How long should a test be run?

[unstable bob]

Just wondering if there is a rule of thumb as to how long a test should be run before it can be declared a "winner" or a "bomb?" I'm trying to get awy from my "one tank wonder" testing...

[theclencher]

http://www.gassavers.org/showthread.php?t=4016

[Bill in Houston]

Ya, see that thread. It depends on how stable your results before and after are, and what size change you are trying to prove.

[KARR]

The longer the test the more accurate your results will be.

[Jim Dunlop]

You can test for statistically significant differences between two means using Microsoft Excel. It's in the Data Analysis toolpak add-in thingy I believe. Just install it and use Help. I'm sure similar directions were given in the thread referenced.

[Bill in Houston]

Ah, cool. Thanks for posting that, JD.