Statistical Tests
For reliability growth data analysis only.
When you use the Crow-AMSAA (NHPP) or Crow Extended models, Weibull++ automatically performs statistical tests on the calculated data set. The results of the tests help you to evaluate how well the model fits the data. Only the test that applies to the data type and model you are using will appear in the results. The results are displayed on the growth data folio control panel, and can also be displayed in a report by choosing Growth Data > Analysis > Statistical Tests Report.
The following tests evaluate the hypothesis that the failure times follow a non-homogeneous Poisson process (NHPP). The ReliaWiki resource portal provides more information about these tests at http://www.reliawiki.org/index.php/Crow-AMSAA_(NHPP).
The Chi-squared goodness-of-fit test is applied to grouped failure times.
The Cramér-von Mises (CVM) goodness-of-fit test is applied to non-grouped failure times where there are no gaps in the data.
The following tests apply to multiple systems analysis only. The ReliaWiki resource portal has more information about these tests at http://www.reliawiki.org/index.php/Hypothesis_Tests.
The Common Beta Hypothesis (CBH) test indicates whether all the systems in the data set have similar beta values so you can evaluate whether the systems should be combined into a single representative system (i.e., the equivalent system, superposition system or cumulative timeline).
The Laplace Trend test evaluates the hypothesis that a trend does not exist in the data. It can determine whether the system reliability is improving, deteriorating or staying the same. When the Crow Extended model is used without BC failures modes, it is assumed that there is no trend (i.e., the system is neither improving nor deteriorating).
Tip: You can set the default significance level for the statistical tests by entering a value in the RGA Growth Data Folios page of the Application Setup, or by changing the value on-the-fly from the Analysis page of the growth data folio control panel.
The following tools are also available in Weibull++, but are not automatically performed when you calculate the data set:
The Interval Goodness-of-Fit Test tool helps you to determine which intervals should be used to group the data so that the goodness-of-fit test passes. This option is available only for fielded data and some of the multiple systems data types (Concurrent Operating Times, with Dates and with Event Codes).
The Test for Fix Effectiveness tool helps you to assess whether or not applied fixes have been effective across test phases. It is available only for the multi-phase data types.