Iraq Death Toll?
Nadagast
Join Date: 2002-11-04 Member: 6884Members
Join Date: 2002-11-04 Member: 6884Members
in Discussions
<div class="IPBDescription">to not derail the other thread</div>This is a new thread for the discussion about the Lancet study between me and Marine01. I'll start off with my last post in the other thread <img src="style_emoticons/<#EMO_DIR#>/smile-fix.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile-fix.gif" />
<!--quoteo(post=1571469:date=Oct 27 2006, 03:34 AM:name=Marine0I)--><div class='quotetop'>QUOTE(Marine0I @ Oct 27 2006, 03:34 AM) [snapback]1571469[/snapback]</div><div class='quotemain'><!--quotec-->I really dont think you understand what an argument from incredulity is. It is also more commonly described as an argument from ignorance - basically ignorance is all that leads a person to disbelieve a certain claim.<!--QuoteEnd--></div><!--QuoteEEnd-->
No... I do.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->When IBC claims that Lancet has stuffed up, it goes on to show why. For example:
Their claims of extrapolation just dont fit the facts. They found 91 deaths from between January and June 2006, which extrapolated all the way up to 180,000 deaths. Thats about 1000 deaths per day. Current death reports pick about 100 deaths per day. Basically, the media and the government and everyone involved over there is either viciously imcompetent or covering it up, because its borderline impossible to miss an extra 900 deaths per day.<!--QuoteEnd--></div><!--QuoteEEnd-->
This is like the definition of argument from incredulity. How many times do I have to ask you to find problems with their method (NOT their results)?
Not to mention, the study cites several other studies that show that passive surveillance (which the IBC and the Iraq government are doing) commonly underestimate deaths by a factor of ~10. I've mentioned this before and I haven't even seen you comment on it yet.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->But it gets worse from then on. There is an old standard for estimating casualties which puts the amount of wounded for every person killed at 3 for every death. This means that not only do you have are huge swath of the Iraqi population exterminated without the media or the officials knowing, but you have 800,000 Iraqi's who were wounded who didn't seek medical help. This is on top of the deaths. It's not possible for half a million excess deaths to just hide away, its even less possible for a further 740,000 people to simply not seek medical help.<!--QuoteEnd--></div><!--QuoteEEnd-->
Ok, so you're saying that a peer reviewed study is wrong because of some random rule of thumb that I've never seen before? Sounds solid to me...
Also, the hospitals don't seem to be a good place to go:
<a href="http://www.cbsnews.com/stories/2006/10/04/eveningnews/main2064668.shtml" target="_blank">http://www.cbsnews.com/stories/2006/10/04/...in2064668.shtml</a>
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->And if they are seeking medical help, why are Iraqi hospitals only reporting 60,000 people over the last 2 years asking for medical help? Is it in their best interest to tell the Government "hey guys, we dont need more help or funding, we're handling fine" when masses of wounded Iraqi's are pouring through their doors daily. Or is there some kind of backyard medical craze going on in Iraq I dont know about? Despite the decayed state of their nation, do you really think the hospitals would just have failed to notice overwhelming numbers of violence victims like that?<!--QuoteEnd--></div><!--QuoteEEnd-->
<a href="http://www.cbsnews.com/stories/2006/10/04/eveningnews/main2064668.shtml" target="_blank">http://www.cbsnews.com/stories/2006/10/04/...in2064668.shtml</a>
Again this really boils down to an argument from incredulity. You personally find it implausible that in a chaotic nation, we might receive numbers that are wrong, or that people won't go to hospitals because of the death squads. When are you going to actually critique their method and not the results? Do you understand how science works? Maybe I'm wrong, but as I understand it, to cast doubt on a study, you critique their method, or you repeat the same method and get different results. It's not about saying, "I personally find it implausible that this could be happening. So the study is wrong. Somehow."
