Sunday, December 2, 2007

Reforming Assessment Practices: Short List

Assessment can be a wonderful tool for gauging how well students are learning key material and how well teachers and schools are fostering learning. But, like any tool, assessment must be handled properly to achieve any benefit. Using a hammer to drive a screw only destroys the screw and damages the wood. The current educational policy climate in the United States has driven the growth of educational assessment, which has the potential to foster powerful cultures of data awareness in order to target interventions for the improvement of an ailing system. Unfortunately, lack of technical capacity and a touch of shortsightedness in places has led to a number of unhelpful, and even at times unintentionally damaging, policies. Below I give my short list of recommendations for purging the dross among assessment practices to allow the beneficial qualities of assessment shine through.

1) Some localities have responded to federal and state assessment requirements by entering a hyper-testing frenzy, in which even a month of instruction over the course of the year can be lost to large-scale, standardized assessment. This is overkill and hinders the primary goal of teaching. Classroom teachers certainly need to be assessing their students continually, formally and informally, in ways that they have been for years, in order to identify students' learning needs and adjust instruction. Likewise, large-scale, standardized assessments also provide a necessary external check of student learning, which can form a piece of evaluations of teacher and school effectiveness and draw out macro-patterns in the educational system as a whole. However, large-scale assessments shouldn't take the place of "light-weight" classroom assessments, nor should they displace significant portions of that which they are attempting to measure, viz instruction. Instead, states should require just a few days of testing both at the beginning and again at the end of each year, leaving the remaining time for teachers to perform their m├ętier.

2) I have heard it claimed that some states and localities have been creating standardized assessments before establishing a strong statement of what they expect students to learn. This is utterly backwards and is largely a waste of time for everyone involved: A number resulting from a measurement of an unknown entity yields very little useful information, and asking teachers and students to perform on what purports to be an evaluation of their progress during the year without first telling them what will be expected is not only unfair, but creates an undue psychological burden on teachers and students alike that can only degrade the learning environment. Instead, any testing authority operating without a clear reference curriculum or with vague or flimsy standards should as soon as possible work out the skills and knowledge they hope students to master (ideally developing the list through a participatory process), publish the standards as widely as possible with accompanying professional development for teachers and local administrators to understand them as well as possible, and then ensure that their assessment is properly aligned with their expectations.

3) No teacher should spend more time teaching test-taking strategies than teaching content. While students do need to have some familiarity with the format of the test (so they spend their time demonstrating their skills instead of decoding the directions), teaching only test-taking skills not only does a huge disservice to the students (after all, one supposes that the students actually need to acquire the knowledge and skills being tested) but violates the spirit and intent of the measurement. For most teachers this is a no-brainer, but there are still
a few rogue teachers out there who, perhaps due to social pressures or out of fear of negative consequences, sacrifice their primary professional responsibility for a better image. These teachers need to be identified for professional development and potentially eventual dismissal if they consistently fail to demonstrate a capacity for improvement.

4) Testing authorities should endeavor to return detail-rich test results to schools and teachers as timely as possible. The primary purpose of large-scale, standardized assessments is to provide an external check on student learning to identify large-scale problems in order to target interventions in the system as a whole (or parts of the system, such as individual districts or schools or sub-populations). Nevertheless, one should not waste the opportunity to provide information that individual teachers could use to target local interventions for individual students. Assessment yields knowledge about students' learning to date--where the competencies are strong, and where they are weaker. Teachers could profit immensely from this knowledge in order to provide targeted instruction to help students improve in their weaker areas more efficiently. This information would be especially helpful if it came in the Fall--to inform instruction through the coming year--and in the late Spring--to help identify students who may need extra help during the summer before beginning the next grade.

5) In the upcoming reauthorization of the Elementary and Secondary Education Act (No Child Left Behind), the federal government should move away from requirements that (a) compare one cohort of students with the next, and (b) use the percentage of students achieving a passing score to determine progress. The former only adds extra variance to the scores and muddies their interpretability, and the latter yields a measure of progress that isn't comparable between groups. Percentage passing is certainly important, since passing indicates achieving one of the goals of education (namely the mastery of certain skills and knowledge deemed important for students' lives), but the percentage passing isn't suitable for gauging progress since the same level of improvement in absolute scores could yield a large increase in the percentage passing if the cohort's average score is close to the cutoff or a small increase if the cohort's average score is further away. Instead, progress should be measured in terms of growth models comparing students' abilities at the beginning and the end of the year. These models provide a wealth of information about not only student's absolute level of mastery but their rate with a given teacher in a particular school. Under a highly skilled teacher, students who are very far beyond the state's expectations can make phenomenal progress, many times faster than students who are "on track," exhibiting one and a half or more years worth of gain relative to state expectations. These students may still take several years to catch up to their on-track peers, but examining raw passing rates fails to acknowledge their fantastic progress in the interim and masks their teachers' true skill. Looking instead at growth rates within the year shows whether students are likely to make sufficient progress to reach proficiency by the end of their schooling and gives a much better basis for comparing the effectiveness of teachers and the improvement of schools.

No comments: