GIGO is a good principle to keep in mind, but nobody's numbers can be absolutely trusted here; every government has incentives, both from good motives and ill,

GIGO is a good principle to keep in mind, but nobody's numbers can be absolutely trusted here; every government has incentives, both from good motives and ill, to make things look as good as possible—and even when there is no conscious bias, various decisions will at least mean that numbers from different countries are probably not really comparable. Thus the normally scientifically sound principle of not allowing in suspect data is indistinguishable, in terms of its effect, from simply ignoring the available data. Better to regard the data as suspect, and do as much statistical analysis as possible to deal with it. (I think, for example, of data gathered on the enemy during wartime, which are also constantly somewhere between unreliable and falsified; but we use them anyway, knowing that they cannot be completely trusted.)