Study Shows Tests Can Be Linked To Make Comparisons
ALBUQUERQUE, N.M--Researchers said here last week that they have demonstrated a way to link different national and international tests to show how students in the United States over all, and those in individual states, compare with students from other countries.
In a study released during a national meeting on assessment sponsored by the Council of Chief State School Officers, the researchers said they were able to use statistical methods to link the results of the National Assessment of Educational Progress with those from an international mathematics test administered to 13-year-olds in 20 countries in 1991.
The study showed how students taking the international test performed according to the NAEP scale, which indicates the number of students who attain the "basic,'' "proficient,'' and "advanced'' levels of achievement.
"The whole country is now trying to find a way to link different tests,'' said Archie E. Lapointe, the executive director of the Center for the Assessment of Educational Progress at the Educational Testing Service, which conducted the federally funded study.
"This just demonstrates that, if you have two tests that are reasonably similar, that are administered in reasonably similar ways, you can be successful [in linking them],'' Mr. Lapointe said.
U.S. Fares Poorly
The results show, however, that, even on the new scale, American students fare just as poorly as they have on other international comparisons of math achievement.
The United States would rank14th among the 15 nations for which data are available, according to the researchers. Only 13-year-old students in Jordan were predicted to have scored lower.
South Korea was ranked first, with between 78 percent and a little more than 82 percent of that country's students predicted to have scored at or above the basic level on the NAEP scale, the researchers found.
The model also predicts that between 34 percent and 39 percent of Korean students would have scored at the proficient level or higher, and between 5 percent and 7 percent would have scored at or above the advanced level on the assessment.
In comparison, only between 55 percent and slightly more than 60 percent of American students scored at or above the basic level. The percentage of American students who ored at the advanced level or higher ranged between just under 1 percent and 1.7 percent.
"This is all very interesting, considering no one in Korea has actually taken the NAEP and couldn't do it anyway because of the language,'' said Peter J. Pashley, a researcher at the E.T.S.
The researchers' model also allows individual states participating in NAEP to compare themselves internationally.
It shows, for example, that student-achievement levels in Wyoming would be roughly similar to those for French 13-year-olds taking the NAEP tests. And North Dakota students--who performed at the top of national rankings in the first state-by-state comparisons of math achievement--would compare favorably with those in Korea.
Researchers cautioned, however, against reading too much into the results.
"This is strictly experimental,'' Mr. Pashley said. "We did find what we think is a reasonable and stable link between the two estimates.''
Opposition to Single Test
The need to find ways to link differing tests has become prominent in debates over developing a national assessment system. In the face of opposition to a single national test, researchers have sought ways to insure that various state and local tests could be linked to a common set of standards.
They have warned, though, that the degree to which results from different tests could be compared depends on the extent to which the tests measure similar content and were administered in similar ways.
In the first experiment of its kind, officials from Kentucky this
year compared the results of its statewide tests with those from NAEP.
(See Education Week, March 17, 1993.)