Excuses, Expectations, and Learning Gaps
One of the most significant advances in educational assessment over the past decade has been the development of the Tennessee Value-Added Assessment System (TVAAS) and its recent application by SAS inSchool as a general diagnostic tool. The SAS system takes a school district's year-to-year assessment data and measures the performance of the school system, different schools, and individual teachers--all at a very modest cost.
The fundamental idea underlying value-added assessment, developed by Dr. William L. Sanders and his colleagues at the University of Tennessee, is that all students can improve academically each year at the same rate as all other students. In other words, regardless of current academic achievement level--and regardless of race, sex, family income, and parent educational level--each student can gain a year's worth of learning in a year of schooling. Whether or not this happens depends largely on the quality of the student's teacher.
"Effective schooling trumps socioeconomic circumstances," declared Sanders in a September 20 address to Chicago United. "Schooling matters," he added. "It matters a lot."
Value-added assessment measures what fraction of each student's expected one-year learning gain is actually delivered by each teacher. The pattern of gains achieved by differently skilled students provides insight into how well a teacher's lessons are reaching different students--the high and the low performers as well as the average performers.
"The biggest factor affecting student achievement is teacher effectiveness," said Sanders, who emphasizes that class size effects and differences in ethnicity, family income, and urban-suburban location fade into insignificance when compared to teacher effects.
The way Sanders’ system works can be seen by applying the concept of value-added to data published by the National Center for Education Statistics in the most recent edition of its Digest of Education Statistics, 2001. Table 48 provides Spring and Fall test scores, sorted by child and family characteristics, from the Early Childhood Longitudinal Study for children in kindergarten and first grade.
The test scores reported for general knowledge and mathematics with children grouped by socioeconomic status are shown in the accompanying table, together with calculations by School Reform News to convert the Fall-to-Spring test score gains to percentages of the average gain.
The first question prompted by the raw data has to do with social promotion: Is it good educational practice to promote the lowest-performing students to the challenge of first grade work when their Spring kindergarten test score for general knowledge (21.55) is lower than the average test score for students entering kindergarten (22.67)?
Teaching to the Middle
When the relative achievement gains for each socioeconomic group are placed on a chart, the pattern of gains in general knowledge in kindergarten exhibit what Sanders calls a "tent" pattern, where gains are higher for average-performing students than for low- or high-performing students. This "tent" pattern is the result of a teacher's lesson being targeted to average students.
If repeated for several grades, the long-term result of this teaching pattern is that low-performing students fall further behind, and the achievement of once high-performing students regresses to the average.
Teaching to the Top
The pattern of gains in mathematics in kindergarten exhibits what Sanders calls a "reverse shed" pattern, where higher gains are achieved by previously higher-performing students. This is the result of a teacher's lesson being aimed at challenging the top-performing students.
If repeated for several grades, the long-term result of this teaching pattern is that low-performing students fall further behind both average and high-performing students. According to Sanders, this pattern is seen more frequently in suburban school systems.
Teaching to the Bottom
The pattern of gains in mathematics in first grade exhibit what Sanders calls a "shed" pattern, where higher gains are achieved by previously lower-performing students. This is the result of a teacher's lesson being aimed at bringing up the lowest-performing students.
If repeated for several grades, the long-term result of this teaching pattern is to raise the achievement of low-performing students at the expense of once high-performing students, whose achievement levels fall back to the average.
According to Sanders, this is a pattern often seen in urban and rural school systems. In those systems, this erosion in the performance of early high performers historically has been attributed to cultural influences. Value-added assessment shows educational practices also contribute significantly to the erosion.
"We cannot overlook these [previously high-scoring] children," urged Sanders. "We have to give them the opportunity to make progress from where they are."
The Ideal Pattern
According to Sanders, an ideal gain pattern would be a downward-sloping "shed" pattern with the previous top scorers achieving a gain of 100 percent, the previous bottom scorers achieving gains in the range of 110 to 130 percent, and average students achieving gains somewhere between those two extremes.
"This is how you close achievement gaps and do it the right way," said Sanders.
George A. Clowes is managing editor of School Reform News.
For more information ...
SAS inSchool is a division of the SAS Institute. Information on SAS inSchool's value-added assessment, together with pricing information and links to related articles, is available on the Internet at http://www.sasinschool.com/index.shtml
The longitudinal test score data are from Table 48 of the Digest of Education Statistics, 2001, published by the National Center for Education Statistics and available at http://nces.ed.gov/pubs2002/digest2001/tables/dt048.asp
Dr. William L. Sanders may be reached at email@example.com. An exclusive interview with Sanders, published in the November 1999 issue of School Reform News, is available on the Heartland Institute Web site at http://www.heartland.org/Article.cfm?artId=11119.