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->What's going on here? If its reasonable to extrapolate upwards as these researchers did,<!--QuoteEnd--></div><!--QuoteEEnd-->
Yes it is, unless you object to the field of statistics.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->where are those extra 500,000 death certificates coming from? We treat their word as reliable because they are relying on Iraqi's producing death certificates.... but there <b>aren't that many death certificates in the entire country by a factor of 10</b>.<!--QuoteEnd--></div><!--QuoteEEnd-->
I don't know where the 500k extra death certificates came from, but it just might be because of:
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->the morgue itself is believed to be controlled by the same Shiite militia blamed for many of the killings: the Mahdi Army, founded and led by anti-American cleric Moqtada al-Sadr.<!--QuoteEnd--></div><!--QuoteEEnd-->
According to the CBS news article I linked.
Also, remember that many previous studies have shown that surveillance like this consistently underestimates casualties by a large factor. <-- YOU HAVE NOT ADDRESSED THIS. WHY?
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->If you take a random sample of cars going into an empty car lot per hour, and decide from that that there must be 100 cars inside, but you find only 10 when you do a count inside, you cant talk about how your methodology was perfect - it aint. Its failed. Try again.<!--QuoteEnd--></div><!--QuoteEEnd-->
You have yet to show me why the methodology is wrong.
And again, passive surveillance always underestimates casualties. It's nothing new.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->It just gets worse and worse for the Lancet. Maybe one of these factors could be ignored individually, but all up it leads to one inevitable conclusion... there is simply far too much evidence against their figures being true for them to be treated with any credibility at all.<!--QuoteEnd--></div><!--QuoteEEnd-->
Are you serious? I'm going to object to Quantum Mechanics because I don't like the results of the theory. Would you believe me or call me crazy?
What evidence? A bunch of passive surveillance that <b>we already know underestimates casualties by a large factor</b>. Not to mention Bush's unbelievable hypocrisy in stating that the method is discredited while his own government spends millions a year training people to do the exact same thing.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->I'm sorry, but there is a very painful life lesson to be learned here. <b>Scientists and researchers are just you like you and me</b>. They have biases, they make mistakes, they refuse to capitulate to evidence, and not even peer review can proof them from it. There may be more or less deaths in Iraq than we know, but Lancet's study doesnt help at all, because its clearly wrong.<!--QuoteEnd--></div><!--QuoteEEnd-->
I'm sure that scientists make mistakes and have biases, BUT YOU HAVEN'T SHOWN ME A SINGLE ONE IN THIS STUDY. All you've done is objected to the results and made like 50 arguments from incredulity.
<!--sizeo:5--><span style="font-size:18pt;line-height:100%"><!--/sizeo-->In your next post please find something wrong with their method. Not their results...<!--sizec--></span><!--/sizec-->
<!--quoteo(post=1571469:date=Oct 27 2006, 03:34 AM:name=Marine0I)--><div class='quotetop'>QUOTE(Marine0I @ Oct 27 2006, 03:34 AM) [snapback]1571469[/snapback]</div><div class='quotemain'><!--quotec-->I really dont think you understand what an argument from incredulity is. It is also more commonly described as an argument from ignorance - basically ignorance is all that leads a person to disbelieve a certain claim.<!--QuoteEnd--></div><!--QuoteEEnd-->
No... I do.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->When IBC claims that Lancet has stuffed up, it goes on to show why. For example:
Their claims of extrapolation just dont fit the facts. They found 91 deaths from between January and June 2006, which extrapolated all the way up to 180,000 deaths. Thats about 1000 deaths per day. Current death reports pick about 100 deaths per day. Basically, the media and the government and everyone involved over there is either viciously imcompetent or covering it up, because its borderline impossible to miss an extra 900 deaths per day.<!--QuoteEnd--></div><!--QuoteEEnd-->
This is like the definition of argument from incredulity. How many times do I have to ask you to find problems with their method (NOT their results)?
