The Stability of the Sat‐verbal Score Scale

In his 1961 report to the College Board, Wilks recommended that “The College Board should conduct a solid research effort to make an assessment of the extent of drift which may have occurred in the SAT‐Verbal scale since 1941.” In the present study, which is based on a design suggested by Angoff, raw scores on each of four old forms of SAT‐Verbal were placed on the reporting scale for a recent form so that the resulting scaled scores could be compared with the scaled scores derived when the old form was first administered in the College Board Admissions Testing Program. The data on the recent form were collected as part of the December 1963 administration of the Scholastic Aptitude Test; data on the old forms were retrieved from files for the April 1944, January 1948, March 1953, and February 1957 SAT administrations. For each of these forms, the scale derived by equating scores on the old form to scores on the December 1963 form was compared with the scale used for reporting scores on the old form. The results may be summarized as follows: For both the 1953 form and the 1957 form, the experimentally derived scales and the reporting scales differed by less than five points in the vicinity of the mean. The degree of agreement was good enough to warrant the conclusion that the SAT‐Verbal scale remained highly stable during the period from 1953 through 1963. With the reservation that the standard errors of both the operational and the experimentally derived scales for the January 1948 form probably are relatively large, it may be concluded that the SAT‐Verbal score of a candidate tested in 1963 and scoring in the vicinity of 500 might be 20‐35 points higher than the score of a candidate of equal ability who was tested in 1948. Since the scale as it existed in 1953 was maintained through 1963, it may be inferred that the shift in the meaning of the scale took place between 1948 and 1953. Scores on SAT‐Verbal reported in April 1944 may not have had the same meaning as scores reported in December 1963, but the methodological limitations inherent in the experimental equating of those two forms preclude any reliable quantitative assessment of whatever differences may exist between the scale of the early 1940's and the scale of the early 1960's.