The SAT.

I was one of the two psychometricians on the panel that advised the College Board on the issue of whether the SAT scores of disabled students who take the test with accommodations should be "flagged" (see Miriam Kurtzig Freedman, "Disabling the SAT," Feature, Fall 2003). In my opinion, the decision to end the practice of flagging was the right one.

Standard testing conditions are problematic for many people with disabilities. This is why the College Board grants testing accommodations, such as extended time to finish the test. However, many people with disabilities do not want others to know about their disability. The practice of flagging SAT scores essentially informs college-admissions officers that the applicant has a disability. The opportunity for bias against the student is obvious.

There are two arguments in favor of flagging. The first is that providing an accommodation may give students with disabilities an unfair advantage. The second is that scores from accommodated test administrations are less valid than scores from standard administrations.

As Freedman argues, the current version of the Standards for Educational

and Psychological Testing can be interpreted to support the practice of flagging when information on the comparability of scores across standard and nonstandard test administrations is lacking. However, these standards also state: "If a modification is provided for which there is no reasonable basis for believing that the modification would affect score comparability, there is no need for a flag." Furthermore, the Standards are silent on the issue of where to draw the line between "comparability" and "noncomparability"

The College Board has sponsored more research on the effects of testing accommodations than any other organization in the world. The findings for the SAT show that students with disabilities perform much better when given accommodations such as extended time. Meanwhile, students without disabilities score only a little better when given an accommodation. The only evidence in support of "noncomparability" was for male students with learning disabilities who had extended time. Their first-year college GPAs were lower than the SAT predicted. However, this finding did not hold for females with learning disabilities, and it was reduced when males' high-school grades were factored in. When evaluating the size of this "overprediction" for learning-disabled males, the majority panel noted that it was smaller than the overprediction for other groups of students, such as African-American males. The practice of flagging could not be supported by the relatively small effect noted just for learning-disabled males.

STEPHEN G. SIRECI

University of Massachusetts 

Amherst, Massachusetts 

Miriam Kurtzig Freedman responds: I thank Stephen Sireci for his thoughtful letter. Nevertheless, I am not prepared to assume that college-admissions officers will be biased against applicants with disabilities. Neither the panel that voted to end flagging nor the College Board has presented any evidence to support this assumption.

Sireci writes that the only evidence of noncomparability is the "relatively small effect noted just for learning-disabled males" and concludes that this could not support the continued practice of flagging. Yet when the panel's two psychometricians asked themselves the key question--Are the scores from standard and nonstandard administrations of the test comparable?--one answered, "No," and Sireci answered, "Not sure." We still have no "reasonable basis" to believe the scores are comparable. Thus the College Board's decision to end flagging is difficult to reconcile with the evidence and the panel's own deliberations.