Abstract: | Educators have need for a procedure to generate alternate forms of tests. The reliability of alternate forms generated from a table of specifications is examined, using 78 high school remedial mathematics students as subjects. Ten forms of a test were constructed and administered; seven of these forms were readministered. Alternate forms correlation, .85, is as high as the test-retest correlation, .82, lending support to the hypothesis that alternate forms generated from a table of specifications are reliable. Discussion includes educational uses for a table of specifications in text books to generate test forms. |