-- Preliminary, Do Not Quote Without Permission -- VALUE-ADDED MODELS AND THE MEASUREMENT OF TEACHER QUALITY Douglas HarrisTim R. Sass Dept. of Ed. LeadershipDept.

Slides:



Advertisements
Similar presentations
Value Added in CPS. What is value added? A measure of the contribution of schooling to student performance Uses statistical techniques to isolate the.
Advertisements

Introduction Describe what panel data is and the reasons for using it in this format Assess the importance of fixed and random effects Examine the Hausman.
Teacher Training, Teacher Quality and Student Achievement Douglas Harris Tim R. Sass Dept. of Educational Dept. of Economics Policy Studies Florida State.
Mark D. Reckase Michigan State University The Evaluation of Teachers and Schools Using the Educator Response Function (ERF)
Random Assignment Experiments
Teacher Effectiveness in Urban Schools Richard Buddin & Gema Zamarro IES Research Conference, June 2010.
Lecture 8 (Ch14) Advanced Panel Data Method
Are Teacher-Level Value- Added Estimates Biased? An Experimental Validation of Non-Experimental Estimates Thomas J. KaneDouglas O. Staiger HGSEDartmouth.
Sampling: Final and Initial Sample Size Determination
Explaining Race Differences in Student Behavior: The Relative Contribution of Student, Peer, and School Characteristics Clara G. Muschkin* and Audrey N.
Using State Longitudinal Data Systems for Education Policy Research : The NC Experience Helen F. Ladd CALDER and Duke University Caldercenter.org
Comparing Two Population Means The Two-Sample T-Test and T-Interval.
“A Unified Framework for Measuring Preferences for Schools and Neighborhoods” Bayer, Ferreira, McMillian.
Technology’s Edge: The Educational Benefits of Computer-Aided Instruction Lisa Barrow Federal Reserve Bank of Chicago Lisa Markman Princeton University.
Using Growth Models for Accountability Pete Goldschmidt, Ph.D. Assistant Professor California State University Northridge Senior Researcher National Center.
Magnet Schools and Peers: Effects on Student Achievement Dale Ballou Vanderbilt University November, 2007 Thanks to Steve Rivkin, Julie Berry Cullen, Adam.
Treatment Effects: What works for Whom? Spyros Konstantopoulos Michigan State University.
Assessments to VAM to VAS to EES Points July 28,
Chapter 9 Hypothesis Testing II. Chapter Outline  Introduction  Hypothesis Testing with Sample Means (Large Samples)  Hypothesis Testing with Sample.
Understanding Student Achievement: The Value of Administrative Data Eric Hanushek Stanford University.
Special Education Teacher Quality and Student Achievement Li Feng Tim R. Sass Dept. of Finance & Econ.Dept. of Economics Texas State UniversityFlorida.
What Makes For a Good Teacher and Who Can Tell? Douglas N. Harris Tim R. Sass Dept. of Ed. Policy Studies Dept. of Economics Univ. of Wisconsin Florida.
Chapter 9 Hypothesis Testing II. Chapter Outline  Introduction  Hypothesis Testing with Sample Means (Large Samples)  Hypothesis Testing with Sample.
Production Functions and Measuring the Effect of Teachers on Student Achievement With Value-Added HSE March 20, 2012.
Analysis of Clustered and Longitudinal Data
Review of normal distribution. Exercise Solution.
Regression and Correlation Methods Judy Zhong Ph.D.
Arizona’s Federal Accountability System 2011 David McNeil Director of Assessment, Accountability and Research.
Student Engagement Survey Results and Analysis June 2011.
The Determinants of Student Achievement: Different Estimates for Different Measures Tim Sass Department of Economics Florida State University CALDER Conference.
T tests comparing two means t tests comparing two means.
Sensitivity of Teacher Value-Added Estimates to Student and Peer Control Variables October 2013 Matthew Johnson Stephen Lipscomb Brian Gill.
Chapter 9 Hypothesis Testing II: two samples Test of significance for sample means (large samples) The difference between “statistical significance” and.
