Weibull analsis of suspended data

[Ali Khatri]

We have 2 samples of 6 each to compare. First sample has 2 failures and 4 suspensions. Second sample has all 6 suspensions. What is best way to compare in this situation?

[Adam]

It all depends on what is that you want to compare. For example, are you interested in a particular reliability at a given time, or a particular time given reliability. In either case, I would use confidence bounds and I would try not to extrapolate way beyond the range of the data. There are also methods for comparing two distributions over their entire range. However, I would be very cautious in making such a comparison when having very few data and failures, as in your situation.