Student Seminar Series - October 17, 2005
University of Minnesota
School of Statistics
College of Liberal Arts

A Discussion of Statistical Modeling Options for Evaluating the Effect of Teachers on Student Academic Achievement


Sarah Heimel


Monday, October 17, 2005
8:30 AM, 170 Ford Hall
Minneapolis, East Bank Campus

Refreshments at 8:15 AM
300 Ford Hall


Abstract

Methods of teacher evaluation have been discussed for decades. In America, recent political initiatives, including the No Child Left Behind Act and Minnesota's proposed Q-Comp plan, have pushed the discussion into the public realm. Statistical models, often called "value-added models," have become a common tool in evaluating teachers. These models utilize standardized test score data to determine how much effect a teacher has on a student's academic achievement. There are a variety of these models currently in use across the country, each with advantages and disadvantages. Ultimately, a community that is formulating a statistical method of teacher evaluation should be knowledgeable of both what the community needs and what statistical options will accurately meet those needs. The magnitude of a teacher's estimated effect on students may depend on the chosen statistical model and its assumptions.