I came across this from Jeff Sachs:
[Bill Easterly in his 2006 book] went on to write that “a study of a program to hand out free [malaria bed] nets in Zambia to people … found that 70 percent of the recipients didn’t use the nets.” Yet this particular study, which was conducted by the American Red Cross and CORE, actually showed the program was a success, with high rates of net adoption.
Sachs provides a link to this 2004 study.
I followed the link and took a quick look at the study, and indeed in the intro it says, “Household ITN coverage increased from 28.9 percent (pre-campaign) to 85 percent (with a greater than 80 percent coverage across all wealth quintiles). . . . At six months post-campaign . . . and 60 percent of the pregnant women and children under 5 years old were reported to have slept under the net the previous night.”
60% of 85% is far from ideal but it’s a lot higher than “70 percent of the recipients didn’t use the nets.” But I am no expert in this area, so I sent a message to Easterly asking if there was actually a different study that he was referring to in his book.
My book footnoted a PSI study on social marketing of bed nets (“the Malawi model”) right before the Zambia sentence. [The footnote is 18. “PSI Malaria Control, the Malawi ITN Delivery Model,” February 2005.] The book does not give the 2004 study Sachs mentions as a source.
Through Google, I found the document right away. But I couldn’t find where it said that 70 percent of the recipients didn’t use the nets. It would be helpful for me to track down this particular point, just as a way of getting a better understanding of this debate.
P.S. I’ve met Sachs and Easterly (at different times) and respect them both but don’t know either of them well. Also I am working on a project with Sachs and am affiliated with his Earth Institute.
P.P.S. Someone writes in:
So far the only thing I found in that document was this:
I guess if you look at Feb 2004 pregnant women, coverage is only 34%, which means 66% are not sleeping under bed nets, which is close to 70%?
For a more recent study, see p.18, Table 2, the mean of the free group usage conditional on take-up is 70% (take-up looks like 99%, see p.17).