Still, any effect is a small one. The real question here is what on earth is going on with the number 7? And, to a lesser extent, 21? I know this survey is far from perfect, but the google returns for 7 deaths are so far out from the rest of the data set that it is hard to believe it is merely a random fluke. Looking at the Google results provides no obvious clues to explain the difference.
The number 21 also has an oddly high number of reports. This is a bit counter-intuitive, as if there was an attempt by the PR machine to make the numbers more appealing, I would guess that they would aim for 18 or 19 instead — you know, the old salesman’s trick of setting the price at $19.99.
I wish I had something clever to say here about what might be going on, but I don’t. Still, if we are going to be questioning the accuracy of military civilian death counts, the alleged ubiquity of the number 30 may be a red herring — 7 and 21 seem like much more promising anamolies to investigate.
There's at least one problem with the graph-- the vertical axis is number of google hits rather than number of uniquie accounts. I don't know whether a few stories which are interesting enough to get linked more could be causing the anomalies. I also don't know whether a statistician would say that blips are to be expected if the situation was random.
in any case, View from LL2, Security Crank, and Moon of Alabama seem to be part of a community that's trying to figure out what's going on, and I'm planning add it to my blog reading list.