Not to mention, the study cites several other studies that show that passive surveillance (which the IBC and the Iraq government are doing) commonly underestimate deaths by a factor of ~10. I've mentioned this before and I haven't even seen you comment on it yet.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->But it gets worse from then on. There is an old standard for estimating casualties which puts the amount of wounded for every person killed at 3 for every death. This means that not only do you have are huge swath of the Iraqi population exterminated without the media or the officials knowing, but you have 800,000 Iraqi's who were wounded who didn't seek medical help. This is on top of the deaths. It's not possible for half a million excess deaths to just hide away, its even less possible for a further 740,000 people to simply not seek medical help.<!--QuoteEnd--></div><!--QuoteEEnd-->
Ok, so you're saying that a peer reviewed study is wrong because of some random rule of thumb that I've never seen before? Sounds solid to me...
Also, the hospitals don't seem to be a good place to go:
<a href="http://www.cbsnews.com/stories/2006/10/04/eveningnews/main2064668.shtml" target="_blank">http://www.cbsnews.com/stories/2006/10/04/...in2064668.shtml</a>
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->And if they are seeking medical help, why are Iraqi hospitals only reporting 60,000 people over the last 2 years asking for medical help? Is it in their best interest to tell the Government "hey guys, we dont need more help or funding, we're handling fine" when masses of wounded Iraqi's are pouring through their doors daily. Or is there some kind of backyard medical craze going on in Iraq I dont know about? Despite the decayed state of their nation, do you really think the hospitals would just have failed to notice overwhelming numbers of violence victims like that?<!--QuoteEnd--></div><!--QuoteEEnd-->
<a href="http://www.cbsnews.com/stories/2006/10/04/eveningnews/main2064668.shtml" target="_blank">http://www.cbsnews.com/stories/2006/10/04/...in2064668.shtml</a>
Again this really boils down to an argument from incredulity. You personally find it implausible that in a chaotic nation, we might receive numbers that are wrong, or that people won't go to hospitals because of the death squads. When are you going to actually critique their method and not the results? Do you understand how science works? Maybe I'm wrong, but as I understand it, to cast doubt on a study, you critique their method, or you repeat the same method and get different results. It's not about saying, "I personally find it implausible that this could be happening. So the study is wrong. Somehow."
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->What's going on here? If its reasonable to extrapolate upwards as these researchers did,<!--QuoteEnd--></div><!--QuoteEEnd-->
Yes it is, unless you object to the field of statistics.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->where are those extra 500,000 death certificates coming from? We treat their word as reliable because they are relying on Iraqi's producing death certificates.... but there <b>aren't that many death certificates in the entire country by a factor of 10</b>.<!--QuoteEnd--></div><!--QuoteEEnd-->
I don't know where the 500k extra death certificates came from, but it just might be because of:
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->the morgue itself is believed to be controlled by the same Shiite militia blamed for many of the killings: the Mahdi Army, founded and led by anti-American cleric Moqtada al-Sadr.<!--QuoteEnd--></div><!--QuoteEEnd-->
According to the CBS news article I linked.
Also, remember that many previous studies have shown that surveillance like this consistently underestimates casualties by a large factor. <-- YOU HAVE NOT ADDRESSED THIS. WHY?
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->If you take a random sample of cars going into an empty car lot per hour, and decide from that that there must be 100 cars inside, but you find only 10 when you do a count inside, you cant talk about how your methodology was perfect - it aint. Its failed. Try again.<!--QuoteEnd--></div><!--QuoteEEnd-->
You have yet to show me why the methodology is wrong.
And again, passive surveillance always underestimates casualties. It's nothing new.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->It just gets worse and worse for the Lancet. Maybe one of these factors could be ignored individually, but all up it leads to one inevitable conclusion... there is simply far too much evidence against their figures being true for them to be treated with any credibility at all.<!--QuoteEnd--></div><!--QuoteEEnd-->
Are you serious? I'm going to object to Quantum Mechanics because I don't like the results of the theory. Would you believe me or call me crazy?