Special Education Teacher Quality and Student Achievement Li Feng Tim R. Sass Dept. of Finance & Econ.Dept. of Economics Texas State UniversityFlorida.
Slide 1 Estimating Performance Below the National Level Applying Simulation Methods to TIMSS Fourth Annual IES Research Conference Dan Sherman, Ph.D. American.
Lesli Scott Ashley Bowers Sue Ellen Hansen Robin Tepper Jacob Survey Research Center, University of Michigan Third International Conference on Establishment.
Introduction Multilevel Analysis
COURSE: JUST 3900 TIPS FOR APLIA Developed By: Ethan Cooper (Lead Tutor) John Lohman Michael Mattocks Aubrey Urwick Chapter : 10 Independent Samples t.
The Inter-temporal Stability of Teacher Effect Estimates J. R. Lockwood Daniel F. McCaffrey Tim R. Sass The RAND Corporation The RAND Corporation Florida.
© 2011, Tulsa Public Schools Copyright © Tulsa Public Schools 2011 © 2011, Tulsa Public Schools Jana Burk, Tulsa Public Schools Fellow Office of Teacher.
Modeling Student Growth Using Multilevel Mixture Item Response Theory Hong Jiao Robert Lissitz University of Maryland Presentation at the 2012 MARCES Conference.
Portability of Teacher Effectiveness across School Settings Zeyu Xu, Umut Ozek, Matthew Corritore May 29, 2016 Bill & Melinda Gates Foundation Evaluation.
Propensity Score Matching for Causal Inference: Possibilities, Limitations, and an Example sean f. reardon MAPSS colloquium March 6, 2007.
Public Policy Analysis ECON 3386 Anant Nyshadham.
DRE FLDOE “Value-Added Model” School District of Palm Beach County Performance Accountability.
Public Finance Seminar Spring 2015, Professor Yinger Public Production Functions.
Impediments to the estimation of teacher value added Steven Rivkin Jun Ishii April 2008.
Strategies for estimating the effects of teacher credentials Helen F. Ladd Based on joint work with Charles Clotfelter and Jacob Vigdor CALDER Conference,
Chapter 8: Confidence Intervals based on a Single Sample
American Educational Research Association Annual Meeting AERA San Diego, CA - April 13-17, 2009 Denise Huang Examining the Relationship between LA's BEST.
Using School Choice Lotteries to Test Measures of School Effectiveness David Deming Harvard University and NBER.
Free Education and Student Test Scores in Chad Gbetonmasse B. Somasse Worcester Polytechnic Institute (WPI) International Conference on Sustainable Development.
Human Capital Policies in Education: Further Research on Teachers and Principals 5 rd Annual CALDER Conference January 27 th, 2012.
Florida Department of Education’s Florida Department of Education’s Teacher Evaluation System Student Learning Growth.
VAM Training. Florida’s value-added model developed by Florida educators  The Department convened a committee of stakeholders (Student Growth Implementation.
Using Prior Scores to Evaluate Bias in Value-Added Models Raj Chetty, Stanford University and NBER John N. Friedman, Brown University and NBER Jonah Rockoff,
Value Added Model Value Added Model. New Standard for Teacher EvaluationsNew Standard for Teacher Evaluations Performance of Students. At least 50% of.
Florida Algebra I EOC Value-Added Model June 2013.
Comparative Analysis of Aggregate Educational Data Between Children in Foster Care and the General Population Florida Department of Children and Families.
 Mark D. Reckase.  Student achievement is a result of the interaction of the student and the educational environment including each teacher.  Teachers.
Chapter 15 Panel Data Models.
Stephen W. Raudenbush University of Chicago December 11, 2006
School Quality and the Black-White Achievement Gap
Educational Analytics
Dr. Robert H. Meyer Research Professor and Director
Ch. 13. Pooled Cross Sections Across Time: Simple Panel Data.
Tutorial 1: Misspecification
What are their purposes? What kinds?
Public Finance Seminar Professor Yinger
Ch. 13. Pooled Cross Sections Across Time: Simple Panel Data.
Presentation transcript:

