valid or not?
- "One problem is that statistical tests are open to fraud. Scientific theories must make logical sense in order to be true, and can thus be easily recognised by anyone as true or false. Statistical tests on the other hand have no such requirement, so it is possible for researchers to fabricate data without easily being caught."
This doesn't sit well with me. Anyone else? Alecwh 23:51, 5 February 2009 (EST)
an easier way
"Another problem is that standard statistical tests randomly give a false positive (an incorrect result saying it works) a fixed percentage of the time. Statistical tests usually accept as proven anything that has less than a 5% chance of happening at random. What that means is that if you test something that doesn't work, 5% of the time the test will incorrectly show that it does work. That allows researchers to continually repeat the experiment until they get the false positive, and only publish that study."
It is unethical (scientific fraud) to do as you say and not report the previous tests that showed no effect. It is the same as (although more time-consuming) throwing out data that does not match your intended result. If you're going to go ahead and lie about something like that, might as well take that research money to a casino and then make up some data. This sort of thing isn't really a valid concern, at least not in this article, perhaps so in one discussing scientific fraud. — Unsigned, by: 18.104.22.168 / talk / contribs