Right Censoring data for ANOVA

[#1]

Hello,

I am trying to carry out ANOVA on some data from correlation experiments I am currently running across 3 different oven chambers. It is a reliability test with each unit having a time-to-fail. For the below plot, I have circled 32 data points in red which have not failed yet. For my Fit Life by X analysis, I have right censored these units as they have not yet failed but I'm wondering is there a method of doing something similar in the Fit Y by X platform for ANOVA? I would rather not hide & exclude them as it is valid data. I would like the impacts of these data points to still factor in the calculations for my analysis. My sample size is 16 units per lot. Thanks.

[#2]

Unfortunately Fit Y by X does not support censored data.

Generalized Regression in JMP Pro does support censoring and would enable you to fit a linear model.

Can you let us know why you want to model these data in Fit Y by X?

Phil

[#3]

Adding to @Phil_Kay, the Analyze > Life Distribution platform can be used to correctly model life time data with censoring and compare groups. See the Help for this platform.

[#4]

Hi Mark/Phil,

Thank you for the feedback. The different datasets above are actually the exact same source material in the same stress conditions, but each data set is using a different set of hardware. I am trying to carry out ANOVA to determine if all sets of hardware return the same approximate MTTF. I have been using the each pair, Student t's test and ANOVA options in the Fit Y by X platform for this work. Is this also available in the life distribution platform? Thank you.

[#5]

Did you read the documentation about Life Distribution as I suggested. I provided a direct link to this information.

No, those features are not available, but it is crucial that the model behind the analysis correctly accounts for the censoring. Otherwise, it will lead to biased estimates and tests. Excluding the censored observations will also lead to biases estimates. So it makes sense to get the correct model, then use available comparisons.

Another platform that might help is Parametric Survival, which is available through Analyze > Fit Model. This analysis is analogous to ANOVA when you have a categorical factor for Equipment.