Under the state's new teacher evaluation system, observations by principals make up half of their scores, but a first glimpse at those observation scores shows they are all over the map.

An open records request from The Tennessean found that in Murfreesboro City Schools, for example, nearly half of the teachers were given a top score of five, while in Fayette County, only one percent earned the top rating (http://tnne.ws/z2Soie).

At the bottom end of the scale, only one Williamson County teacher received the lowest score, while at Humbolt City Schools 7.6 percent of teachers earned a one. And in at least 16 other school districts, no teachers received the bottom score.

The state had predicted that districts would rate 3-5 percent of teachers as ones; 10-25 percent as twos; 40-50 percent as threes; 10-25 percent as fours and 5-10 percent as fives. No district that submitted data hit all those ranges.

Some critics of the new evaluation system say the differences between the districts show that the system is too arbitrary.

"I question how evaluators are evaluating if the scores vary greatly across the state," said Marshall Winkler, wellness teacher at Franklin High School. "I feel like the (state) jumped into this new plan too soon."

But state officials say it's too early to draw conclusions.

"(The data) is inconsistent with what the research would project, but we are midyear on this," said Emily Barton, the state's assistant commissioner for curriculum and instruction. Barton said they will know at the end of the year whether the observation scores line up with student achievement measures. If not, districts could penalize principals by taking 10 percent off their own evaluations.

Williamson County Schools Director Mike Looney filed his own open records request for the ratings data after state officials questioned why Williamson principals rated 97 percent of t eachers a three or higher. Looney said the scores are justified because his county has good teachers.

"To come to some conclusion that our scores are too high ... is preposterous," Looney said. "We are not going to feel compelled or pushed into making our teachers fit some bell curve."

Murfreesboro City Schools also got a visit from state officials after issuing more fives on teacher observations than any other district in the state but one - Clinton City Schools.

"When first seeing the score distributions, I questioned why they were different than the predicted distribution," Murfreesboro City Schools Director Linda Gilbert wrote in an email. "I have made the principals aware of the state average and expected distribution ... but I have not asked them to change what they are doing."

Gilbert said the scores are justified by high student achievement in the district.

The evaluations won't affect teacher tenure this year, Barton said, because it tak es five years for consideration and teachers can't lose tenure unless they have two years of low evaluation scores.

Under the new evaluation system, 35 percent of a teacher's final score is based on student learning gains and 15 percent on data the school chooses, such as ACT scores. Principals use a long list of measures for success during their observations, which count for the other 50 percent.

The ratings data is based on 47,000 teacher observations between August and Dec. 13. Some districts did not submit enough observations to be included in the data. Districts using their own evaluation models, such as Memphis and Hamilton County, are not included either.