Monday, August 5, 2013

The importance of reading and writing carefully

Two interesting recent blog posts underscore the importance of both reading and writing very carefully when grappling with an argument.

First, AidData's blog The First Tranche (disclaimer -- I'm associated with AidData, though I am not directly involved with the blog) posts a valuable overview of Chinese aid and influence in Zimbabwe, by Amber Will. The post on the whole is very good and well worth reading.

However, there is one key sentence that is problematic: "AidData uncovered $3.82 billion in official Chinese finance to Zimbabwe, accounting for 4.9% of all official Chinese financing to Africa." The issue here is that the second part of the sentence needs the same qualifier as the first: "AidData uncovered". After all, we don't know the total amount of all Chinese financing to Africa -- all we know is the total amount found by AidData's media-combing/crowd-sourcing approach.

Indeed, even the qualifier "AidData uncovered" is less than ideal: strictly speaking it should be "AidData uncovered references to" aid -- whether those references are accurate is unknown. I suspect that both the $3.82bn figure and the 4.9% estimate for the portion of total aid are, if anything, under-estimates. Still, we really don't know enough to be as confident of either as the post suggests.

In a second example, the Duck of Minerva blog recently featured a debate about what caused the Iraq War (here and here), inspired by an article in the current issue of International Organization by Alexandre Debs and Nuno P. Monteiro. Again, the blog posts are well worth reading, but Debs and Monteiro's argument is weakened by the following statement introducing it: "we argue that the 2003 U.S.-led invasion of Iraq can be accounted for strictly within a rationalist framework."

I have not yet read the original article in IO, but this is a very lame claim: it is trivially true that the US-led invasion can be accounted for within a rationalist framework -- all you need is to specify appropriate preferences and information. The claim becomes interesting only if you can show that key actors held those preferences and had that information, and that other models fall short.

The invasion can also be accounted for by a divine inspiration framework: all we need to do is posit that Bush believed God talked to him directly and told him to invade Iraq. In the absence of additional data showing what Bush believed, that claim is just as credible as the claim Debs and Monteiro make: constructing a model is only a first step, and not a particularly challenging one at that (in this particular example).

In the remainder of their post, Debs and Monteiro do in fact provide some data to move their argument beyond simply positing a particular model. This suggests that they simply do a poor job of summarizing their main claim at the start (although some critics may not be convinced by the data they supply).

Be that as it may, both of the examples given here underscore the importance of writing (and reading) carefully.