-- Preliminary, Do Not Quote Without Permission -- VALUE-ADDED MODELS AND THE MEASUREMENT OF TEACHER QUALITY Douglas HarrisTim R. Sass Dept. of Ed. LeadershipDept. of Economics and Policy StudiesFlorida State University Florida State IES Research Conference – June 2006

-- Preliminary, Do Not Quote Without Permission -- Evaluating Value-Added Methodology The recent availability of panel data has produced a flood of research studies using various “value-added” approaches The recent availability of panel data has produced a flood of research studies using various “value-added” approaches Research Questions Research Questions Are assumptions underlying the value-added approach valid? Are assumptions underlying the value-added approach valid? Are some methods more likely to produce reliable estimates than others? Are some methods more likely to produce reliable estimates than others? What data are most important to obtaining consistent estimates? What data are most important to obtaining consistent estimates?

-- Preliminary, Do Not Quote Without Permission -- Evaluating Value-Added Methodology Basic Model Types Basic Model Types Cumulative Model Cumulative Model Unrestricted Value-Added Model Unrestricted Value-Added Model Value-Added Models with Persistence Restrictions Value-Added Models with Persistence Restrictions Restricted Value-Added or “Gain-Score” Model Restricted Value-Added or “Gain-Score” Model Contemporaneous Model Contemporaneous Model Specification Issues for Value-Added Models Specification Issues for Value-Added Models Treatment of teacher heterogeneity Treatment of teacher heterogeneity Measures of classroom/school inputs Measures of classroom/school inputs Treatment of student heterogeneity Treatment of student heterogeneity Aggregation Aggregation

-- Preliminary, Do Not Quote Without Permission -- General Cumulative Model of Student Achievement

-- Preliminary, Do Not Quote Without Permission -- Basic Assumptions of Value-Added Models Cumulative achievement function does not vary with age and is additively separable. Cumulative achievement function does not vary with age and is additively separable. Family inputs are time invariant. Family inputs are time invariant. Parents do not compensate for poor school inputs or poor outcomes Parents do not compensate for poor school inputs or poor outcomes Todd and Wolpin (2005) reject exogeneity of parental inputs at 90 percent, but not at 95 percent confidence level Todd and Wolpin (2005) reject exogeneity of parental inputs at 90 percent, but not at 95 percent confidence level The marginal inputs of all school-based inputs, parental inputs, and the initial student endowment each decline geometrically (at potentially different rates) over time. The marginal inputs of all school-based inputs, parental inputs, and the initial student endowment each decline geometrically (at potentially different rates) over time. Lagged achievement serves as a sufficient statistic for prior inputs Lagged achievement serves as a sufficient statistic for prior inputs We find twice-lagged inputs do not provide additional information We find twice-lagged inputs do not provide additional information

-- Preliminary, Do Not Quote Without Permission -- Unrestricted Value-Added Model

-- Preliminary, Do Not Quote Without Permission -- Persistence Restrictions Restricted Value-Added or “Gain-Score” Model Restricted Value-Added or “Gain-Score” Model is assumed to equal 1 (no decay in effect of past inputs) is assumed to equal 1 (no decay in effect of past inputs) Alternatively, can interpret as an achievement growth model where growth is independent of past school inputs. Alternatively, can interpret as an achievement growth model where growth is independent of past school inputs. Contemporaneous Model Contemporaneous Model is assumed to equal 0 (complete decay in effect of past inputs). is assumed to equal 0 (complete decay in effect of past inputs).

-- Preliminary, Do Not Quote Without Permission -- Decomposition of School-based Inputs in Value-Added Model

-- Preliminary, Do Not Quote Without Permission -- Modeling Teacher Heterogeneity Substituting teacher time-invariant measured characteristics for teacher fixed effects

-- Preliminary, Do Not Quote Without Permission -- Classroom and School Inputs Exclusion of peer variables (P -ijmt ) Number of peers (class size) and peer characteristics (gender, race, mobility, age) If peer variables are correlated with student and teacher characteristics (X it and T kt ), omission will produce inconsistent estimates Exclusion of school fixed effects (  m ) Given that teachers do not frequently change schools, omission of school effects will mean that teacher fixed effects will capture both teacher effects and some of the school effect, leading to inconsistent estimates

-- Preliminary, Do Not Quote Without Permission -- Modeling Student Heterogeneity Substituting measured time-invariant student characteristics for student fixed effects Race/ethnicity, foreign/native born, language parent speak at home, free-lunch status As with teachers, if unmeasured time-invariant student characteristics are correlated with independent variables, will get inconsistent estimates

-- Preliminary, Do Not Quote Without Permission -- Modeling Student Heterogeneity Fixed vs. random student effects Fixed effects allow for a separate intercept parameter for each student (equal to the mean error for that student) whereas random effects assume that the student-specific intercepts are drawn from a known distribution (typically normal) Since random effects are part of the error structure, they must be orthogonal to the model variables (X it, P -ijmt, T kt ) in order to yield consistent estimates Given that fixed effects estimates are always consistent (whether or not unobserved student heterogeneity is correlated with other variables in the model), can test orthogonality assumption by applying a Hausman test Multilevel fixed effects models have been computationally burdensome

