Test-Based Teacher Evaluation Earns F, Again
Teacher quality, including teacher evaluation and pay scales, is a growing national and state concern in education reform -- a trend that can be traced to, and remains reflected in, the Los Angeles Times' repeated publication of value-added models (VAM) of ranking teacher quality.
Many researchers have raised concerns regarding the use of value-added models (VAM) for teacher evaluation. Briefly, VAM do not provide guidance for improvement, are comparative rather than absolute measures, assess a small part of teacher’s responsibilities, force different kinds of teaching into one scale, do not produce consistent results for given teachers over time, and may not identify effects actually caused by the teachers.
As noted above, the relatively new but growing body of evidence on the validity and reliability of test-based teacher evaluations reveals that data are unstable ; in other words, as the populations of students change or the school settings change, the rankings of the teachers fluctuate. Test-based teacher evaluation can be of value only if it can offer a stable message that a teacher is strong or weak. If that label isn’t predictive, it has no positive contribution to policy and personnel decisions.
In order to implement test-based teacher evaluation that shows student growth, SC will have to create and implement two tests per course for every teacher in the state and in every content area taught. Since there is no compelling evidence test-based teacher evaluation data are stable or valid, this investment in time and money is a catastrophic failure by our state leaders. How will we identify growth in music, P.E. and art, and how can we justify the costs associated with generating all of these tests?
Test-based teacher evaluations create a competitive environment in which teachers must choose between the welfare of their students and their own professional security as that becomes threatened by the outcomes of other teachers and their students. In effect, each teacher must seek to use her/his students against the outcomes of other teachers’ students for their own personal gain. Education is best served by a collaborative, not competitive, environment.
- Test-based evaluations of teachers place far too much weight on flawed assumptions. First, high-stakes testing distorts how well any tests reflect student learning. Next, high-stakes testing decreases the quality of both teaching and learning since it encourages teaching to the test. Further, linking teacher evaluations to student outcomes confuses “high-quality teaching” with student test scores, although we have no way to insure that teacher quality is always positively correlated with those scores. And finally, test-based evaluations of teachers imply that student outcomes are or can be linked only to the teacher’s room that any student sits in when the test is administered. Here is the great failure of test-based evaluations of teachers: Data linked to a student are correlated with dozens of conditions (that cannot be controlled for) that distort any one teacher’s quality.