Non statistically significance in interaction variable for DID-model.

[Daniel Stockholm]

I conducted a regression to study property prices over time. First, I created a graph comparing house prices for the control and treatment groups. It's clear that prices significantly increased for the treatment group after the intervention (see the first picture).

Next, I performed a difference-in-difference regression using the command reg Price treatment##i.time_dummy treatment i.time_dummy. The 'time_dummy' variable showed statistically insignificant results before the treatment, and then statistically significant results after. However, the 'treatment##i.time_dummy' interaction effect, which measures how the treatment's impact changes over time, didn't yield statistically significant results neither before or after. But we do see a clear change in prices between the treatment and control group after the treatment!

I'm seeking insights on why this occurred and possible solutions. Any assistance would be greatly appreciated!

[Andrew Musau]

The graph itself cannot tell you whether there is a significant treatment effect. In addition, you also see an increase in the control group. You may want to include covariates in the outcome model and this may yet change things. Also look at

help didregress

if you have Stata 17 or later.