How to test for significant difference between paths in SEM?

[Frank Flieger]

Hi,

how can I test for significant difference between two paths in SEM? I have four paths in a model with 4 variables, and the standardized coefficients of two paths are kind of similar.

Chat-GPT suggests a Wald test like this: test (_b[y1:x1] = _b[y2:x1]) but Stata answers "equation y1 not found

Help appreciated!

[Clyde Schechter]

The names that are used for the various coefficients in _b are not always easily predictable in complicated models, and different models use different schemes for doing it. Even the same model may do it differently in different versions of Stata. So you need to find out what the right names are.

Rerun your -sem- command, this time adding the -coeflegend- option. You will get the coefficients, without the inferential statistics, but Stata will show you their names. Then you can write your -test- command using those names and it will work.

[Frank Flieger]

great, thanks!