Policy Analysis Using Interrupted Time Series

Interrupted time series analysis and regression discontinuity designs are two of the most rigorous ways to evaluate policies with routinely collected data. ITSx comprehensively introduces analysts to interrupted time series analysis (ITS) and regression discontinuity designs (RD) from start to finish, including selection and setup of data sources, statistical analysis, interpretation and presentation, and identification of potential pitfalls.

At the conclusion of the course, students will have all the tools necessary to propose, conduct and correctly interpret an analysis using ITS and RD approaches. This will help them position themselves as a go-to person within their company, government department, or academic department as the technical expert on this topic.

ITS and RD designs avoid many of the pitfalls associated with other techniques. As a result of their analytic strength, the use of ITS and RD approaches has been rapidly increasing over the past decade. These studies have cut across the social sciences, including:

• Studying the effect of traffic speed zones on mortality
• Quantifying the impact of incentive payments to workers on productivity
• Assessing whether alcohol policies reduce suicide
• Measuring the impact of incentive payments to physicians on quality of care
• Determining whether the use of HPV vaccination influences adolescent sexual behavior

Week 1: Course overview

• Introduction to ITS and RD designs
• Assumptions and potential biases
• Data sources and requirements
• Example studies
• An introduction to R (optional)

Week 2: Single series ITS

• Data setup and adding variables
• Model selection
• Addressing autocorrelation
• Graphical presentation

Week 3: ITS with a control group

• Data setup
• Adding a control to the model
• Graphical presentation
• Predicting policy impacts

Week 4: Extensions

• Advanced modeling issues in ITS and RD
• Non-linear Trends · Differencing
• “Wild” Points and Transition periods
• Adding a Second Intervention

Week 5: Regression Discontinuities and Wrap-up

• Regression Discontinuities
• Any Remaining Questions