It's world lab animal libration week next week

It's hard enough to do biomedical research involving live animals without having to worry about bomb threats from activists, thanks very much.

Though I wonder if it is really as world-wide as they claim. My God -- can you imagine someone trying to "liberate" the lab animals from research in Egypt? More confusion! Or perhaps it is confusing enough over there that this is a small enough bleep that it won't even register on the radar....

[edit 8:24]
Forgot to comment on this XKCD comic. Or more precisely, how much it reminds me of my PI, since abuse of statistics (which, sadly, you do see in journals, even the higher tiered ones) is one of his favorite pet peeves. For historical reasons we use 0.05 as cut off and for basic tests, that's fine, but when you are testing a large number of hypotheses...well 0.05 is still a 5% chance of false positive, which means you'll find a "significant" result in one in twenty tests just by chance. (And why are they not using ANOVA, any way?) An example in my field, for instance, is testing the linkage between a certain allele of dopamine receptors and...say schizophrenia. There are different types of dopamine receptors. The researchers could start off by testing, say, D4, which they found no linkage. Lack of result = no publication and so, naturally they move on to test receptor D2...and so on. If they test 20 receptors (I don't think there are 20 types of dopamine receptors, but for the same of argument let's say they are testing all neurotransmitter receptors) there'll be one that's linked...if they test 40 there'll be two. If you repeat the 40 tests there might still be two that are positive, but it most likely be a different two because those two happened to be positive by chance. There are various ways to counter that. Bonferroni correction is the simplest and most commonly used for multiple-hypothesis-testing.

Also, yes, replication is important, technical replicates don't count (but biological replicates do), a minimum of 3 reps is good for the standards of most people, but our PI INSISTS on 5 reps whenever possible because of the robustness of stats. Scientific claims in popular news tend to give me headaches. Even pop-sci journals are better but alas, they don't have as strategic placement of splash pages.