Dear all,

My study is concerned with the causal effect of development aid on support for democracy in a given geographical area. I theorize that receiving development aid will have a positive impact on local support for democracy. In order to test this theory I match spatial data on the location of aid projects to survey data on local citizens' support for democracy.

Aid is not randomly allocated and in order to account for the potential selection effects of aid targeting, as well as pre-correlated levels of local support for democracy, I use a difference in difference type design where I compare differences between three groups (a group that presently receives aid, a group that will receive aid and a group that does not receive aid) in order to isolate the causal effect of aid from potential selection effects.

I believe I can use the "geonear" command to match respondents to aid projects which are within a certain cutoff distance of aid projects.

However, I am unsure of the structure my hypothesis should take for a difference in difference type design as I have not used stata to run such an analysis before.

Would this be acceptable:

Citizens in areas where aid projects are active will have greater levels of support for democracy than citizens in areas where aid projects will be implemented in the future but are not currently active.

Or, do I also need to refer to my control condition in analysis, something like:

**Testing is based on comparisons to a control group in which citizens are not exposed to aid projects

Any help would be greatly appreciated.