|State the problem|
For a control on this test, I'm using someone else's blog and their software over which I have no control. Also, I've hired a bunch of trained monkeys to post comments there to see what happens as we approach the limit imposed by Overlord PZ Liars.
Test procedure was as follows:
trained monkeys typed a lot. As we know, 20 monkeys on 20 keyboards . . .
Data collected. (see figure 8, by a FuQu2 recursive algorithmic analysis, we can see the following pattern emerge - note, not to scale)
We have collected so much data that at the 100% confidence level, we can reject the null hypothesis.
Ruminations: this section is much better than the conclusion. Note how many threads PZ indicates have not been censored. Also note that this says nothing whatever about how many threads were censored. For instance, if I show that I haven't killed 10 people, does this in any way indicate that I have killed no one? (This supports my 100% confidence level - like PZ Liars I too ignore inconvenient data!)
(this version is only a draft to be used internally; not yet ready for the referee process before submission to peer-review in Pharyngula)
[edit: for the record - in this one blog post I have come as close to publishing a scientific paper in a reputable journal as PZ Liars has since being graduated from grad school. Just sayin'!]