Glenda Russell didn't need someone to tell her how to be a teacher. She'd been doing it for two decades in a small elementary school in the southern US state of Tennessee. But, like many teachers in the US, she suddenly found herself under scrutiny from state legislators who demanded new assessments be put in place as part of a plan to increase education spending. Politicians reasoned that no amount of money would improve student performance if the teachers didn't know how to do their job.
Almost unnoticed, in 1992 Tennessee adopted an educational assessment scheme that would turn out to be revolutionary - and one which would take years for its significance to be truly understood. While other states were judging teachers based on their students' performance in comparison with students of the same age in other schools, Tennessee was following the same individual students and groups of students throughout their educational careers. By charting where the students' test scores rose or fell, they could pinpoint which teachers were effective and which were not.
"I hated it," says Ms Russell. "I thought it was the most horrible thing that was put on teachers." Worse still, the testing quickly found that 87 per cent of the 11-year-old students in her class had failed to master punctuation, basic maths, and other subjects required at their age.
"When I saw these terrible scores, I was so embarrassed," she says. "I had been teaching for 20 years. I had been a mentor for new teachers. I was supposed to be good at this." The low marks of the students in her school were published in the local newspaper, and while no teachers were identified by name, it was easy to infer that one was Ms Russell, since there was only one other teacher in her grade.
But then she looked more closely at the data. "When I sat down and analysed what I was doing, I realised that I was not reviewing," she says. "If you don't review and review, you don't get through. When I looked at my scores, I realised my students were not retaining what I was teaching."
So Glenda Russell came up with a structured review programme. Soon the students in her class at South Clinton elementary school rose to the top 5 per cent nationally in subjects they had failed to master in the past.
"Are all teachers enthusiastic? Certainly not," says Ms Russell, now an avid believer in the assessment system. "Some of them are hurt and disappointed. But I have seen tremendous improvement."
The self-contained nature of the Tennessee value-added assessment system, by which students are judged against themselves and not against their counterparts in other schools, is not the only reason it's unique. The most peculiar thing about it is that it was devised by a professor of statistics who specialises in agricultural research and who has no background in educational evaluation.
William Sanders, a 58-year-old professor at the University of Tennessee school of agriculture, had looked on with casual interest as lawmakers tried to come up with the best means of assessing teachers while screening out students' social and economic backgrounds. Then he remembered a methodology developed in the 1950s to track the outcomes of animal breeding. It allowed large amounts of data to be broken down while partitioning genetic factors from environmental influences.
"I thought to myself, 'Golly, the problem is the same'," says Dr Sanders. In this case, it was students' educational influences that would have to be separated from environmental influences. "What I argued, coming from a different tradition was, 'Let's look at the kid as the blocking factor, to use the statistical term. Let's let the kid be his own control, because he has the same innate ability this year as he had last year, comes from the same home environment, etc. That way, we can filter out those other things'."
The "other things" include wealth, parental education, and other variables known to affect student scores. Under Dr Sanders's methodology, students are compared only with themselves from one year to another. The degree to which they improve is quantified as the "value added" by the teacher. The consequences are not just academic. High-scoring schools receive additional state funding; districts that show little or no improvement are subject to sanctions including continuing state scrutiny.
"If a kid catches a sequence of ineffective teachers, I don't care how innately bright he is, that kid is going to suffer," says Dr Sanders.
Every spring, students in grades 3 to 8 (nine to 14-year-olds) take a test called the Tennessee comprehensive assessment program, which covers reading, language, maths, science, and social studies. The scores are compared with the same students' results from previous years.
"The problem is that traditional statistical approaches require you to have complete information about the same kids over time. But kids move, kids get sick, kids miss tests. That was always a stumbling block, and the methodology we brought to the table enables one to do this longitudinal analysis using this data no matter how incomplete."
Dr Sanders's approach has been criticised by other statisticians. Teachers, understandably, also resent being judged using methodology developed to evaluate farm animals, and say it ignores the impact of learning that occurs outside school.
So far Dr Sanders's system is being used only in his home state. But a few school districts outside Tennessee have also signed on, and now Florida is considering adopting the idea. Dr Sanders will leave the university this summer, with his wife and four other members of his research team, to join a private firm that is expected to market the concept aggressively.
Teacher assessment is a hot topic in the US. Opinion polls show that, by more than six to one, people think that teacher quality is the most important element in educational reform. President Clinton has called for teachers who "aren't competent" to be weeded out, and has threatened to cut off federal funds to schools that hire unqualified teachers. The education secretary, Richard Riley, has said "a growing number of school districts are throwing a warm body into a classroom, closing the door, and hoping for the best".
Some politicians have proposed testing teachers, and making the results public. Dr Sanders disagrees, and has advocated keeping assessment scores private. "I really believe that you get the most benefit from having individual teachers and principals confront their own data," he says. "When teachers and principals begin to see that there's a certain portion of the kids who are not making progress, they will deal with it without pressure or force."
That's what happened in the Maryville middle school, according to its principal, Joel Giffin, another one-time sceptic who is now an advocate of Dr Sanders's methods. "To be honest, any time you talk about evaluation or accountability, people get nervous," says Mr Giffin. "That's because we've had a system for so long that never really had the accountability it should have had. But this is real data that gives you a chance to make all kinds of changes."
When the scores come in, Mr Giffin meets teachers individually to "pat them on the back and tell them how good a job they're doing. After that, if somebody's got a weak area, then we start talking about how to do better."
Teachers who do poorly in certain measures are paired with teachers who do well; they even sit in on each other's classes. "Any area you're weak in, I can find you somebody that's very strong in that," says Mr Giffin. Unlike assessments that compare teachers, Tennessee's "does not force teachers to compete against each other. They're competing against a benchmark. I reassure them that this is not something they have to fear."
Last year the data helped identify 20 students who were not advancing in maths. On closer inspection, Mr Giffin found that most came from low-income, single-parent homes. He provided them with extra help with their homework. In the next tests, their maths scores had quadrupled.
"This does not tell anybody what to do," says Dr Sanders, "but it certainly begins to inform the individual teacher and the principal which pupils might need more attention."
Since the system began to be used, according to Dr Sanders's research, 45 per cent of the state's 137 school districts have shown measurable progress; most of the rest have held steady. "I don't want to paint this rosier than it is," he says. "You can still go to places where people want to buy me a one-way ticket to the middle of the Atlantic."
Now Dr Sanders plans to expand his work by looking at teachers' training to determine what makes them high-achieving or ineffective. Using the 6 million student records and evaluations of more than 30,000 teachers he has collected, he has already found that teachers improve their effectiveness steadily year after year during the first 10 or 12 years of their career. Then they reach a plateau that lasts about 10 years, before beginning to decline.
For Glenda Russell, learning that her students weren't advancing in her classroom helped her correct a problem she says she would not otherwise have known about. "If it was in the newspaper that you were doing a poor job, and thenyou had this improvement," she says, "you'd be motivated, too."
* What makes an effective teacher? Analysis of the Hay McBer Report: TES main section pages 21-25