What evidence? A bunch of passive surveillance that <b>we already know underestimates casualties by a large factor</b>. Not to mention Bush's unbelievable hypocrisy in stating that the method is discredited while his own government spends millions a year training people to do the exact same thing.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->I'm sorry, but there is a very painful life lesson to be learned here. <b>Scientists and researchers are just you like you and me</b>. They have biases, they make mistakes, they refuse to capitulate to evidence, and not even peer review can proof them from it. There may be more or less deaths in Iraq than we know, but Lancet's study doesnt help at all, because its clearly wrong.<!--QuoteEnd--></div><!--QuoteEEnd-->
I'm sure that scientists make mistakes and have biases, BUT YOU HAVEN'T SHOWN ME A SINGLE ONE IN THIS STUDY. All you've done is objected to the results and made like 50 arguments from incredulity.
<!--sizeo:5--><span style="font-size:18pt;line-height:100%"><!--/sizeo-->In your next post please find something wrong with their method. Not their results...<!--sizec--></span><!--/sizec-->
Comments
<a href="http://www.thestar.com/NASApp/cs/ContentServer?pagename=thestar/Layout/Article_Type1&c=Article&cid=1161814211186&call_pageid=970599119419" target="_blank">http://www.thestar.com/NASApp/cs/ContentSe...id=970599119419</a>
That first group must have been extremely guillble ignorants, I call on this test the sheenanigans.
<!--QuoteEnd--></div><!--QuoteEEnd-->
What? ... Elaborate please <img src="style_emoticons/<#EMO_DIR#>/smile-fix.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile-fix.gif" />
He means Moultano's link. Basically, another argument from incredulity. "I can't believe this study is true, so I declare it false."
<!--QuoteEnd--></div><!--QuoteEEnd-->
Ah yeah I thought he was talking about the Lancet study.
Yeah obviously you know better than those tricky psychologists Epidemic. Good work. Any other cognitive biases that you can disprove with simply a hand wave?
<a href="http://en.wikipedia.org/wiki/List_of_cognitive_biases" target="_blank">http://en.wikipedia.org/wiki/List_of_cognitive_biases</a>
PS. Don't let this derail this thread! What's wrong with the method in the Lancet study? I want to know!
The rebuttal you're looking for is aptly described <a href="http://news.bbc.co.uk/2/hi/middle_east/6045112.stm" target="_blank">in this report from the BBC.</a>
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->An article in the Wall Street Journal by Steven Moore, who worked as a pollster for the coalition authorities in Iraq, attacked the sampling: "The key to the validity of cluster sampling is to use enough cluster points. In their 2006 report... the Johns Hopkins team says it used 47 cluster points for their sample of 1,849 interviews. This is astonishing: I wouldn't survey a junior high school, no less an entire country, using only 47 cluster points."
And on 20 October, Science Magazine reported the queries of researchers at Oxford and Royal Holloway universities. One of them, Sean Gourley of the Physics department at Oxford, said their studies "have found fundamental flaws [in the Lancet report] that lead to an over-estimation of the number of deaths. "
One aspect they questioned was the selection of sample households chosen for interviews. There could be "main street bias", they said, in that households on main streets were more likely to suffer casualties from car bombings. They want an inquiry into the methodology. "It's almost a crime to let it go unchallenged," said Neil Johnson of Oxford.
And other groups that track deaths in Iraq dispute the findings. Michael O'Hanlon of the Brookings Institution, which tracks statistics in its Iraq Index, said: "I do not believe the new numbers. I think they're way off." The Brooking Index, relying on the UN (which gets figures from the Iraqi health ministry) and the Iraq Body Count (IBC), estimates the civilian death toll at about 62,000. <!--QuoteEnd--></div><!--QuoteEEnd-->
When asked why the physical evidence of bodies cannot be verified, one of the authors conveniently notes that it's too dangerous to travel the countryside counting corpses. Sounds like Captain Super Science to me.
BTW, here's a clip of one of the Lancet editors sharing his thoughts during a "gathering": <a href="http://www.youtube.com/watch?v=v7BzM5mxN5U" target="_blank">Dr. Richard Horton.</a> In light of those ideals, it's clear we can trust him to provide unbiased scientific data.
