Will Value-Added Data Help Move Out Ineffective Teachers?

In this article in Educational Researcher, Marcus Winters (University of Colorado/ Colorado Springs) and Joshua Cowen (University of Kentucky) examine the efficacy of using value-added data to dismiss ineffective teachers. Drawing on retrospective test-score data from Florida, they come to two conclusions: 

• First, students assigned to teachers who would have been dismissed one or two years earlier based on a single year of value-added data did significantly worse than students assigned to teachers with more positive data. In other words, value-added data can predict many teachers’ future classroom performance.

• Second, the way value-added evaluation is being implemented – waiting for two consecutive years of data before taking job action – undercuts its effectiveness and results in very few ineffective teachers being dismissed. “Consecutive-year policies that set relatively low percentile cutoffs for satisfactory performance will tend to remove very few teachers because even very bad teachers might score above the threshold in 1 of 2 years due to random fluctuation in the estimates of their effectiveness… and many ineffective teachers will remain unidentified,” say Winters and Cowen. “[O]ur evidence also indicates that no system of evaluation will eliminate flaws from the measure of teacher ability.” 

Winters and Cowen conclude that implementing value-added evaluation of teachers will result in few if any improvements in the quality of teaching. 

“Who Should Stay, Who Should Be Dismissed? An Empirical Consideration of Value-Added Teacher Retention Policies” by Marcus Winters and Joshua Cowen in Educational Researcher, August/September 2013 (Vol. 42, #6, p. 330-337), 

http://edr.sagepub.com/content/42/6/330.abstract; the authors can be reached at 

mwinters@uccs.edu and joshuacowen@uky.edu.

From the Marshall Memo #501