Skip to main content
Ed. Magazine

Taking the Long View: Tracking Changes in Achievement

[caption id="attachment_8578" align="alignright" width="185" caption="Conant Professor Judith Singer and Eliot Professor John Willett (photo: Kris Snibbe/Harvard News Office)"][/caption] Imagine a student named Jack. In kindergarten, he is the kid itching for recess, the lone voice chattering through quiet time, the boy who is always in “time out.” The next years are no better... He is falling behind... He acts out in class... He ignores directions... The principal’s office has a seat with his name on it. But the following September, things begin to change. Jack starts to listen to his teachers. His hands grab for pencils instead of ponytails. Books no longer pose as enemies. Something, indeed, has clicked.

“New advances enable us to accurately monitor, map, and assess human progress,” says John Willett.
Today, statisticians or social scientists looking at Jack—or a class full of Jacks, or a country full of them—possess the tools to tell his success story. They can study his progress over long periods of time, examine the causes of his initial setbacks—developmental delays, a troublesome home life, or, perhaps, poor schools or teachers—and determine the reasons for his eventual triumph. For most of the last century, research has examined singular moments and discrete events, often truncating data collection before a full picture came into focus. But, in the last two decades, science has made great strides in quantitative analysis, developing new methods to measure change over time. “In life, everything that is truly important is longitudinal,” says John Willett, the Eliot Professor of Education at HGSE. “New advances enable us to accurately monitor, map, and assess human progress.” Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence, an award-winning new book coauthored by Willett and Judith Singer, the School’s Conant Professor of Education and academic dean, offers the first accessible, in-depth presentation of the field’s most recent advances. Focusing on two complementary longitudinal methods—individual growth modeling and survival analysis—Willett and Singer provide new avenues for researchers to address critical questions about how things evolve and how long it takes for change to occur. “These are truly the most central questions that anyone can ask of the human condition,” says Singer.
Advances achieved in one arena often find effective applications in another.
She explains that such queries are essential to all areas of substantive scholarship, from education, to economics, to biology, and beyond. In fact, the two statistical methods in the book originated not in the behavioral sciences but in the medical and biological fields, where they were initially developed to track physical growth and the survival of patients with terminal diseases. Applying Advances across Disciplines But nothing in science occurs in isolation, Singer continues. Advances achieved in one arena often find effective applications in another. In the field of education, for example, these new statistical methods have already led to more reasonable answers to questions, including: What affects children’s ability to read? When and why do teachers leave their profession? What is the impact of a GED on the earning potential of a high-school dropout? “These methods tell us not only if, when, and why shifts occur,” Singer says. “They show us the big picture of risk over time: which groups of people are most susceptible to an event, why that might be so, and at what point in their lives the event might occur.” In today’s age of educational accountability and assessment, these new statistical methods will play a crucial role in analyzing academic achievement. Children reach developmental milestones at different times; as in the case of our student, Jack, some arrive after great delays. Yet policies like the No Child Left Behind Act increasingly tie students’ and schools’ successes to a single year’s standardized test scores. Willett and Singer worry that this method for measuring school and student performance reflects not how efficiently individual students learn, but what different groups of children know at a particular moment in time. To make accurate decisions about the quality of teaching and the impact of school programs, Willett and Singer urge policymakers to examine not what different classes of children know at each grade level, but how the knowledge and skills of specific groups change over time. “If you track individual cohorts of children, you will see actual changes in achievement,” says Willett. “In the end, that’s the definition of learning-change in achievement. That is what schools should be held accountable for.” About the Article A version of this article originally appeared in the Spring 2004 issue of Ed., the magazine of the Harvard Graduate School of Education. For More Information More information about John Willett and Judith Singer is available in the Faculty Profiles.

Ed. Magazine

The magazine of the Harvard Graduate School of Education

Related Articles