topic Interpret the reliability test plan precision number in Discussions

Interpret the reliability test plan precision number

georgerapko — Tue, 31 Mar 2015 17:00:13 GMT

I would like help interpreting the precision value that is calculated with the Reliability Test Plan tool.

Re: Interpret the reliability test plan precision number

michael_jmp — Tue, 31 Mar 2015 17:44:18 GMT

You can read about the precision value in the documentation for Reliability Test Plan online:
Reliability Test Plan and Demonstration

We also have a white paper available that goes into more details about the power and sample size calculations in JMP:

Power and Sample Size Calculations in JMP (See Section 6.2.)

Basically, the precision value is a measure of the width of the expected confidence interval for your test plan.

Hope that helps,

Michael

Re: Interpret the reliability test plan precision number

georgerapko — Tue, 31 Mar 2015 18:28:27 GMT

Thanks Michael that did help a bit. What I'm having trouble understanding is the results give here:

Sample size 1, Censor time 106.9, and Precision 6.134 With the inputs: alpha 0.2, Weibull alpha 66.42, beta 1.

I think I understand the precision value isn't symmetrical around the estimated Time of 6.99 given in the CDF plot with an estimated failure probability of 0.1, but I don't understand how to describe that value. I used a Lower One-sided Interval Absolute Width to get the precision estimate.

Is it correct to say the Point Estimate until we reach 0.1 failures is 6.99 hours? If it is, how is the precision value described? Can I say that I'll reach 0.1 failures at 6.99-6.134 hours?

Re: Interpret the reliability test plan precision number

michael_jmp — Tue, 31 Mar 2015 19:59:47 GMT

The Time value in the plot is a Weibull quantile based on the failure probability and the Weibull alpha and beta values:

Weibull Quantile( .1, 1, 66.42 ) = 6.99804544999282.

For your example, the one-sided confidence interval for the time associated with a 0.1 failure probability is (6.99-6.134, +infinity) or (0.856, +infinity).

Regards,
Michael