Oh no! You're spoiling my line! Interesting findings, nonetheless, Jerry. We appear to be taking a slightly different approach - I'm looking at N(crisis + X)/N(crisis) and you're looking at N(crisis + X)/N(X). So what I'm computing is the % of pages with "crisis" that also have "education", and you're computing the % of pages that have "education" that also have "crisis". So basically what you're showing is that (very, very roughly) about 10% of the discussion of education involves "crisis", where I'm showing that (again, very roughly) about 2/3 of the discussion of "crises" involve "education." I think the latter is the more relevant thing to look at here, but your point about controls is well taken. (Incidentally, I think the high number for mental health may have to do with the fact that mental health issues can lead people to have personal crises - there are crisis hotlines, etc - like I said, all these things are very crude)

I definitely agree that things are already ugly, and I sympathize about your grants. The current pay lines are really low. The reason I am concerned about the next few years is that there are still a lot of people supported by the large numbers of R01s that were granted in 2002 and 2003. A lot of those grants won't be renewed, and the number of new R01s to replace them will be lower. Also, a lot of the new hires that Paula Stephan documented from the 2003 time frame are going to be reaching the end of their startup packages, so as people have said in The Scientist article, we may well be looking at a fair number of lab shutdowns.