The rebuttal you're looking for is aptly described <a href="http://news.bbc.co.uk/2/hi/middle_east/6045112.stm" target="_blank">in this report from the BBC.</a><!--QuoteEnd--></div><!--QuoteEEnd-->
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->An article in the Wall Street Journal by Steven Moore, who worked as a pollster for the coalition authorities in Iraq, attacked the sampling: "The key to the validity of cluster sampling is to use enough cluster points. In their 2006 report... the Johns Hopkins team says it used 47 cluster points for their sample of 1,849 interviews. This is astonishing: I wouldn't survey a junior high school, no less an entire country, using only 47 cluster points."<!--QuoteEnd--></div><!--QuoteEEnd-->
This just doesn't make sense... The sample size is reflected in the 95% confidence interval.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->And on 20 October, Science Magazine reported the queries of researchers at Oxford and Royal Holloway universities. One of them, Sean Gourley of the Physics department at Oxford, said their studies "have found fundamental flaws [in the Lancet report] that lead to an over-estimation of the number of deaths. "
One aspect they questioned was the selection of sample households chosen for interviews. There could be "main street bias", they said, in that households on main streets were more likely to suffer casualties from car bombings. They want an inquiry into the methodology. "It's almost a crime to let it go unchallenged," said Neil Johnson of Oxford.<!--QuoteEnd--></div><!--QuoteEEnd-->
The way I read the study, they chose a main street then chose a residential street off of it. Which would imply that none of the houses were on a main street. No? Also, this was done randomly, which is entirely standard practice for choosing samples.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->And other groups that track deaths in Iraq dispute the findings. Michael O'Hanlon of the Brookings Institution, which tracks statistics in its Iraq Index, said: "I do not believe the new numbers. I think they're way off." The Brooking Index, relying on the UN (which gets figures from the Iraqi health ministry) and the Iraq Body Count (IBC), estimates the civilian death toll at about 62,000.<!--QuoteEnd--></div><!--QuoteEEnd-->
IBC does not measure total deaths in Iraq, that's why the numbers don't match. Like I've said before, passive surveillance like IBC underestimates deaths by large factors, in situations like Iraq.
I've never heard of The Brooking Index, but the Iraqi health ministry numbers are likely to be inaccurate, for the same reasons, plus: <a href="http://www.cbsnews.com/stories/2006/10/04/eveningnews/main2064668.shtml" target="_blank">http://www.cbsnews.com/stories/2006/10/04/...in2064668.shtml</a>
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->When asked why the physical evidence of bodies cannot be verified, one of the authors conveniently notes that it's too dangerous to travel the countryside counting corpses. Sounds like Captain Super Science to me.<!--QuoteEnd--></div><!--QuoteEEnd-->
I can't imagine traveling around counting corpses would be a good way to calculate death totals. Most people are buried... in the ground...
Why add the insult at the end? Is that necessary?
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->BTW, here's a clip of one of the Lancet editors sharing his thoughts during a "gathering": <a href="http://www.youtube.com/watch?v=v7BzM5mxN5U" target="_blank">Dr. Richard Horton.</a> In light of those ideals, it's clear we can trust him to provide unbiased scientific data.<!--QuoteEnd--></div><!--QuoteEEnd-->
Yes he has a political opinion, but he didn't write the report. I don't see how he could change their numbers without writing (falsifying) the report. What exactly are you accusing him of doing?
Edit: I'm going to try to find a free link to one of the papers that discusses the underestimates of passive surveillance.
