Laurence wrote: > Mr Ulrich, > > No it's only to make a Chi square test on residuals against a Chi square > value. > This test is needing to prove if the model i have find by least square > method is valid or not. > Then i've to verify is the residual distribution is Normal or not. > I've to do a Chi square test. But in this test we have to find different > frequencies in different intervals. > I' dont know how to construct this columns, how many columns. How to > determine the size of this columns. > How to calculate the frequency in different columns. > > If someone know it particularly for the residuals.
This is called the chi-squared goodness of fit test. Any introductory text should have at least an example in the section on the chi-squared distribution. I don't think there is a particular modification for residuals. I guess you found this material in the book, given what you write above. The basic idea is to have as many columns as possible. I've seen informal rules saying that you should have an expected number of at least 5 in each column. To compute the expected number you convert to a probability --- you have the sample size --- and use standard normal probabilities from a table. For instance, if n=10, use two columns and split at the mean. Then count how many residuals are below zero, how many are above. There are better methods than the chi-squared test, by the way, even for assessing the normality of an arbitrary sample. . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
