Apart from the sillyness of actually engaging in a calculus to stick some kind of moral degree on people I did use the tables from that study and did a quick calculation.
The result I got was that there have been 232 casualties extra, because of what the study called “High mention weeks”.
I did what he wanted and got no reaction.
Strangest I found was that, using the figures, I came to a success rate of 8.28 casualties per attack.
Surely that cant be right.
I was hoping for someone (maybe the OP perhaps) to do what he claimed his interest was, and correct the figures. Point out a mistake here and there.
If protesting a war hypothetically led to more casualties in that war, I do not think the protesters should share moral responsibility for those deaths. Iraq war protestors aren’t responsible for he war’s casualties any more than Henry Ford is responsible for traffic fatalaties.
In the movies, the crazy badguys like to kill a bunch of people if they don’t get what they want. They say stuff like “you don’t want their blood to be on YOUR HANDS do you?” but it’s not the hero’s fault when the badguy does decide to execute someone.
Magellan, you’re like that badguy in the movies. You want the people who are trying to stop the war to feel responsible for the death that it’s causing. We are, understandably, quite fucking irritated at the implication, especially when it’s based on nothing but a dumb hypothetical that’s not even likely to relate to reality.
So let’s change the hypothetical. Obviously, the current war is a loaded subject, and there’s no way to step back and discuss it objectively. Too many people have too many strong feelings about it, which are contaminating the discussion. We’re unable to examine the principles, in the abstract, because of the concreteness of the real-world examples in play. Therefore, let’s alter the terms of the debate.
New hypothetical:
Every time somebody disagrees with one of magellan01’s posts, and replies to say so, magellan01 becomes upset, and rapes a child. Should we take this into consideration before replying to anything magellan01 says? Do those who disagree with magellan01 bear any moral responsibility for the children he is raping as a result?
Since, contra tomndebb, it appears that the study remains on the table as “evidence”, if not proof, perhaps a few more facts about the study might help to put a stake through its heart. They certainly cast doubt on Ms. Chenowith’s cavalier assertions about the likelihood of publication of this paper.
The authors exclude Baghdad from the analysis!
They exclude it even though they acknowledge that something in the neighborhood of one third of the U.S. fatalities occurred there, and according to their own data, on average, 200 more attacks per week occur there than elsewhere. Their justification is that it is too disparate in other factors from the other provinces to include. If one were for some reason concerned about it being too distinct from other regions to include in those analyses, I would wonder why it couldn’t be examined separately, given that the numbers there are so high.
The authors include Sunni versus Shiite ethnicity as an afterthought, rather than controlling for this crucial factor in Iraqi violence in their primary models.
They do not present the main effects of the models, but only the interaction terms. They do not explain how they handed the apparently markedly non-normally distributed outcome variables in their statistical analyses.
Their analyses are broken out into 8 time periods, but they fail to note that they find significant interactions for only a few time periods. The authors use a non-conventional denotation system, starring interactions with p values less than 0.1, but yet fail to achieve a p value of .05, which is typically that used to describe something as statistically significant. For instance, out of nine time periods examined for attacks and high mention weeks, only one achieved significance at p=.05. The results for fatalities were that in no time period did the effect exceed a p value of .05.
The release of US polling was found to be significantly (again, p = .05) associated with attacks in only 4 out of 9 comparisons.