
    PW-the statistics here are above my head so this point may be irrelevent. In terms of numbers of dead soldiers, aren't the numbers only counting soldiers who died in Iraq? What is left out are the soldiers that died from wounds after being airlifted out of Iraq. Not a small number I imagine.

    Re: Iraq Update (none / 0) (#57)
    by Talkleft Visitor on Sat Feb 11, 2006 at 08:10:16 AM EST
    pw made a reasonable point early in this thread, and thereafter, that if a data set is to be best understood it should be subjected to an appropriate statistical analysis. But I think that we've reached a point in this discussion where one can suggest that wonkish debates about slopes, confidence intervals, and the null hypothesis is at least in part a way to avoid having to think directly about the awfulness, suffering, and death that the numbers represent, irrespective of determinable trend or forecast.

    Re: Iraq Update (none / 0) (#58)
    by Sailor on Sat Feb 11, 2006 at 08:36:45 AM EST
    pw, sorry for doubting your numbers, I didn't understand the sampling.

    Re: Iraq Update (none / 0) (#59)
    by pigwiggle on Sat Feb 11, 2006 at 09:47:34 AM EST
    Squeaky-
    "In terms of numbers of dead soldiers, aren't the numbers only counting soldiers who died in Iraq?"
    Right, but if we make the assumption that the proportion is fixed; for example, if 90% of the dead died in Iraq and 10% died elsewhere, then it wont change the trend. However, if the proportion isn't fixed for some reason, say the military is steadily increasing the number of dieing soldiers evacuated, i.e. in 2004 10% died elsewhere and in 2005 20% died elsewhere, then the trend of deaths in Iraq will be unrepresentative of the true trend.
    "pw, sorry for doubting your numbers, I didn't understand the sampling."
    You should doubt them, skepticism is always a good thing. Punisher-
    "a way to avoid having to think directly about the awfulness, suffering, and death that the numbers represent,"
    A thought that made me question if deaths were the best metric. If the insurgency is truly rising and attacks are increasing, then injuries due to IEDs and the like may be up while deaths are not. Incidentally, my wife occasionally rotates through he VA; she is starting to see a number of young amputees. You know, the VA is a real bad place to get care. Her patients there are those that are too poor or too crazy and poor to go elsewhere. Real substandard care.

    Re: Iraq Update (none / 0) (#60)
    by squeaky on Sat Feb 11, 2006 at 09:53:50 AM EST
    PW-
    You know, the VA is a real bad place to get care....Real substandard care.
    And Bush is determined to make it worse. Read and weep.

    Re: Iraq Update (none / 0) (#61)
    by Talkleft Visitor on Sat Feb 11, 2006 at 10:08:40 AM EST
    pw: ...the VA is a real bad place to get care. I've done my time at the VA as well, and can't entirely agree with this assessment. It's something, and better than nothing, but not nearly what our vets deserve. And worse, though this is OT, NavyTimes reported last week that under FY 2007 budget plans, vet retirees under age 65 are getting hit by substantial increases in enrollment and deductible costs. But as disappointing as the resources often were, the vets that I had the privilege and opportunity to work with were most often grateful to the VA and proud to get their care there.

    Re: Iraq Update (none / 0) (#62)
    by Talkleft Visitor on Sat Feb 11, 2006 at 10:34:27 AM EST
    pw: A thought that made me question if deaths were the best metric. An important metric, no doubt. But looking at mortality only, and not looking at morbidity makes no sense.

    Re: Iraq Update (none / 0) (#63)
    by Sailor on Sat Feb 11, 2006 at 10:48:43 AM EST
    Hi PW, I ran the numbers again using your time period but my monthly sampling and came up with your trend. I do think the pre 3/04 data is accurate as a monthly sample, and I charted 1/04 to 12/05 and separately for 1/05-12/05. Both show an upward trend. Amazing how sampling choices affect outcomes. The wounded would probably be a better metric, but I was trying to avoid anything but binary data.

    Re: Iraq Update (none / 0) (#64)
    by pigwiggle on Sat Feb 11, 2006 at 04:42:57 PM EST
    From what I understand the VA was put in place to handle large volumes of casualties during wartime. During WWII for example, the average death rate for US soldiers was ~ 200/day; I don't know the injury rate but it must be at least as large. The private healthcare infrastructure was unable to cope with the volume of wounded soldiers. So we have the VA. It really wasn't meant for quality long-term healthcare. My wife hates working at the VA. She can't prescribe certain brands or dosages of medication; their computer system is unable to handle non-uniform requests. Often the medications she prescribes are substituted with another she specifically avoided. The quality of staff and their moral is typical of what you would expect of a bloated entrenched inflexible bureaucracy. If we are going to provide soldiers with long-term healthcare as a benefit they should get at least the quality typically offered a professional through an employee benefit package. Sq-
    "Amazing how sampling choices affect outcomes."
    It is. Folks play games with this all the time. See my previous comments about the percent change graph the NY Times printed. I think it's theirs; I don't recall seeing it in the GAO reprint.

    Re: Iraq Update (none / 0) (#65)
    by Talkleft Visitor on Sat Feb 11, 2006 at 05:07:33 PM EST
    pw: My wife hates working at the VA. She can't prescribe certain brands or dosages of medication; their computer system is unable to handle non-uniform requests. Often the medications she prescribes are substituted with another she specifically avoided. The quality of staff and their moral is typical of what you would expect of a bloated entrenched inflexible bureaucracy. The VA could be improved, but I sure don't hate it, and I'm proud of every minute that I spent in it. Limitations on med options is a drag, though Kaiser patients get the same. And any time that I wanted to prescribe off-formulary, all I needed to do was fill out a simple form and fax it to the pharmacy. I don't think I was ever rejected. Our staff was great, physicians and nurses were top-notch. Everyone was over-worked, just like every hospital that I've worked in. Moral was better there than in some other places I've been, maybe because there was a sense that we were helping the guys who've done so much for us. It's been about four years since I've been in a VA hospital though, and I'm sure that GWB could've inflicted plenty of pain on the system in that time.

    Re: Iraq Update (none / 0) (#66)
    by pigwiggle on Sun Feb 12, 2006 at 08:15:23 AM EST
    Punisher- Well, I guess my point is that the VA has become outdated. We don't fight wars with a WWII like volume of casualties. The private domestic healthcare system is more than capable of providing daily long-term healthcare. Heck, soldiers don't even get VA care unless their condition is service related. I say we provide comprehensive insurance similar to what any other professional might have for all active duty, and life-time coverage for those who have seen combat. This would remove the problem of access vets have; most everyone has a hospital in their town, few have a VA. It would also increase the quality, as folks will now have a choice. Some do have a choice, they fund their care elsewhere. When vets in my town eschew free care, emperically I know something is wrong here.

    Re: Iraq Update (none / 0) (#67)
    by pigwiggle on Sun Feb 12, 2006 at 08:16:16 AM EST
    and by outdated I mean redundant