Campus News - October 15, 2001

Stevens taking a long look at school testing

By Carolyn Gonzales

Joe StevensJoe Stevens teaches statistical and research methods courses, but he can tell you that numbers don’t always tell the whole story. As an educational psychology professor he teaches graduate students “to master the tools they need to do research successfully.” And that means the organization and structure of data.

News abounds about accountability and school effectiveness. But, Stevens says, the way data is used to judge schools often fails to tell the real story. One reason is that the wrong tests and statistical methods are frequently used.

The most common method of rating a school’s performance is to assess students at one point in time. “What needs to be done is to look at the students longitudinally, or over the long term,” says Stevens. “We need to look at patterns of learning and behavior over time.”

Stevens has been working with the Albuquerque Public Schools, other New Mexico school districts as well as the state Department of Education on how to apply better statistical models including longitudinal models.

Stevens explains that, looking at only one point in time, students’ performance levels are most strongly related to where they start out socio-economically, culturally and linguistically in school. Using this one point in time or “status” approach, schools that are “low-performing” are usually those who serve students on the lower end of the economic spectrum and those that are “high-performing” are those who serve students on the higher end of the economic spectrum.

“To better understand how schools are working we need to look at not just what students’ status is but more importantly we need to look at students’ subsequent growth,” he says.

“If you look at growth, socio-economics are no longer the best predictor of student performance,” Stevens says. He also says that once schools are identified with low performance or low growth, those schools need to be offered both additional financial and teacher support.

Stevens presents his results at local and national conferences and in journal publications. He says that four or five states besides New Mexico are now looking at longitudinal statistical models to assess school effectiveness.

“Nationally, there is an almost desperate press for immediate results. But it is very important that we do this right, not fast,” Stevens says. “Careful consideration must be made to make sure these systems are fair and reliable.”

Testing is familiar territory for Stevens. He worked for the Educational Testing Service, the folks who put together the SAT, National Assessment of Educational Progress (NAEP) and the Advanced Placement tests. The psychology of assessment is also part of his background as well.

“I earned my undergraduate, master’s and PhD in psychology, all at the University of Arizona in Tucson,” he says.

The University of New Mexico
Albuquerque, New Mexico USA
Copyright ©1998 The University of New Mexico.
Comments to: paaffair@unm.edu
Campus News
Public Affairs Department
Hodgin Hall, 2nd floor
Albuquerque, NM 87131-0011
Telephone: (505) 277-5813, Fax: (505) 277-1981