How do I interpret the Turnbull-Weiss statistic that is reported for the calibration step?
• Dave Lorenz writes: “If the p-value from the Turnbull-Weiss (T-W) test is low, then it is unlikely that the residuals are normally distributed. I suggest using the p-value from the T-W test not as a accept-reject criterion, but as a guide to further exploring the normality of the residuals. I have seen nearly normal residuals fail the T-W test at the .05 level, but I have not seen non-normal residuals have a p-value greater than 0.05. If the p-value from the test is less than 0.05, I look very carefully at the normal plot and other diagnostic plots to try to determine what might be the problem.” Tim Cohn writes: “Turnbull-Weiss is a test for non-normality of the residuals. If the p-value is close to zero, then the data are not likely to have come from a normal distribution. I don’t think there is a good critical value in this case. My sense is that you will always reject the null hypothesis if you have enough data. The real question is whether or not the residuals are approximately b
Related Questions
- Can I use the barometric pressure as reported by the local weather channel to check the calibration of the internal IABP transducers?
- The EPA data is hard to interpret. How do you know what portion of the reported power output is gas-generated?
- How many American children were reported abused and neglected in 1998?
