Glenn Gee wrote:
>
> I have a question in regards to which test statistic to use. Example:
> my problem is a test of glue "A" vs glue "B". I have a test that just
> determines if glue "A" remains stuck to the surface or not. The test
> does not have any other measurements. It holds or does not hold. If I
> test 5 samples of glue "A" against 5 samples of glue "B" with the
> following results:
> glue "A" held 4 times, glue "B" held 2 times.
> Can I make a valid decision? Do I need more samples? Suggestions?
> Thanks
> Glenn Gee
> Director, Manufacturing Assistance Services
> [EMAIL PROTECTED]
You need more samples. Look up the Chi Square test or Fisher's exact test. In
principle your procedure can be used to test the effectictiveness of the two
glues (though other measurements e.g., time until failure might provide more
sensitive tests).
Thom
