balancing tests after collapse command.

[Ayse Uygun]

Hello, I wanna do balancing test between 2 different groups. but in my case, after i collapsed my variables, i'm running the tests but giving me error about "more than 2 groups found, only 2 allowed" because I'm also using if commang while usong ttwst command. How can I solve this problem.

[Ayse Uygun]

Plus, when reduce my groups, i have this error

[Hemanshu Kumar]

What is the output of

tab treatment if sending_week == 1, missing

[Ayse Uygun]

here it is, total observation must be 1329 but here is 665? Am I missing something?

collapse (mean) ID_numeric atleast1scanning_12h atleast1opening_12h scanning_12h_week opening_12h_week SR_waste_responsible SR_weekly_disposals_num stay_1_year_numeric compost_at_home nature_and_you_vignete_num years_group SR_morally SR_free_time SR_guilty SR_resp_others SR_annoyed SR_responible_self SR_kantian_1 SR_kantian_2 SR_ecologist sending_week SMS_received treatment POST, by(ID_num)

this is the code I used before the ttest

[Ayse Uygun]

What is the output of

tab treatment if sending_week == 1, missing

this is the code I used before the ttest:

  collapse (mean) ID_numeric atleast1scanning_12h atleast1opening_12h scanning_12h_week opening_12h_week SR_waste_responsible SR_weekly_disposals_num stay_1_year_numeric compost_at_home nature_and_you_vignete_num years_group SR_morally SR_free_time SR_guilty SR_resp_others SR_annoyed SR_responible_self SR_kantian_1 SR_kantian_2 SR_ecologist sending_week SMS_received treatment POST, by(ID_num)

results not showing truth, obvervation must 1329 in total

[Hemanshu Kumar]

You said you want to do a balance test between two different groups, but as you can see, your treatment defines three groups. If treatment == 0 represents your control group, balance tests will typically want to compare 1 with 0 and 2 with 0. To do the first, you would do

ttest atleast1scanning_12h if sending_week == 1 & inlist(treatment, 0, 1), by(treatment)

and similarly for 2 vs. 0. To simultaneously check equality of means across all three groups, you would need an F-test; a t-test will not do it.

[Ayse Uygun]

You said you want to do a balance test between two different groups, but as you can see, your treatment defines three groups. If treatment == 0 represents your control group, balance tests will typically want to compare 1 with 0 and 2 with 0. To do the first, you would do

ttest atleast1scanning_12h if sending_week == 1 & inlist(treatment, 0, 1), by(treatment)

and similarly for 2 vs. 0. To simultaneously check equality of means across all three groups, you would need an F-test; a t-test will not do it.

Hello Hemanshu, Can I have your contact mail/number?