It’s fine to defend the conclusion that development aid does not help, and may in fact hurt, economic growth, but oversimplified graphs aren’t much help, as Jim rightly complains.
However, I think the oversimplification is partially a result of cramming one side of an extensive development economics debate into a one thousand word article. In his 30-page policy brief on the subject of aid and development, Fredrik Erixon includes the caveat that “these figures alone do not overthrow the idea of aid.” He then develops his argument by referring to econometric work by William Easterly and Peter Boone (which Jim criticizes elsewhere).
The oversimplification problem is common in articles on development written for laypersons. For example, the accompanying article by Jeff Sachs, which defends aid as useful, features a graph that is equally misleading.
Obviously government health expenditure is correlated with a number of other features that are important determinants of infant morality, and this graphic provides little useful guidance to the policy issues in questions.
Trade Diversion 
Good point on the Sachs graph, I’ll have to mention that in an update to my post. I suppose government health expenditure as % of GNP would be a better variable to use, but it would be interesting to see an econometric analysis that tried to control for various factors.
More dodgy aid charts
A few days ago I posted on the subject of a graph by Fredrik Erixon that mysteriously omitted the recent years in which his posited negative relationship between aid and growth seems to have been reversed. Here’s his graph again….