The first two I looked for I found. Haven't had a chance to look at them yet, I have to go. But here they are:
<a href="http://www.who.int/infectious-disease-news/IDdocs/Lancet_CDs_complex_emergencies.pdf" target="_blank">http://www.who.int/infectious-disease-news...emergencies.pdf</a>
<a href="http://www.cdc.gov/MMWR/preview/mmwrhtml/mm5220a3.htm" target="_blank">http://www.cdc.gov/MMWR/preview/mmwrhtml/mm5220a3.htm</a>
This just doesn't make sense... The sample size is reflected in the 95% confidence interval.<!--QuoteEnd--></div><!--QuoteEEnd-->
Cluster size does not equal sample size.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->The way I read the study, they chose a main street then chose a residential street off of it. Which would imply that none of the houses were on a main street. No? Also, this was done randomly, which is entirely standard practice for choosing samples.<!--QuoteEnd--></div><!--QuoteEEnd-->
If I walk down MLK blvd in Detroit and interview every house on the corner to see if anyone has witnessed gun violence can I then say that all of Michigan is under attack? Looks like the definition of "random" is flexible to these gents. <a href="http://www.techcentralstation.com/110104H.html" target="_blank">It wouldn't be the first time they've played with numbers.</a>
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->I can't imagine traveling around counting corpses would be a good way to calculate death totals. Most people are buried... in the ground...<!--QuoteEnd--></div><!--QuoteEEnd-->
And if they've been burying 500 bodies per day, every day, for the last 2 years then you'd think somebody might notice...
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->Yes he has a political opinion, but he didn't write the report. I don't see how he could change their numbers without writing (falsifying) the report. What exactly are you accusing him of doing?<!--QuoteEnd--></div><!--QuoteEEnd-->
He's an editor of the journal. That means he gets to pick and choose what goes in and what stays out. <a href="http://www.techcentralstation.com/102904J.html" target="_blank">Maybe this will help clear that up.</a>
Cluster size does not equal sample size.<!--QuoteEnd--></div><!--QuoteEEnd-->
No, but they are linearly related. This just isn't a valid objection unless you want to disbelieve statistics.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->If I walk down MLK blvd in Detroit and interview every house on the corner to see if anyone has witnessed gun violence can I then say that all of Michigan is under attack? Looks like the definition of "random" is flexible to these gents. <a href="http://www.techcentralstation.com/110104H.html" target="_blank">It wouldn't be the first time they've played with numbers.</a><!--QuoteEnd--></div><!--QuoteEEnd-->
First, that URL shows nothing close to them playing with the numbers. I don't want to refute the entire thing, so quote specific sections that you feel show that they played with the numbers.
Second, they chose the clusters randomly, this is absolutely standard procedure. To deny this or to claim it's biased is to deny statistics. Or are you speculating that they didn't choose them randomly? If so, I'd like to see some evidence.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->And if they've been burying 500 bodies per day, every day, for the last 2 years then you'd think somebody might notice...<!--QuoteEnd--></div><!--QuoteEEnd-->
This is a lame argument from incredulity.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->He's an editor of the journal. That means he gets to pick and choose what goes in and what stays out. <a href="http://www.techcentralstation.com/102904J.html" target="_blank">Maybe this will help clear that up.</a><!--QuoteEnd--></div><!--QuoteEEnd-->
Yes, so what? Are you <b>speculating</b> that he somehow maniacally edited the report and only left in the bad information, and removed the good information (which is?...)? Can I see some evidence of this? Otherwise it's nothing more than <b>speculation</b>. Nothing in the link jumped out at me except for the fact that whoever wrote it doesn't know what they are talking about. From the article:
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->Have a look at those confidence levels. Yup, 95%. That is, a one in twenty chance that the effect simply does not exist.<!--QuoteEnd--></div><!--QuoteEEnd-->
<b>Wrong</b>. It's a 2.5% chance that the effect is lower than the lower end of the confidence level, and a 2.5% chance that it's higher than the top end of the confidence level. Also this percentage isn't the chance that 'the effect' doesn't exist. His wording suggests to me that he simply doesn't understand statistics, or what this study measures. When the person who wrote this doesn't even have a high school level understanding of statistics, I worry about the resilience of the criticisms. Don't you?
Quote me specifics from the article, I don't have the time nor desire to reply to the entire thing.
You want to believe them, go right ahead.
Man, did the author of that article ever mangle this and what they've written here sounds a lot more like what the <i>power</i> of a study is and not so much the confidence level. To understand what a confidence level actually does, we need to look at what it actually means and how they are derived. Firstly a confidence interval is calculated by taking independant samples from the same population and then deriving the chance that at least a certain percentage of the confidence intervals will accomodate the unknown population factor. The confidence <i>limit</i> is hence the lower and upper boundaries of a confidence interval. We then determine to within 90%, 95% or 99.9% confident that the data lie <i>between</i> the values in question. Taken together, we can use such data to make a statement like the following:
We can be 95% confident that the Labour party will get 60% of the vote with a margin of error of 3%.
