Weibull & Reliability/Failure Time Analysis - Raw Data Tab

Select the Raw data tab of the Weibull & Reliability/Failure Time Analysis dialog box to access options to specify the variables for the Weibull analysis & reliability/failure time analysis. When analyzing raw data, specify the variables for the analysis and any censoring variables and codes, and then click the OK button. By default, Statistica computes maximum likelihood parameter estimates for the two-parameter Weibull distribution, and displays the Weibull Analysis: Results (Raw Data) dialog box. Note that if the maximum likelihood parameter estimation procedure fails, the values 0, 1, and 1 will be used as the default parameters for the location, shape, and scale parameters, respectively.

Input variables

This group box contains the following two options:

List of variables with failure times

Select this option button if your data contains actual failure times. When this option button is selected, you can specify a list of variables (with failure times) for the analysis.

Single variable, start & stop times, or dates

Select this option button if your data contains dates (or other measurements of time) that specify the start and end (failure or censoring) times for each observation. While reading the data, for each observation, Statistica computes the difference between the end (stop) time and the start time, and later fits the Weibull distribution to the difference scores.

Variables (failure times & censoring indicator)

Click this button to display a variable selection dialog box, in which you select the variables for the analysis. When the List of variables with failure times option button is selected (see above), Statistica expects one or more variables with failure times and an optional censoring indicator (grouping) variable with codes that identify the censored and complete cases.

When the Single variable, start & stop times, or dates option button is selected (see above), you can specify in the first variable list either 1) a single variable with failure times, 2) two variables with dates or other time measurements; when reading the data Statistica will subtract the values found in the first variable from those found in the second variable (i.e., subtract the start-times from the stop-times) before fitting the Weibull distribution; or 3) you can specify 6 variables, which will be interpreted as start and stop times [as in 2) above]; specifically, the 6 variables will be interpreted as start month, start day, start year, stop month, stop day, and stop year, respectively; for more information on this special type of input format, see also the Survival Analysis - Life Tables and Distributions. In either case, you can also specify an optional censoring indicator (grouping) variable with codes that identify the censored and complete cases.

Code for complete responses (failures)

This box is available only if a censoring indicator variable is selected. Specify the codes or text labels that were used in the censoring indicator variable to uniquely identify complete (uncensored) observations. To review all codes in the respective variable, double-click on the respective box (or press the F2 key on your keyboard) to display the Variable dialog box. The first two valid values found in the censoring indicator variable are taken as the default codes for uncensored (complete) and censored data, respectively.

Code for censored responses

This box is available only if a censoring indicator variable is selected. Specify the codes or text labels that were used in the censoring indicator variable to uniquely identify incomplete (censored) observations.

Add constant to 0 times

The Weibull distribution is bounded on the left side; i.e., all values must be greater than the respective location parameter (greater than 0, by default). When the Add constant to 0 times check box is selected, Statistica replaces zero failure times with the constant specified in the adjacent box, before fitting or plotting the Weibull distribution. When the check box is cleared, cases with zero failure times will be excluded from the analysis (treated as missing data).

For related procedures, see also the Survival Analysis module; for non-censoring (complete) data, you can also use the Quantile-Quantile and Probability-Probability plots on the Stats 2D Graphs menu as well as the distribution fitting options on the Process Capability Analysis Setup - Raw Data: Distribution tab.