Not All Alike

Abstract

THE HIGH-STAKES testing movement has reinforced a tendency to think of students in dichotomous categories: passers and failures. Part of this derives the capriciousness of high-stakes testing, in which a single right answer can shift children one category to another. For instance, when Virginia changed its equating procedure, the new procedure required one less correct answer for a passing score (something the state has never explained). The change turned 5,625 failures into passers, although the students in question didn't know this for some months. Marsha Riddle Buly of Western Washington University and Sheila Valencia of the University of Washington decided to look beyond this categorical divide. Their research appears in the Fall 2002 issue of Educational Evaluation and Policy Analysis. Their concern, in part, was that all students who fail a high-stakes test tend to get the same treatment, as if all had the same problems. More disturbing is that some of the policies that states have developed are based on assumptions that are, at best, untested. At worst the policies are founded on misguided assumptions, misunderstandings, or skewed data sources. In reading, the instruction of children who fail high-stakes tests in the third or fourth grade (the most common elementary grades tested) assumes that the students have failed to learn some skills that were taught earlier. In particular, the instruction tends to emphasize phonics. Such a focus might occur in Washington more than in some other states because Washington legislators, like those in California, have pushed phonics. In a Seattle Times op-ed titled Can't Read Because Johnny Needs Phonics, Harold Hochstatter, chairman of the Senate Education Committee, declared, Anyone who claims to teach reading by some method other than phonics isn't teaching reading at all, but is offering some weak substitute for it. Hochstatter trumpeted, READING IS PHONICS! and WRITING IS PHONICS! (capitals in the original). In one 18,000-student district, Buly and Valencia looked at skill patterns among fourth-graders who had scored at level 1 or 2 out of a possible 4 on the Washington Assessment of Student Learning (WASL). The district was diverse, with 43% white, 19% Asian, 11% African American, 11% Hispanic, and 3% Native American students. The researchers used WASL results along with a battery of five individually administered reading assessments so that got a variety of indicators of word-identification and word-attack skills, phonemic awareness, comprehension, fluency, reading rate, and vocabulary. Looking at the averages for the whole group, determined that from average scores, we might conclude that students had similar needs. However, as the cluster analysis reveals, [student needs] are quite different. Among the clusters found: * Automatic word (18%). Some were English-language learners, some were native English speakers, and most were poor. The researchers affixed the label automatic word callers because they can read the words quickly and accurately but fail to read for meaning. * Slow word (17%). These students are accurate in word identification, especially for real or nonsense words in isolation or in context, but are slow. Their slowness cannot be attributed to weak decoding skills. * Struggling word (15%). While better in word identification than in reading for comprehension, have some difficulty with word- attack and word-identification skills. * Word stumblers (18%). Consisting primarily of native speakers who are not poor, these students read well for comprehension, but their word- attack and word-identification skills are so weak that have great difficulty reading. Consider the profile for Sandy, a member of this cluster: Sandy stumbled on so many words initially that she could not be expected to successfully read and comprehend. …