Interpreting Tests of School VAM Validity

New Findings, School Reform, May 2016

Public school dis­tricts increas­ingly look to the esti­mates gen­er­ated by value-added mod­els (VAMs) to mea­sure school and teacher qual­ity. A typ­i­cal VAM com­pares aver­age stu­dent test scores across teach­ers or schools while using regres­sion mod­els to con­trol for stu­dents’ past scores and demo­graphic char­ac­ter­is­tics. The result­ing esti­mates serve as inputs into teacher reten­tion and pro­mo­tion poli­cies, school report card sys­tems, and deci­sions about which schools to close, restruc­ture, or expand. This paper lays out the econo­met­ric the­ory behind lottery-based VAM validity tests.