Fix Testing—Don't End It
“Trust me—there aren’t going to be any 7th grade English-language learners next year.”
“But there are three in the 6th grade this year… How do you know they will test out of ESL?”
How can I say, “Trust me”? It’s simple: I trust the test, and I knew my students were prepared. As many teachers rail against their subject’s state assessments, I count myself lucky to be an English as a second language teacher. Since its redesign two years ago, the test my students take, the New York State English as a Second Language Achievement Test (NYSESLAT), produces results that mirror my students’ proficiency on classroom tasks. And the data I receive from the exam is disaggregated by language usage (listening, speaking, reading, writing), so I can design classroom tasks that allow students to practice the skills they most need help on. Reliable results and relevant data make the NYSESLAT a test I can trust.
Many teachers have spoken out against testing and assessment, which is why I was glad to be able to address some of these concerns on a recent Center for American Progress panel. On the panel, we all agreed on the need for better, fewer, and fairer tests, but we also acknowledged the benefits of well-designed, well-administered assessments with readily-disseminated, disaggregated results. Recently, I worked with a team of teachers through the nonprofit teacher-advocacy organization Educators 4 Excellence New York to produce “None of the Above: A New Approach to Testing and Assessment,” a paper in which we recommend changes that strike a balance between those who demand an end to all testing and those who see the current system as infallible. There is a middle ground, and it is one in which standardized tests mirror the high-quality instruction students are receiving while providing one type of valuable data for teachers and schools.
Let’s first talk about the problem of teaching to the test. The solution is in one sense simple: If teachers could trust that the tests would accurately measure their students’ abilities, they may not feel pressured to “teach to the test.” Indeed, research shows that excessive test preparation does not actually result in higher test scores. In fact, one study found that students who were expressly prepared for the ACT did not perform as well as students who were given no test prep, just high quality instruction. To prepare for the NYSESLAT, I spend a total of 90 minutes (two class periods) looking at the format of the test with my students—that is the sum total of my explicit test preparation. I want them to be familiar with the types of questions they will answer, but I do very little coaching. I tell them, “Everything we do all year prepares you for this test. You don’t need anything extra.” The best part of this is that it’s the truth—they perform well and have come to trust the test themselves. When teachers and students trust the test, we can spend our time on rich, authentic teaching and learning that really matters.
Adapting to Learners
My ESL students also have to take the state English/language arts and math tests, and unfortunately the results from these exams are not always helpful—that’s why we need tests that can truly measure student progress. Many of my English-language learners read far below grade level. They continue to score at the lowest level (in New York, a 1 out of 4) year after year after—does that mean they haven’t grown as readers? A 6th grader who comes to me reading at a 2nd grade level and leaves at the end of the year reading at a 4th grade level is going to score “1”… again. And yet she has progressed two full grade levels in one school year! Computer-adaptive testing—as opposed to paper-pencil exams—would allow that growth to shine through.
Already in use for such exams as the GRE and GMAT, computer-adaptive testing adjusts to a student’s answers, changing the difficulty level of the questions accordingly, so that the result gives a true picture of where the student is and how much he grew over the course of a year. Computer-adaptive testing could provide teachers with more accurate pictures of student abilities, which would increase teacher support for state assessments.
Teaching and learning could be aided by standardized testing if the results of the tests were quickly released and fully disaggregated, so that teachers can use the assessments to inform our instruction. We do it every day with exit slips, unit tests, student conferences, class surveys… just not with state tests. The results arrive during the summer, when, for most teachers, our students have moved on to the next grade. We can’t even adjust our own teaching because the scores are calculated with a complicated metric and give little information about which skills the students mastered.
Imagine how much more rich and targeted a teacher’s instruction could be if she knew that her weak spot was in teaching students to draw conclusions or divide fractions. If teachers felt that the test scores gave us information we could use, there would be more teacher buy-in and less vitriol surrounding standardized testing.
The prospect of new exams designed by the two common-core assessment consortia, the Smarter Balance Assessment Consortium and the Partnership for Assessment of Readiness for College and Careers, is hopeful. Both groups’ assessments will be computer-based, which increases the likelihood of providing teachers, schools, and families with timely results. Both consortia appear to be testing higher-order thinking skills, which should reduce “teaching to the test” and promote high-quality instruction for all students.
Unfortunately, however, PARCC’s summative assessments will not be computer-adaptive, and SBAC’s will be adaptive only within the grade level, not across grade levels. In short, these new assessments are steps in the right direction but should continue to improve in order to best support teaching and learning.
I know—I’m lucky. The test that evaluates both my students and me is well-designed and reliable, producing relevant results I can actually apply in my classroom. I firmly believe that other teachers would come to trust their tests and use them as the valuable tools they should be if those tests were thoughtfully designed by fellow teachers, computer adaptive to show the full spectrum of performance, and produced results that were readily disseminated to schools, teachers, parents, and students.
Whether we want to acknowledge it or not, there are data showing correlations between standardized test scores and long-term life outcomes. Moreover, whether students want to get into college or pursue a variety of careers, they will likely face a battery of standardized tests throughout their lives.
Let’s give our students the best chance at lifelong success by administering high-quality assessments that mirror the high quality instruction we all strive for.