chowtest from SSC, need some clarification

[Alice Yang]

Hello!

I want to do a chow test to see whether the coefficients estimated over one group of the data are equal to the coefficients estimated over another, and I found this user-written command chowtest. However, when I see the help of this command, there's very little information.

Title

  chowtest --  Chow test for structure break


Syntax

    chowtest varlist [if] [in], group(varname) [ restrict(varlist) het detail ]

    options                   Description
    ------------------------------------------------------------------------------------------
      group(varname)          specify the group variable
      restrict                independents which are restriced to have constant coefficients
      het                     heteroskedasticy
      detail                  display regression result
    ------------------------------------------------------------------------------------------
    varlist are dependent variable and unrestriced independent variables.

Example 

        +-------------------+
    ----+ The auto.dta data +-----------------------------------------------------------------

        . sysuse auto,clear
        . chowtest price wei mpg, group(foreign)
        . chowtest price wei mpg, group(foreign) restrict(headroom)
        . chowtest price wei mpg, group(foreign) het

I just wonder, in using chowtest, do I need to put control variables in, as well as fixed effects? Or just DV, IV and the group variable? The examples given look like there is a DV (price), an IV (wei), and one control variable (mpg).

[Vincent Li]

One more question, I wonder if it functions similarly to -suest- and -test- or -lincom- Xb

[Alice Yang]

I use -chowtest- because -xtreg- is not supported by -suest-. No matter use -test- or -lincom-, need -suest model1 model2- first. That's why I'm seeking another test that could be applied to -xtreg-.