-- Preliminary, Do Not Quote Without Permission -- Aggregation Measuring characteristics of specific teachers vs. grade-level- within-school averages Since Texas data does not identify specific teacher, work by Rivkin, Hanushek and Kain (2005) relies on average characteristics of teachers within a grade Advantages/Disadvantages of aggregation Eliminates problems associated with non-random assignment of students to teachers within a school May reduce problem of measurement error since individual errors may cancel out at grade level May upwardly bias estimated impacts of school resources in the presence of omitted variables Tends to reduce precision of estimates

-- Preliminary, Do Not Quote Without Permission -- Data Florida’s K-20 Education Data Warehouse Florida’s K-20 Education Data Warehouse Census of all children attending public schools in Florida Census of all children attending public schools in Florida Student records linked over time Student records linked over time Covers 1995/1996 – 2003/2004 school years Covers 1995/1996 – 2003/2004 school years Includes student test scores and student demographic data, plus enrollment, attendance, disciplinary actions and participation in special education and limited English proficiency programs Includes student test scores and student demographic data, plus enrollment, attendance, disciplinary actions and participation in special education and limited English proficiency programs Includes all employee records including individual teacher characteristics and means of linking students and teachers to specific classrooms Includes all employee records including individual teacher characteristics and means of linking students and teachers to specific classrooms

-- Preliminary, Do Not Quote Without Permission -- Sample for Analysis Middle school students (grades 6-8) who took SSS- NRT (Stanford-9) math test in three consecutive years during 1999/2000 – 2003/2004 Middle school students (grades 6-8) who took SSS- NRT (Stanford-9) math test in three consecutive years during 1999/2000 – 2003/2004 Enrolled in a single math course in the Fall Enrolled in a single math course in the Fall Up to 4 years of achievement gains Up to 4 years of achievement gains 4 cohorts of students 4 cohorts of students Includes a variety of math courses, from remedial to advanced and gifted classes Includes a variety of math courses, from remedial to advanced and gifted classes Use random sample of 100 middle schools Use random sample of 100 middle schools Reduces computational burden of estimating fixed effects Reduces computational burden of estimating fixed effects Represents about 12% of middle schools in state Represents about 12% of middle schools in state

-- Preliminary, Do Not Quote Without Permission -- Value-Added Model Estimates With Varying Degrees of Persistence

-- Preliminary, Do Not Quote Without Permission -- Correlation of Estimated Teacher Effects From Models with Varying Degrees of Persistence

-- Preliminary, Do Not Quote Without Permission -- Restricted Value-Added Model Estimates With Differing Controls for Teacher Heterogeneity

-- Preliminary, Do Not Quote Without Permission -- Restricted Value-Added Model Estimates With Differing Classroom/School Controls

-- Preliminary, Do Not Quote Without Permission -- Correlation of Estimated Teacher Effects From Models with Differing Classroom/School Controls

-- Preliminary, Do Not Quote Without Permission -- Restricted Value-Added Model Estimates with Differing Controls for Student Heterogeneity

-- Preliminary, Do Not Quote Without Permission -- Correlation of Estimated Teacher Effects From Models With Differing Controls for Student Heterogeneity

-- Preliminary, Do Not Quote Without Permission -- Restricted Value-Added Model Estimates -- Teacher-Specific vs. Within-School Grade-Level Averages

-- Preliminary, Do Not Quote Without Permission -- Summary of Findings Model Selection Model Selection Restricted value-added model seems to be a good approximation of the full cumulative model Restricted value-added model seems to be a good approximation of the full cumulative model Specification Specification Use of student and teacher fixed effects (rather than covariates) important Use of student and teacher fixed effects (rather than covariates) important Random effects may yield inconsistent estimates Random effects may yield inconsistent estimates Important to include school fixed effects, but classroom peer variables relatively unimportant Important to include school fixed effects, but classroom peer variables relatively unimportant Aggregation to the grade level has some effect, though estimates not radically different from estimates with teacher- level data Aggregation to the grade level has some effect, though estimates not radically different from estimates with teacher- level data