t-Test for equivalence

[Azriel Alfajri]

Hi Everyone,

I've conducted logistic regression to examine life goals and socio-demographic variables on happiness in 2007 and 2014. And now i'm struggling with t-test for equivalence to determine whether the results between 2007 and 2014 were significantly differences. I've run paired t-test also McNemar's test for binary, but I'm not sure that i'm doing this right. Do you know, which t-test i should perform? and also do you have any examples?

thanks in advance!

[Carlo Lazzaro]

Azriel:
I'm not clear with your dataset being a panel or a cross-sectional one.
That said, the best way to get what you want is to interact variables with -i.time- in a unique logistic regression.

[Azriel Alfajri]

Dear Carlo,

my data structure is balanced panel, with exactly 7747 obs per period (2007 & 2014)

[Carlo Lazzaro]

Azriel:
go -xtlogit- with interactions (-i.year-) then.

[Leonardo Guizzetti]

Also as a note, equivalence is not the same thing as equality when it comes to statistical inference. You want tests of equality based on the other tests you have said you've run and the limited description in your post.