Do You Understand My School? (Part Two)
Yesterday, I posted “Do You Understand My School (Part One)” and argued that policy makers place unwarranted faith in state test scores as a tool for teacher evaluation. In large part, I believe that happens because they don’t sufficiently understand how schools operate. The argument continues:
To further complicate matters for the test advocates, they have no formula to account for the students who change classes, teachers, or schools during the year. You can’t just weight the test score according to the percentage of instructional time spent in a class, since classes do not cover the same content and skills all year long. That’s not a trivial matter affecting a small number of students. In some school communities, student transiency is a huge factor; additionally, in some schools, students can drop a class, or be removed from it for disciplinary reasons, but the student still must take the state test in that subject. In some schools, semester long courses mean teacher changes for an entire student body at the middle of the year. How do you use state tests to evaluate a teacher’s performance across semesters when that test is given before the completion of the second semester, but months after the first?
If my school expects me to take on a different teaching assignment, new curriculum, or new methodology next year, am I allowed a one-year grace period on producing test results? After all, the first year of teaching a new curriculum presents some unique challenges during a period of adjustment. Or, should I fight any change to my duties, in order to protect my livelihood? If I’m deemed effective as a teacher of juniors but my principal asks me to teach freshmen, that change might involve some adjustment, or I just might not be as good with one age group as I am with another. Why take a chance? If I teach seniors, they don’t even take a state test, depriving me of a chance to (supposedly) prove my effectiveness.
If I teach courses for two grade levels, will I be evaluated with student test results separated by grade levels? If so, we should be concerned about sample size – how much weight is given to each single student’s results – and the likely volatility that results from small samples. If not, it seems likely an average might mask evidence of strengths and weaknesses. (By critiquing both approaches, my point is not that teachers should therefore ignore student growth; we need multiple and more robust measures of student growth – a topic I will return to in later blogs, and which will be addressed in an upcoming policy report by Accomplished California Teachers). Further complicating the matter of grade levels, how do you measure effectiveness when students in the same class are in different grades? Is it reasonable to expect the same growth from freshmen that you expect from sophomores, or even juniors, who might all be enrolled in the same math, language, or arts class?
An additional problem too often overlooked is that students are not randomly distributed. I might teach the exact same course at the same grade level as one of my colleagues, but differences in the students’ schedules and activities will affect the make-up of the class. If computer science classes are only offered at certain times of the day, you increase the odds of grouping those students in the same math classrooms, while the same math course offered at other times, with another teacher, will have a low concentration of students currently enrolled in computer science. If my English classes are mostly in the morning, my students will be present for more instructional time, while my afternoon classes are thinned out when athletes are excused for competitions. Then, at every school you will find differences in the ways students with special needs are placed in mainstream classes, further complicating efforts to compare results.
What about extra support? If one-third of my students have a study skills class to help them keep up, and that class is effective, would you expect to see performance results on state tests? And if so, what does that have to do with my teaching? Consider also that some students are repeating a class, though we have no means of figuring out precisely which content and skills they picked up in each time they attempted the course. What about students being tutored, or those who have better academic support at home? We have no means of tracking that information reliably, no way to measure the tutoring effect, and again, no method to ensure that well-supported and under-supported students are distributed randomly among teachers.
Just to be safe, maybe I’ll start offering more personal assistance after school. That would help my students achieve better results. Students may be more likely to come in for extra help if they come with friends, but I’d better not help the friends if they are not in my class, lest I make myself appear less effective relative to their teachers. I certainly shouldn’t promote the idea of having a team of teachers operate an after school program, because well-distributed results do little to enhance my evaluation in a competitive atmosphere, or increase my job security when layoffs are looming, and seniority no longer matters (another currently popular goal of education reformers).
Education reformers, especially those who want to weaken teacher unions, like to tell us that schools should be run according to the needs of the students – not the teachers. I agree. So, why do they argue for policies that encourage teachers to neglect the best interests of students and put self-preservation first?
It’s because they don’t understand schools.
They might say I’m indulging in hypotheticals, or presenting exceptions as if they are the rule. That would be further evidence of their misunderstanding. Here’s the thing: exceptions – collectively – are the rule.