July 27, 2013 by juicefong
Ugh. Standardized testing. Such a double-edged sword. On the one hand, it gives us a faulty yet still meaningful measurement whereby we can assess school’s performance on teaching and learning. On the other hand, tests have been overemphasized by schools because they are so high-stakes. On the low end, teachers are doing test prep year-round. On the high end, great public schools feel handcuffed and have to dumb down their curriculum to optimize against tests that are actually a low bar for learning.
I’m not ready to remove standardized testing altogether, though. Even if you have a great school in middle class suburbs, you still want to make sure various subgroups are being served equally—across low-income students, English Language learners and all racial groups. I considered just doing random sampling of students within a school (say 25%), but you may not have enough of a sample size to measure these subgroups. You could test every other year or every three years instead of each year, grades 3-8 per No Child Left Behind. But then I worry that there may be too much slide year-to-year and the urgency would evaporate.
In talking to Dr. John Thompson yesterday (whose beliefs will appear in my upcoming “What’s Your Beef” series), a thought came to me: What if we tested students in the beginning of the school year, rather than the end? I could see this accomplishing a few things: 1) The test could be used much more like a diagnostic, and if teachers received an item-by-item analysis in the first month of school, that would help them decide how to spend the rest of the year with their students. 2) Test results could still be used as one way to assess the quality of schools. We’ll get to how this would work in a second. 3) By testing at the beginning of the year, schools would probably cut down on test prep, and might even do more to prevent summer learning loss. 4) This would blur the lines that directly tie teacher performance to high stakes test scores, in a productive way.
Tests as diagnostics, finally
The diagnostic element should be pretty clear. If students take a test in the first week of school, let’s revise the system so that test scores can be made available just a few weeks later. That’s both multiple choice and open-ended responses. Then let’s make sure teachers have a great analysis of strengths and weaknesses of all of their students by various standards or sets of skills, down to the specific question, to inform their teaching for the rest of the year. When I was a teacher in New York, we gave the tests in March and got the scores around May, but all we got was the scale score for each kid. We didn’t get a question-by-question analysis of how kids performed. Arm teachers with that kind of data in the fall and the test can truly be used as a diagnostic tool.
One new thought I’m adding in: because the sixth grade teacher gets her results that are mostly dependent on what happened in fifth grade and before, it might naturally lead to her having conversations with the fifth grade teachers and others. “I’m looking at the kids’ data, can you tell me what your curriculum was like last year? What do you think worked or didn’t work? What worked with this student?” When we test at the end of the year, the data for a cohort seems to rest squarely on one teacher, proliferating that “teacher as island” trap. Under this new model, it would encourage great collaboration between colleagues about their teaching and learning.
Tests can still play a part in measuring school performance
This data can still be used to softly assess school performance. We’d still test every child and I’d even be willing to do it every year, 3-8 per NCLB. We would still see trends in performance by various subgroups and we’d have a good picture overall at where all students stand as entering fifth graders or eighth graders. Of course you have transition years (e.g. a school that begins at sixth grade owes its sixth grade scores mostly to students’ previous school), so this gets a little tricky, but there will still be plenty of data to give you one snapshot of the school’s performance.
Dr. Thompson described to me how test scores in England are sent to an inspectorate, who every few years comes to the school and does a full assessment; presumably observing classrooms, interviewing teachers, students, parents and looking at achievement and other sets of data, such as teacher turnover or student attendance and suspensions. Gosh, that seems sensible. When I worked at a charter school in NYC, our authorizer did exactly this. A group of four people camped out with us for the better part of a week, observing classrooms and talking with various constituents and looking at data. The report they produced was full of helpful guidance—it wasn’t “gotcha” accountability. All schools deserve something like this: I could see standardized test scores contributing to that assessment and I could see value in doing an inspection every few years, as school quality does not typically change that much year-over-year.
Cut down on test prep and summer learning loss
This is a rather simple argument, but testing in the first or second week of school would hopefully cut down on test prep. Test prep seems to be most intense in schools in the months and weeks leading up to the high-stakes exams. Hopefully we’d see schools get away from that and instead offer great instruction year-round, knowing that good teaching, not test prep, is the best cause of students performing well on standardized tests. Currently, I see some of the best instruction in schools after testing period is over and the pressure is off. How ironic, right? By testing in the beginning of the school year, maybe we’d also see schools do more to prevent summer learning loss, and maybe summer school programs would actually be rigorous and helpful—most I know of in big urban districts seem to be lackluster.
Blur the straight line between teachers and test scores
Finally, and perhaps to the cheer of many teachers: by testing in the fall, we could blur the direct line between teachers and test scores and once and for all soften the notion that good teaching only means students doing well on standardized tests. If you’re a principal or anyone evaluating teachers and you’re relying solely on the results of one test to assess your teachers, then you’re just terribly mistaken. Teaching is more than what shows up on standardized tests. It’s about caring for and nurturing your kids, it’s about expanding their minds, it’s about helping them see a world they don’t know, it’s about inspiring them to greatness, it’s about filling them up with knowledge and ideas.
Don’t get me wrong, a good teacher should also see students do well on tests, but that is a low bar, and it’s just one bar. The trap we’ve found ourselves in with standardized testing is that it’s the only bar all of a sudden, because of how high-stakes they are and in the absence of less concrete and quantifiable ways to assess teachers. It’s time for districts and principals to put some real thought into how they evaluate their teachers more holistically.
Now, if I’m a fifth grade teacher, you better believe I want to know how every one of my students tested in the fall after they’ve left my classroom, and this data should be made available on the school level. But by just slightly blurring the connection between teacher performance and standardized test scores, we start to open the playing field to consider a variety of ways in which we assess teachers and perhaps we more strongly consider the notion that it is a holistic set of factors that influence student outcomes.
This is a new thought from the past 24 hours, so please send in your comments and poke holes in my argument.
Justin Fong (@jgfong) heads the small but mighty internal communications team at Teach For America. He writes freely about whatever he feels like, and of course does not expect his views to represent those of TFA or its leadership, ya know?