There used to be an ad on TV that claimed one toothpaste was proven to be more effective in fighting tooth decay than another. One of statistics professors was interested in the study, and after 6 ads finally got the address to write away for it from the TV.
The study was classically perfect - randomly assigned individuals into the test and control group, double blind administration of the tooth paste samples, etc. Everything looked great until step 2 of the hypothesis testing procedure, where alpha was defined as .80! Now this was a one tail test: My toothpaste has fewer cavities than the other (alternate hypothesis).
What is the impact of a .80 alpha? For those who do not understand statistical procedures, this probably sounds like a good 80% mark - but what is the 80% all about?
"Alpha" measures the type I error. This is the probability that a statistical test will generate a false-positive error; that is, you rejected the ...
A real life hypothesis example is analyzed. The impact of 0.80 alpha.