USE CASES
Start with the decision.
Then test the design.
Explore four study designs, the data each needs, and the questions to settle before analysis. For a worked example with data and computed results, open the sample report.
RESEARCH QUESTION
Policy & program evaluation
Did a phased program change the outcome it was intended to improve?
Typical units
Regions, schools or organizations observed over time.
What to bring
A clear start date for each participating unit, consistent outcomes, and credible untreated or not-yet-treated comparisons.
Questions to settle
Why did some regions join earlier? Record other policy changes and check whether nearby comparison regions were also affected. These answers help define the comparison group.
I want to evaluate a program introduced at different times across regions. Help me define the treatment, comparison group and assumptions before selecting a method.
RESEARCH QUESTION
Business & operational rollouts
Did the rollout improve performance beyond the movement we would otherwise expect?
Typical units
Stores, branches, business units or customer groups observed over time.
What to bring
Consistent measurement, rollout dates, and a reason why the comparison units are informative about the participating units.
Questions to settle
Were early locations chosen for their growth potential? Add a timeline of pricing, promotions and closures so the rollout can be separated from other changes in performance.
We rolled out an operational change in stages. Help me inspect timing and support, and distinguish a rollout effect from common market changes.
RESEARCH QUESTION
Applied & academic research
What effect is identified, for which population, and under which assumptions?
Typical units
A documented panel or another explicitly supported study design.
What to bring
A declared estimand, variable definitions, reproducible data preparation, and a specification chosen before inspecting favorable results.
Questions to settle
Should the result describe an average effect, its evolution over time, or a particular cohort? Keep that target consistent when comparing methods.
Help me state an overall versus dynamic target, confirm the schema, and draft an analysis plan that preserves the target across eligible comparisons.
RESEARCH QUESTION
Place-based interventions
Did a local intervention change outcomes relative to credible comparison locations?
Typical units
Geographies or local institutions with stable definitions over time.
What to bring
A stable unit boundary, an outcome measured comparably before and after, and attention to nearby places affected indirectly.
Questions to settle
Check whether boundaries, population or nearby locations changed during the study. For one treated place, ask whether a supported synthetic comparison or time-series design fits the available data.
I have outcomes by location and time around a local intervention. First help me assess whether the unit count, comparison group and spillover assumptions support a DiD design.