Academic journal article Phi Delta Kappan

Should Value-Added Measures Be Used for Performance Pay? Value-Added Measures Are Becoming a Critical Piece of Compensation Reform, but Buyers Should Be Wary That Any Single Number Can Express the Effect Teachers Have on Students

Academic journal article Phi Delta Kappan

Should Value-Added Measures Be Used for Performance Pay? Value-Added Measures Are Becoming a Critical Piece of Compensation Reform, but Buyers Should Be Wary That Any Single Number Can Express the Effect Teachers Have on Students

Article excerpt

The current teacher compensation system is obsolete. On that point, there appears to be wide agreement. Virtually every new idea for improving teacher compensation focuses on how to tie teacher pay to teacher effectiveness as measured by improving student learning.

The so-called value-added model is the idea du jour. Value-added models vary in complexity, but they're generally based on an intuitively acceptable premise: By observing at least two data points at different times on vertically aligned tests, we can determine a student's growth over time. This, in turn, can be attributed to the effect of the teacher. The concept of pre- and posttesting students to determine growth is not new. Attributing that growth in a reliable and valid way to teacher effectiveness is the latest potential use of these data.

Evaluating teachers based on student performance makes sense. However, there's little evidence for the effect that this will have. Linking student performance with teacher pay before educators and data systems are ready is comparable to distributing an experimental drug before it's been adequately tested (Farmer 2009). But value-added measures should be included along with other measures of teacher effectiveness in an effort to reform the current approach to teacher compensation. Policy makers, administrators, and teachers must answer three questions in order to move forward on this:

* What are the current limits of value-added measures?

* What combination of measures could be used to determine teaching effectiveness?

* How can we move away from divisive rhetoric toward necessary teacher compensation reform?

THE LIMITS

Tennessee has the longest history of using a value-added system tied to state assessments. The state began using Bill Sanders' model in the early 1990s. Sanders' Tennessee Value-Added Assessment System (TVAAS) is one of the most sophisticated and respected value-added models in use. Tennessee uses TVAAS in grades four through eight in math, reading, science, and social studies. In fact, in an attempt to receive Race to the Top funds from the U.S. Department of Education, Tennessee has passed legislation that requires 35% to 50% of a teacher's evaluation to be based on TVAAS data if it is available. TVAAS is also used in Pennsylvania, North Carolina, and local districts across the country.

However, the model has been criticized for its lack of external review, lack of transparency, issues with missing data, and its lack of consideration of student background variables (Amrein-Beardsley 2008). Indeed, an important part of the model is cloaked in proprietary secrecy. In addition, recent analysis shows that value-added models such as TVAAS can retroactively affect a student's prior year of growth data. In other words, a 5th-grade teacher can affect a student's 4th-grade score. Thus, we need to question the validity of such measures (Rothstein 2008).

Value-added measures must have three elements to be viable:

1. Close, but not perfect alignment of assessments and curriculum;

2. Appropriate reliabilities; and

3. Sufficient "stretch" in the assessments to show growth (Sanders 2003).

Sufficient stretch means assessments can't have ceiling effects that prevent a student from showing the full extent of his or her knowledge. Koedel and Betts (2008) question the ability of current state assessment measures to provide sufficient stretch. "If different tests emphasize different kinds of material or have different effective maximums or 'test ceilings,' the measure effect of a teacher can vary depending on the test instrument being used" (Lockwood, McCaffrey, and Sass 2008:14). Considering all three requirements, current assessment measures may not be sufficient or appropriate to tie to performance pay (Ballou 2002).

Sanders' and TVAAS deserve credit for advancing thinking about assessment beyond the traditional "snapshot of a student" toward a measurement of growth over time. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.