Latent structurePsychometricsEducational AssessmentModel

Value-Added Modeling

Also known as: VAM

OriginatorWilliam Sanders, Sandra HornYear1998Sources3Related methods12

Value-Added Modeling (VAM) is a method for assessing the contribution of schools or teachers to student achievement growth, developed by Sanders and Horn (1998). VAM isolates the effect of a teacher or school by comparing student gains (value added) while controlling for prior achievement and student characteristics.

Key highlights

  • Isolates teacher effect: attempts to separate teacher contribution from student selection
  • Uses growth: measures learning gains rather than absolute achievement levels
  • Accounts for baseline: controls for students' starting points
  • Policy-relevant: directly addresses whether specific teachers improve student outcomes

Intuition

This section is available to Pro members. Upgrade to Pro

How it works

This section is available to Pro members. Upgrade to Pro

When to use it

Apply VAM when evaluating teacher or school effectiveness using growth-based measures, when you want to account for student composition differences across teachers, or when predicting student outcomes under different teachers. Most common in K-12 education policy and evaluation.

Strengths & limitations

Strengths
  • Isolates teacher effect: attempts to separate teacher contribution from student selection
  • Uses growth: measures learning gains rather than absolute achievement levels
  • Accounts for baseline: controls for students' starting points
  • Policy-relevant: directly addresses whether specific teachers improve student outcomes
Limitations
  • Specification sensitivity: results depend heavily on which controls are included; different models yield different rankings
  • Bias from unmeasured confounds: cannot control for unmeasured factors (motivation, parental involvement) that affect both sorting and learning
  • Instability: teacher rankings can fluctuate dramatically year-to-year due to random variation
  • Fairness concerns: teachers in challenging contexts may appear less effective even if working as hard as others

Common pitfalls

This section is available to Pro members. Upgrade to Pro

Applications

This section is available to Pro members. Upgrade to Pro

Frequently asked

What should I include as controls in VAM?

Include student demographics (poverty, race, language, special education status), prior achievement, and family factors. Do not include factors directly caused by the teacher (current grades, attendance, behavior referrals). Sensitivity to controls is a major VAM limitation.

What is the standard error of a VAM estimate?

VAM estimates have large standard errors, especially for teachers with small classes or high student turnover. A teacher's 'true' value-added is usually 2-4 times the estimated standard error away from the point estimate.

Can VAM account for peer effects?

Standard VAM cannot. If student A's classmates are high-achieving, A learns more from peer effects, which may be attributed to the teacher. Some advanced models attempt to control for peer composition.

Is VAM appropriate for non-standardized tests?

VAM requires comparable outcome measures across years. Works best with standardized tests. Open-ended assessments or portfolio-based outcomes are harder to use in VAM.

How do I communicate VAM results fairly?

Avoid rankings. Report confidence intervals. Note that VAM estimates have substantial year-to-year volatility. Use as one input among multiple sources (observations, surveys, student growth) for evaluations.

Sources

  1. 1.
    Kane, T. J., Rockoff, J. E., & Staiger, D. O. (2008). What does certification tell us about teacher effectiveness? Evidence from New York City. Economics of Education Review, 27(6), 615-631.
  2. 2.
    Sanders, W. L., & Horn, S. P. (1998). Research findings on classroom heterogeneity and achievement. Journal of Educational Research, 91(5), 294-303.
  3. 3.
    Koedel, C., Mihaly, K., & Rockoff, J. E. (2015). Value-added modeling: A review. Economics of Education Review, 47, 180-195.

You have read it. What now?

Cite this page

ScholarGate. (2026, June 3). Value-Added Modeling. ScholarGate. https://scholargate.app/psychometrics/value-added-modeling