If you are unable to attend this course meeting in person, the slides and a video will be available here:
http://blogs.cam.abtassoc.com/MethodsCenters/?page_id=1522
(previous recordings and presentations from this course are also available)
The Monitoring and Evaluation Methods Center
presents:
A course: Power Analysis and Design for Random Assignment Studies
Jacob Alex Klerman, Instructor
Power analysis answers the two fundamental questions in research design:
1. How precise will my estimates be?
2. How does that vary with the details of the design?
Formal estimates of precision for a given design is a key component of almost all proposals and analysis plans to estimate program impact. Understanding how precision varies with the design is crucial to identifying the best design. Closely related questions arise in designing surveys; this course will make those connections.
The Power Analysis course develops the basics of power analysis (in particular the computation of Minimum Detectable Effects (MDEs)) and the insights of those computations for the design of random assignment experiments. Topics to be covered include:
Unbalanced treatment/control ratios, varying treatment/control ratios over the term of an experiment
Stratified sampling and unequal sampling weights, regression adjustment, differential subgroup impacts
Clustered data; for simple treatment/control comparisons, for regression models, and for other statistics
While random assignment studies are the primary focus of the course (and the simplest context in which to estimate power), the course will also consider similar computations for observations studies of impact and descriptive surveys.
Course Participants:
This is not a course for “consumers”; this is a course for “producers”.
Participants in the course will develop their own Excel spreadsheets for doing power calculations and use them to do a range of those computations.
The course will not be “plug and chug”; instead, the course will introduce some basic ideas from mathematical statistics and then use those ideas to derive the formulas used in the spreadsheets (and that could be used to do similar calculations for non-standard situations).
Course Prerequisites:
Basic working knowledge of statistics (e.g., mean, standard deviation, variance),
Basic familiarity with Excel (formulas, functions, absolute vs. relative reference)
Comfort with pre-calculus (algebra, summations, square roots).
Those unsure about the prerequisites are encouraged to speak to Jacob Klerman, the instructor.
Course Structure:
6 one-hour sessions, from noon until 1PM:
Thursday October 23
Wednesday October 29
Wednesday November 5
Wednesday November 12
Thursday November 20
Tuesday December 2
Wednesday December 10
Wednesday December 17
Information about the series, and a space for online discussion is available on the Methods Center website, here:
http://blogs.cam.abtassoc.com/MethodsCenters/?page_id=1522
12:00 – 1:00 PM
Locations
Georgetown Room, Bethesda; Ramp Room 1, Cambridge; and Hatteras Room, Durham
Remotely:
https://www1.gotomeeting.com/join/732798833
Call in or use the speakers on your computer: 1-213 493-0619
Access code: 732-798-833
(reminder: all callers will be muted. Please type any questions in the GoToMeeting chat box)