Harder than Hard

(That's what she said). Seriously, though, you know what's harder than hard? Trying to explain multivariate regression analysis to one of your sophomore students from principles of microeconomics. We had a seminar speaker present a paper on racial discrimination, and he came. How do you explain that this speaker "controlled for" a ton of crap and that the correlation coefficient was positive and statistically significant for a variable that reflected the measure of discrimination? Answer: you talk really fast, repeat yourself a million times, don't use any graphs or talk about slopes, and answer all their questions when they say "well what about this or what about that?" I don't think he understood multivariate regression any better, but at least he understood what it was required to topple the evidence in the paper - it must be the case that the discriminator acts differently whenever there is a minority present because something the econometrician doesn't observe is going on. For this particular speaker, it was very hard to imagine what that "something" would have to be, but it at least helped me frame the issue to him much better, and I think he at least understood that.