This project addresses the challenges for estimating effects of treatment in an educational experiment with high attrition rates. The project discusses the study of Principal Stratification, focusing on estimating the Average Treatment Effect of Always Post-test stratum - students who would take the post-test no matter which experimental group they belonged to. Since the previous methods for identifying that stratum resulted in principal scores (probabilities) larger than 1, new methods including Multinomial Regression and Expectation-Maximization algorithm are introduced in this project. In the experiment, middle school students used interventions including to enhance their math skills. The result shows that the treatment had a positive impact on post-test math scores.

• This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.

Creator

• Tao, Anzhe

Subject

• Learning

Publisher

• Worcester Polytechnic Institute

Identifier

• 115278
• E-project-121523-122954

Keyword

• Attrition
• Randomized Controlled Trial
• Principal Stratification
• Multinomial Regression
• Expectation-Maximization Algorithm
• Average Treatment Effect

Advisor

• Sales, Adam

Year

• 2023

UN Sustainable Development Goals

• 4 - Quality Education

Date created

• 2023-12-15

Resource type

• Major Qualifying Project

Major

• Mathematical Sciences

Source

• E-project-121523-122954

Rights statement

• In Copyright

Last modified

• 2024-02-07