Baseline test

[Nirosha Dissanayaka]

Hi All,
I am trying to see the impact of a development program that was introduced to a district in 2006. I use diff-in-diff method in this regard.

So I consider households in this district as my treatment group. There are 25 districts in this country. So, I want to find households in another district as my control group. I need to perform a balance test/ baseline test in order to confirm that both treatment and control groups are not different before the program intervention.

Can anyone help me with how to perform this in stata.

I need to see my table as follows;

Example:

Variables	Mean Control	Mean Treatment	Difference	t-test	P-value
Age	45.391	45.104	_0.287	0.5	0.6181

Thank you in advance...

[Rich Goldstein]

the "goal" of DiD is to see whether the pre-existing difference changes after the introduction of the new program; given that, why do you want to do what you say you want?

[Nirosha Dissanayaka]

Hi Mr Rich,

Yes, as you said the goal of DiD is to see the program's impact.
To see the real impact, we need to see there are no significant differences between the two districts (treatment and control group) before the programme intervention.
Then only we can say that the program caused to increase/decrease our outcome variable, right?

Therefore, I need to find a control group to perform the study. To find the control group(district), I need to find another district that does not show statistical differences from the treatment district(in terms of my varibales) at the baseline year which is 2006.


Thank you

[Rich Goldstein]

apparently I was not clear enough; rather than trying again, however, I point you to (1) the Stata manual on (starts on p. 5 of te.pdf) and (2) Wing, C, et al. (2018), "Designing Difference in Difference Studies", Annual Review of Public Health, 39: 453-469