This means that we are 95% certain (the confidence level) that the labour party will get 60% of the vote (the confidence interval) with a margin of error of 3% (the confidence limit).
So this statement effectively means there is a 95% certainty that the labour party will get between 57%-63% of the vote in the election based on the sample we took. It does not mean there is a 1/20 chance this is wrong, because most studies usually calculate a power as well as a CL. In short for an example, we are 95% confident in X result with an associated P (power) value (whatever that is). The CL tells you if your estimate accomodates at least 95% of the population dataset and the P value tells you how likely this result is simply due to random chance.
Update: Just reading through the paper we can indeed show how simplistic this supposed 'refutation' is by looking at the paper. The confidence intervals are given and indeed, they calculated a p value, which far from being a 1/20 chance is actually shown to be <0.0001 or in other words, there is a 0.01% chance that their results can be explained purely by chance.
Not, as the article above suggests, a 1/20 chance.
When I've calculated confidence levels in the past, this is the sort of procedure I have gone through, although I'll give you that it has been a while since I last did this sort of thing. Not being a statistician though, my explanation above is probably horribly confusing or worded in a manner that is probably not quite statistically correct (I've avoided going into the mathematics behind this deliberately).
I have not read the paper in question, I have only read some of the controversy around it and other things. My initial impression of the paper from a not so deep look is that I'm somewhat skeptical, but I'm not about to dismiss their results out of hand without looking at subsequent studies that should follow up this work. One paper does not itself prove anything, but rather a body of work has to support the initial findings of the first paper. That's what "peer review" actually entails. <i>Other scientists repeating your results</i>. Until this work is replicated, I'm not going to call it gospel but I'm not going to say it's not right either (especially when I've not read the paper in any detail for myself).
However, can people please <a href="http://www.thelancet.com/webfiles/images/journals/lancet/s0140673606694919.pdf" target="_blank">read the study in question</a> and note what the ACTUAL statistics in the study are. Attack the papers methodology, assumptions for their statistics and similar. Also make an effort to understand the statistics that you are trying to attack as well, or you like the poster quoted above, may massacre a statistics fairy.
Edit: I will state, now I've perused the paper that the confidence intervals they have are pretty wide. That's a fair amount of variation.
I'm not here to frame your argument for you. You asked why their methodology is questionable and I provided links.
You want to believe them, go right ahead.
<!--QuoteEnd--></div><!--QuoteEEnd-->
Sorry, I don't feel like going through that sentence by sentence. You're not framing my argument for me, you'd be framing your own argument... Shrug.
I read both articles fully and I don't remember seeing any good criticisms. I'm inclined to not believe them because of what I said earlier, the writer doesn't even understand confidence intervals. That's why I'm asking for you to quote any particularly good points, so I can reply.
I believe the study, but conditionally. If new evidence comes out I'm more than willing to change my views but I don't see any reason that we shouldn't (somewhat cautiously) trust the study. Am I wrong
Interesting. The study estimates that before the invasion, about 140,000 Iraqis died every year for normal reasons. After the invasion, that number approximately doubled, to around 300,000 per year. So over a 3-year period since the invasion, when approximately 420,000 Iraqis would have died in peace time, instead closer to 900,000 died, meaning about 1/2 a million deaths could be attributed to the war.
Interestingly enough, the study reports 629 total deaths, and 1,474 births during this period. By extrapolating, with a little quick math, I estimate Iraq has experienced 1.88 million births since the invasion, increasing their total population by 980,000 despite the high casualty rate caused by the war. Pre-invasion, Iraq's population increased by an average of 1.8% annually, and post-invasion that rate slowed to a mere 1.2% annually. Assuming 5% infant mortality, that drops to 1.7% growth prewar and 1.1% postwar.
Am I going somewhere with this? Not really. I just like playing around with numbers.