I don’t get it. There’s some inference going on here, and I just can’t grasp it. Is “the null hypothesis” supposed to be a generic stand-in for an outdated idea? Or is the guy mistakenly thinking that there’s only one null hypothesis because we call it the null hypothesis? Is it some commentary about our school system being slow to update? If so, what’s that have to do with the null, as opposed to any hypothesis, persisting? Perhaps he’s making fun of people who read a single study and dismiss something that should require some more rigor to be dismissed?
And what’s any of this have to do with the mouse-over about 8th graders and mega-research teams?
The null hypothesis usually comes up in about 7th or 8th grade science. StickMan apparently had an 8th grade class that performed an experiment that disproved that particular experiment’s null hypothesis. He mistook the conclusion as disproving every null hypothesis ever, and is perplexed that some researcher spent millions of dollars conducting experiments that couldn’t manage that much.
My surmise is that the joke is about (1) people who use statistical language without understanding what it means joined with (2) the observation that, generally, a study is only “interesting” when the null hypothesis can be rejected and (3) only “interesting” results get published. Thus if you pursue science journals knowing certain buzzwords, but not really being clear on the scientific method, experimental design, and rules for statistical inference (where might this happen?!?), you might, like poor benighted StickMan, think that “the null hypothesis is therefore rejected” means something other than it really does. To coin a phrase: “Hurr durr.”
The joke is that the guy assumes there is only one the null hypothesis. Therefore, if one experiment proves the null hypothesis wrong, all such hypotheses must be wrong, and it makes no sense to continue testing them.
My take on it is that it claims that a big study years ago demonstrated that there is no such thing as a null hypothesis. The irony comes from the fact that such a demo must have started with a null hypothesis. What could it possibly be? Only that there is no null hypothesis, but how could you possible reject that?
Just for the record, the null hypothesis comes when you ask a question, could such and such an effect be real? Can prayer really affect healing? The null hypothesis is that there is no effect of prayer on healing. You then set up an experiment in which half the patients are prayed for and half aren’t and see if there is any effect on healing. If there is, then you can reject the null hypothesis. Such an experiment was carried out a few years ago and the first results suggested that prayer did affect healing, but it turned out to be unrepeatable and had to be a statistical anomaly. After all, if effects found with p = .05 will still be wrong one time in 20. And effects found at p = .1 will be routinely rejected by people who misunderstand probability even though the effect should be real 9 times out of 10.
Anyway, I think the cartoon is pretty funny in the same way as CS books that have an index entry to “infinite loop” that says “See loop, infinite'" and an entry to "loop, infinite" that says "See infinite loop’”. It is assuredly self-referential.
