Delaware Chapter, ASA

Dinner Meeting: October 2006


Notice: Oliver's Restaurant and the Holiday Inn is closed. Our October and later meetings will be at the Blue and Gold Club on the University of Delaware campus. For details see --
            http://www.udel.edu/ASA/2006-DE ASA-FALL Program.pdf


Speaker: Larry Hotchkiss
University of Delaware

James Raths
University of Delaware

Joseph Glutting
University of Delaware

Title: Student-Teacher Racial Match and Test-Score Performance

Abstract:

A few recent papers in the educational literature conclude that school children learn more when the race and/or gender of their teacher matches their own race and/or gender. One paper by Thomas Dee published in 2004 in The Review of Economics and Statistics has been widely cited. Dee concludes that matching black teachers to black children in grades K through 3 and white teachers to white children improves the reading and math performance of children of both races. This finding is, as you might imagine, controversial because of potentially emotional issues related to affirmative action and racial segregation of schools.

This talk reports our reanalysis of the Tennesse Star Class-Size Experiment data used in the Dee paper. Our approach to the data analysis differs markedly from the regressions reported by Dee. Dee stacked data for reading and math tests for all four grades, K - 3, into one long data set, including dummy indicators for grade in school and type of test, reading or math. In contrast, we propose a difference-equation learning model and study changes between adjacent grades. Not surprisingly, our results differ in important respects from the conclusions in the Dee paper.

Many papers in the technical literature raise objections to regression models which include lagged dependent variables in the set of regressors. The primary problems are due to the high likelihood that the error term in the regressions are serially correlated. The talk will summarize these issues.

Also, the concept of student-teacher racial match is, in a sense, a compound concept. It depends on the more elementary concepts of the race of the student and the race of the teacher. We find that the idea of a "perfect" match effect implies a linear restriction on a full interaction model which includes student race, teacher race and their interaction. This is an important question because including just a variable for the race of the student and a "race-match" dummy does not reveal whether the benefit of the match goes mostly to white children or black children, or is evenly split.


 
Date: Thursday, October 19, 2006
 
Time:

Social hour -- 6:00 pm
Dinner -- 6:30 pm
Speaker -- 7:15 pm
 
Place:

Blue and Gold Club
University of Delaware   Directions/Map
 
Menu:

Entree: Buffet
Cost: $25 for members, $15 for students

 

Reservations: 

Bruce Stanley by Wednesday, October 18, 2006.
Phone: 302-366-5910
Email: Bruce Stanley.

 
Please note: Be sure to RSVP by this date.
 
The following is a list of the dates of future meetings in 2006 (third Thursday of month (except November) from 6 - 9 pm):
November 9, 2006
December 21, 2006
Return to Delaware Chapter ASA home page.