Best and worst teachers can be spotted in first 2 years of teaching

Published: Tuesday, March 19 2013 10:45 p.m. MDT

Itineris Early College High School teacher Karen Cavin teaches Math 1050 in West Jordan Wednesday, Feb. 6, 2013.

Jeffrey D. Allred, Deseret News

Enlarge photo»

A new study finds that it doesn't take long to recognize the best and worst teachers. It found that overall teacher effectiveness improves dramatically during the first two to three years of teaching, but that relative effectiveness of individual teachers changes little over the years.

Teachers ranked in the lowest 20 percent of teachers during their first two years of teaching proved likely to be in that same bracket five years later. And teachers whose scores placed them in the top 20 percent of teachers in the study were even more likely to stay in that top bracket, said the study conducted for the National Center for Analysis of Longitudinal Data in Educational Research.

The experiment

Data from 65,000 to 80,000 students in each grade was used for the study, which followed teachers who stayed in the teaching profession five years or longer. The teachers received value-added scores each school year, based on how well their students performed.

Value-added evaluation is a statistical way of judging teachers by comparing their students' test scores from the current school year to scores of those same students in the previous year, and to scores of similar students from a larger sample, said Allison Atteberry, one of the paper's authors. The goal is to isolate a teacher's contribution to a student's progress from other effects on school performance, such as poverty and parental involvement.

"A lot of information is brought to bear to make a good guess at how a kid would have performed with an average teacher," Atteberry said. "Getting that right is at the heart of a good value-added model."

Atteberry was surprised by the predictive strength of the value-added testing used in the study because forecasting teacher success has been a hard nut for researchers to crack, she said.

Predicting success

Many of the likely predictors that a teacher will be successful don't hold up under statistical scrutiny. Atteberry said that whether a teacher did undergraduate work at a competitive school doesn't predict success reliably, and neither does the preparatory pathway a teacher took into the profession — whether through a traditional college program or by an alternative route from another profession. A teacher's SAT scores or credentialing scores aren't strongly predictive of future success, either, Atteberry said.

An administrator's gut feeling about how effective a teacher is likely to become might be superior to any of those indicators, but it's a hard thing to quantify on paper.

"I think there are probably people in the field of teaching who do have a strong, clear predictive sense of how teachers will serve students, but it's difficult to leverage that for a large-scale policy," Atteberry said. "This study is about figuring out what we can learn from data to help us have some of those same kinds of insights."

Value-added measurements of teacher effectiveness are controversial, especially if they are used as the sole basis of teacher evaluation. A resource guide from the National Educators Association said value-added models should not be the principal basis for making high-stakes decisions about pay or tenure of teachers. There is lack of consensus among researchers about whether such measures accurately isolate the effects of a single teacher, the report said.

Results can be skewed because student achievement is influenced by many factors beyond a teacher's control, and students are not necessarily assigned to teachers on a random basis, which compromises the results. NEA recommends that teachers be evaluated by multiple measures, which could include classroom observations, student portfolios and test results. The NEA considers value-added measurements to be acceptable when used as part of a multiple-measure approach.

Keeping "Irreplaceables"

Try out the new DeseretNews.com design!
try beta learn more
Get The Deseret News Everywhere