Tennessee’s new way of evaluating classrooms “systematically failed” to identify bad teachers and provide them more training, according to a state report published Monday.

The Tennessee Department of Education found that instructors who got failing grades when measured by their students’ test scores tended to get much higher marks from principals who watched them in classrooms. State officials expected to see similar scores from both methods.

“Evaluators are telling teachers they exceed expectations in their observation feedback when in fact student outcomes paint a very different picture,” the report states. “This behavior skirts managerial responsibility.”

The data revealed:

• More than 75 percent of teachers received scores of 4 or 5 — the highest possible — from their principals, compared with 50 percent scoring 4 or 5 based on student learning gains measured on tests.

• Fewer than 2.5 percent scored a 1 or 2 when observed, while 16 percent scored a 1 or 2 when judged by learning gains.

• Of teachers who received the learning gains score of 1, the average observational score was, on average, 3.6.

In this first state review of evaluations — launched last summer — the education department suggests some principals will need to be trained again on how to observe teachers. It’s one of numerous recommendations in a 45-page report that captures thousands of teacher and administrator responses to the evaluation program.

The State Board of Education will consider revisions July 27. Some require General Assembly approval.

A federal Race to the Top grant spurred Tennessee to create an evaluation system tied, in part, to student test scores. Every teacher is evaluated every year, receiving a score between 1 and 5. Teachers can be denied tenure, or lose it, if they score score 1s or 2s for two consecutive years. Some educators criticized the systemas being unfair, time-consuming and rushed into place, and they unsuccessfully pushed for the first year’s results to be considered a trial run.

Half of each evaluation is based on observations. The other half comes from standardized tests and other measures of student performance.

But almost two-thirds of instructors don’t teach subjects that show up on state standardized tests, so for those teachers — including in kindergarten through second grade, and in subjects like art and foreign languages — a score is applied based on the entire school’s learning gains, which the state calls its “value-added score.”

The report recommends relying less on the schoolwide scores, which many teachers fault for failing to capture their individual work. The state suggests bringing in other types of tests to measure these teachers.

That suggestion drew the most praise from Carol Schmoock, assistant executive director of the Tennessee Education Association, which has repeatedly advocated for fairness in the new evaluation system.

Schmoock also applauded a recommendation that principals be allowed to spend less time evaluating teachers who scored well and more time with teachers who need more training.

But Schmoock took issue with the suggestion that too many teachers received passing marks when observed by principals.

“That’s quite an indictment of teachers,” she said. “There’s a suggestion that the data from the (test) scores is superior to the data from the observations, and we’re not at all sure that is true.”

But state education department spokeswoman Kelli Gauthier said it’s reasonable to expect teachers who do well on observations to also post high value-added scores.

“If we don’t have the tough conversations and help teachers who may be struggling, they won’t get the training that they need,” she said.

Schmoock contested the report’s suggestion that Tennessee has more low-performing teachers than demonstrated in observations. “It just could be that all the processes we have in place of preparing teachers may be working,” she said.

The state is also pushing for ways to make sure districts across the state evaluate teachers consistently, although the report doesn’t say exactly how to do this beyond increasing training for evaluators.

During the school year, state officials who reviewed ongoing evaluations approached some districts to question when principals awarded a high percentage of scores of 4 and 5.

Williamson County Schools Director Mike Looney, for example, had to answer questions when county principals rated 97 percent of teachers a 3 or higher. Monday’s data wasn’t broken down by county.

The state had predicted the best districts would rate, at most, 85 percent of teachers at 3s or higher. No district that submitted midpoint data hit all of the state’s predicted ranges.

“I’m very confident that the observation scores and the evaluation scores are reflective of the hard work our teachers did here this year,” Looney said. He praised the state education department for gathering school district leaders’ opinions.

“On the surface, it appears the department has listened to some feedback,” he said.

The state is also looking for clarity regarding when officials can intervene in districts with a wide gap between value-added and observation scores.

Fifteen percent of a teacher’s overall score is derived by a measure of his or her choosing, with options including ACT scores and graduation rates. Teachers tended to choose the method that would most help their scores, the state found.

Changes possible

The report outlines numerous other changes, and anticipates what could be annual tweaks.

The first year drew feedback that included conversations with every school district superintendent, 7,500 conversations with teachers and 17,000 teacher and administrator surveys.

Educators wanted ways to streamline the evaluation process. Principals found their time consumed by class visits, with some responsible for as many as 36 teachers, but they may get a break.

High-scoring teachers may get the chance to undergo fewer observations and to choose to use their value-added scores for 100 percent of their overall scores.

Tennessee Education Commissioner Kevin Huffman has said the evaluation system would likely be reviewed annually. He was not available Monday.

“No one thinks that work is done,” Gauthier said. “But I think we’re moving in the right direction.”