But we want to be able to measure this, right?

The alternative viewpoint is that there are a small number of really excellent scientists and mentors, and you want--need, even--to concentrate resources there to make the most important discoveries in difficult areas, and to give the best training to the people who can in their turn be really excellent. This is how apprenticeship systems work: to become a great master, study under a great master, not because it's incestuous but because they are the ones who have the temperament and skill to be great and, if you have talent, can pass it on.

The question then is: to what extent is that view true and to what extent is it unhealthy incestuousness? And to answer that reliably, you need to be able to measure the effect.

My primary strategy for convincing scientists to change their minds is to collect data and analyze it sensibly. This goes for everything from determining how polyglutamine repeats cause toxicity through to changing institutions of funding and publishing.

One might also appeal to society-wide goals (e.g. "we want to have inexpensive science broadly distributed because that results in a healthier, more thoughtful society"), but getting scientists on board via data is one of the less difficult steps. (Still difficult, of course.)