Subscribe Bookmark RSS Feed

How to properly censor data

daniel

Community Trekker

Joined:

Oct 7, 2014

Hello everyone.

I want to use the jmp software for analysing biological data. I observed how many eggs some worms laid over 40 days after birth. Some did not start to lay eggs. I want to use the survival analysis of jmp to have a look at the timing of the first eggs different groups of worms produced (instead of dying they start laying eggs). My question is how I can censor those who did not lay eggs in the 40 days and how I define the 40 days threshold in the programm.

Ps: Do you think the survival analysis is ok for this question or can you reccommend a better way of analysing this (maybe to analyse the timing and the ammount of eggs the worms produced) within jmp?

Every help is greatly appreciated.

Have a nice day.

Daniel

1 ACCEPTED SOLUTION

Accepted Solutions
Solution

The easiest way to analyze your data is to set up a data table with a separate row for every worm. In one column enter the day on which the eggs were laid. For those worms that did not lay eggs by the end of the study, enter 40 for the number of days. Create another column. Call it what you like. Enter one of two values (e.g., 0 and 1) in the new column to indicate if the number of days for an observation of an actual egg laying event (e.g. 0 for uncensored) or not (e.g., 1 for censorerd). Then select Analyze > Reliability and Survival > Life Distribution. Select the column with the number of days and click Y, Time to Event. Select the second column with the censor indicator values and click Censor. Click the red triangle after Censor Code on the left of the dialog to indicate the value that you used for censored observations. Click OK.

Use the check boxes before the model name on the left side of the plot to fit a distributional model to your data. Note that only uncensored events are shown in this plot. Use the radio buttons after the model name to select axes that should linearize the non-parametric mid-point estimates if the model is a good approximation.

After you select a model, then use the reports below to extract information about the time to laying eggs.

Learn it once, use it forever!
2 REPLIES
Solution

The easiest way to analyze your data is to set up a data table with a separate row for every worm. In one column enter the day on which the eggs were laid. For those worms that did not lay eggs by the end of the study, enter 40 for the number of days. Create another column. Call it what you like. Enter one of two values (e.g., 0 and 1) in the new column to indicate if the number of days for an observation of an actual egg laying event (e.g. 0 for uncensored) or not (e.g., 1 for censorerd). Then select Analyze > Reliability and Survival > Life Distribution. Select the column with the number of days and click Y, Time to Event. Select the second column with the censor indicator values and click Censor. Click the red triangle after Censor Code on the left of the dialog to indicate the value that you used for censored observations. Click OK.

Use the check boxes before the model name on the left side of the plot to fit a distributional model to your data. Note that only uncensored events are shown in this plot. Use the radio buttons after the model name to select axes that should linearize the non-parametric mid-point estimates if the model is a good approximation.

After you select a model, then use the reports below to extract information about the time to laying eggs.

Learn it once, use it forever!
daniel

Community Trekker

Joined:

Oct 7, 2014

Ok I will try that. Thank you very much for your help!