Posted by ALP on July 2, 2009, 21:58:20, in reply to "Re: "one would expect a wider spread of sources cited""
Michael writes:
> I can't see anything in gunnar's posts which is especially
> controversial given the above facts
It's not controversial, but I think it is fundamentally confused. He's saying that he'd expect an "equal distribution" of citations in the IBC database for all the sources monitored.
Actually, the reverse is true. You'd have good reason to suspect something was wrong if there was an equal distribution, since the reporting itself is far from equally distributed. Reuters, for example, picks up a much greater percentage of incidents/deaths than most other news sources (including non-Western sources). You'd expect that to be reflected in the IBC database, just as you'd expect media sources with very low coverage to have far fewer citations.
The real question (as Raoul points out) is whether there's any evidence of documented deaths from any source that IBC isn't including in its database. So far there doesn't seem to be any, despite the fact that some experts have probably been looking pretty hard for it.
Empirically, the IBC database is demonstrating its worth continually, but the ML guys decided to take a break from the empirical world for this particular issue.
Message Thread:
« Back to thread