Question 433:

Asked on October 1, 2008

Tags: Non-Normal Data , Log Transformation , Cycle Time

1

Answer:

No answer provided yet.

Yes you can do that. This approach is called dichotomizing your continuous data. When you do this you do lose a lot of information since you no longer have a measure which is sensitive to the subtle changes of time (seconds, milliseconds etc) and now only have the crude pass/fail.

You should use the raw data if possible to calculate the z-score. Cycle time data is usually non-normal (positvely skewed) since there is usually a lower limit on how fast things can happen. To remedy the normality issue, I recommend transforming your data using the natural log then calculating your sigma on the transformed data.

Let me know if you need assistance with the calculations.