An existing client is looking to perform the Chow test on JMP. Any feedback greatly appreciated. Thanks in advance!
@cweisbart , It looks like it can be done, but it requires making your own subsets. This paper goes into how it can be done.
https://www.lexjansen.com/wuss/2006/posters/POS-Calise.pdf
To run the Chow test, use the Custom Test option in the Fit Least Squares.
I use this SAS example to illustrate https://support.sas.com/rnd/app/ets/examples/chow/index.htm. Data set is attached. It estimates the effects on per capita food consumption of the food price and per capita income in the US for the years 1927-1941 and 1948-1962.
We want to test if there is structural change in the parameters of the model after 1941 when the US entered the WWII:
- Create an indicator variable: Before 1941=1; 0, otherwise.
- Run the SLS with regressors, lnp and lny, and their interactions with the indicator variable;
- Perform the Chow Tests using the Custom Test (shown on the left below).
JMP results, F Ratio=5.066, p-value=0.0073, show that the estimated effects on food consumption of food price and income changed significantly after 1940. The result matches that from SAS Proc AUTOREG’s Chow test.
(For those who are not familiar with the Chow test, it is a F test named after the Princeton economist Gregory Chow who developed it in1960. It’s used to test the equality of two sets of regression coefficients. The Chow test is quite well known since it is taught in every introductory econometrics course .)
