From Science (October 2, p.26–27):
Tuesday, October 27, 2009 :::
Wow.He's the statistician, I'm not, and I don't know how many failures they had in "2 decades of effort", but if I do 20 experiments and finally get a result that has a 1/20 chance of occurring by chance, I'm not jump-up-and-down excited. If my theoretical prior is leaning against the result, I'm that much more not excited. This apparently was a rather expensive study, and was controversial for that reason; it may be that this is a good enough signal to justify another expensive study.
That was the reaction of many HIV/AIDS researchers last week to the news that a large clinical trial of an AIDS vaccine worked— the first success in 2 decades of effort. The two vaccines used in the study as a one-two punch had only a modest ability to protect people from HIV, and the results just barely qualified as statistically significant [at the 95% level]. ...
... One of the most perplexing findings is that the vaccines prevented infection by HIV but failed to reduce levels of the virus in people who did become infected. "Most people expected it would be the other way around," ....
.... "The results tell us two things: These vaccines work, and they don't work well enough" says statistician Donald Stablein, president of the Emmes Corp. in Rockville, Maryland.
p=0.039, by the way. In 3 years after receiving shots, 74 infections in the control group, 51 in the experimental group, about 8,000 people in each group.
::: posted by dWj at 11:17 AM