With the nation spending about $5.2 billion annually on education technology, how and if computers contribute to improved student performance is a crucial question. A new study by Educational Testing Service reveals that how computers are employed determines their efficacy.

Higher scores are obtained when they are used for math simulations, learning games and software applications that ask students to address real-life situations. But more popular applications known as drill-and-practice are linked to lower test scores.

The study analyzed fourth- and eighth-grade math scores from the 1996 National Assessment of Educational Progress.

• Among fourth-graders, students whose teachers used computers mainly for "math/learning games" scored up to 15 percent of a grade level higher than other students.

• Eighth-graders whose teachers used computers mainly for drill-and-practice performed nearly half a grade level worse than other students.

• Students who spent more time on computers in school performed slightly worse in math than those who spent less time on them.

• More than half of black eighth-graders had teachers who used computers mostly for drill-and-practice -- compared with only 30 percent of white students.

Three out of every four U.S. public school classrooms have at least one computer designated for instructional use.