"When Rhee set the percentage at 50 percent back in 2009, after spending millions of dollars to create the assessment system, she based it on absolutely nothing grounded in research. The 35 percent? Also based on no research."
I have recently argued that test-based teacher evaluations and pay are without credibility (HERE and HERE).
Teachers, parents, and the public who are genuinely concerned about public school quality must enter directly the teacher evaluation/pay debate, but do so informed with the following foundational facts, messages, and questions:
(1) When anyone argues that teacher evaluation/pay must be reformed, ask for evidence of why first. No one has ever established that teacher quality is the most pressing problem in education—or even one of the most pressing problems. Demand that education solutions begin with establishing evidence-based problems to be solved.
(2) Teacher quality is a subset of school quality, and the evidence on how much teacher quality impacts MEASURABLE student outcomes is significantly less than many plans to hold teachers accountable for student test scores (lowered from 50% to 35% in DC and piloting a plan using 40% in SC). The Joseph Rowntree Foundation has shown that school/teacher quality accounts for about 14% of measurable student outcome; conservative researcher and merit-pay advocate Hanushek  shows that teacher quality accounts for about 13%-17% of measurable student outcomes. No one can justify holding teachers accountable for student test scores at rates in the teens, much less 35%-50% (although, as I will note below, there is no compelling evidence teachers should be held accountable at all for student test scores).
(3) Holding teachers accountable for student test scores is often posed as evidence-based reform and replacing a subjective system with an objective one. Yet, basic statistics and testing protocol do not support using a test designed for one primary purpose (student testing is designed to measure student learning—and is only a pale representation of that sole purpose) for a different purpose . Second, standardized testing gives the appearance of being "objective" simply because it is quantitative. Standardized testing remains highly biased in ways that disserve the populations of students already being mis-served in society and schools.
(4) Labeling, ranking, and sorting are inherently flawed practices, in both the business world and in education.
(5) Since no one has established that teacher quality is a central problem in education, the excessive cost in time and money needed to create a test-based system that would capture in some fair way teacher quality is disproportionate to the current needs in education reform.
(6) Finally, rejecting test-based teacher evaluation/pay is NOT rejecting teacher accountability. Rejecting test-based teacher evaluation/pay is rejecting a misguided use of high-stakes testing. Teachers should be held accountable for their teaching, but student test scores are not valid proxies of teaching quality.
 Hanushek, E. (2010, December). The economic value of higher teacher quality. Working paper 56. Washington, DC: Calder, The Urban Institute.
 See Bracey.