How to obtain Wald chi-square from running logistic regressions?

[Weidi Qin]

Hello statalists,

I was an SPSS user and switched to STATA a couple years ago. I have a question about logistic regression. Below is a screenshot of an SPSS-run logistic regression output:

My questions is, when I run logistic regressions using STATA, how can I obtain the Wald statistics and df for each dependent variable that are shown in this SPSS output?

I know that I can manually calculate the Wald statistics by squaring the z statistics reported in STATA, but is there any chance that STATA can generate it?

I no longer have SPSS in my computer and it would be really helpful if I know how to get these numbers.

Thank you for your help.

[Clyde Schechter]

I am not aware of any option to the Stata logistic regression command that will give you the Wald chi square. As you note, it is simply the square of the z-statistic. You don't have to calculate those by hand, though. You can automate that process in a loop something like this:

foreach x in age weight gender VO2max _cons {
    local wald= (_b[`x']/_se[`x'])^2
    display "Wald statistic for `x' = " %4.3f `wald'
}

Note: You may need to change the names used for the variables in the -foreach- command to make them match what -logit- puts in _b[] and _se[]. So before writing this code, rerun your logitic regression with the -coeflegend- option to see what those are.

[Hong Il Yoo]

The z statistic that Stata reports is a Wald statistic. If for some reason you would like to have it in the same format as what SPSS reports, you can square your z statistic.

[Weidi Qin]

I am not aware of any option to the Stata logistic regression command that will give you the Wald chi square. As you note, it is simply the square of the z-statistic. You don't have to calculate those by hand, though. You can automate that process in a loop something like this:

foreach x in age weight gender VO2max _cons {
local wald= (_b[`x']/_se[`x'])^2
display "Wald statistic for `x' = " %4.3f `wald'
}

Note: You may need to change the names used for the variables in the -foreach- command to make them match what -logit- puts in _b[] and _se[]. So before writing this code, rerun your logitic regression with the -coeflegend- option to see what those are.

Thank you. This is very helpful!