Schoolchildren in the North East of England are to be given fish oil supplements. The purpose of this trial is to improve exam results [source: BBC News].
Some people might think that improving health is more important than improving exam results. I think it says something interesting about our society that this trial is justified (to participants and their parents, as well as to the tax-payers who are presumably funding the scheme) in terms of the education benefit, rather than the health benefit.
Meanwhile, the Food Standards Agency (presumably aligned more to the health lobby than to the education lobby) has already cast doubt on the education benefits of fish oil [source BBC News]. If I didn't believe in the pure motives of public servants, I might think that the FSA objections were based more upon inter-agency rivalry than upon objective science.
There must surely be some degree of statistical correlation between health and exam results, so you might possibly use exam results as a very rough indicator of general health, but this would need careful interpretation by a trained statistician, as there are obviously many other factors involved. In any case there are much better (more direct) measures of general health.
One possible reason for the nature of the scheme is that the education agencies are able to use their budget to produce educational benefits, but not to produce health benefits. So they downplay the health benefits, in order to avoid having to collaborate with other agencies.
I said above that the purpose of the trial is to improve exam results. A scientist might argue that the purpose of any trial is to produce knowledge, and the knowledge can then be used to produce some benefit, such as improved exam results. But this is not how government agencies work. Trials like these are expected to deliver results, not just knowledge. (Note that there is no mention of double-blind testing - the supplements will be offered to all year-11 children in the region.)
So the purpose of a scheme depends on what benefits are seen as politically acceptable, as well as on the division of responsibilities between different agencies. And the visible effects of the scheme will be determined by what is measured. In situations like these, POSIWID must be seen as a political construction.
Update: For Tim Harford's call for proper scientific trials, see Political Ideas Need Proper Testing (Financial Times, 18 March 2010)