Lots of good points, as always. I particularly like this point for its combination of relevance and generality.
The trick is, there’s data about a lot of things, but if you slice what counts as relevant data finely enough, you can claim there’s no data whenever you’d prefer that there was no data. You want to worry about 16-17 year olds? Say that 18 year olds provide no data. If you wanted to, you could then say that people were the wrong gender, or the wrong race, or the wrong health status, or the wrong lifestyle, and keep doing that until your sample was small enough that there was ‘no data.’ Or you could realize that all data is data, and do something reasonable, but reasonable is not the watchword here.
Lots of good points, as always. I particularly like this point for its combination of relevance and generality.