that there exist individuals for whom a low test score is one of the few indicators of a low likelihood of success, and others for whom a high test score is one of the few indicators of a high likelihood of

success.

If the impact of tests on admissions processes is reduced, as I think it would be with mandatory disclosure, applicants for whom a high test score is one of the few positive indicators would be the losers, in favor of those who can present other positive indicators but who do not score well on tests.

But is this harmful? Yes, since the other indicators will be used regardless of the weight given tests, or whether or not they are used at all.

Thus, tests are providing additional information, and it is axiomatic that additional information cannot harm, that is, can only improve the performance of a selection process. Can it be wrong to allow all segments of society the full range of opportunities to demonstrate potential?

I should like now to examine several other provisions of H.R. 1662. In section 3(a)(3), relationships between test scores and career success would be required. Actually career success is not easily defined, let alone measured, and is affected by choice among subspecialties offered at postbaccalaureate institutions, which information is not available at the time of application.

Tests are intended to be predictive only of success in studying for a specific career and not of success in a career after studying for it. Therefore, no test within the range of tests specified in this bill in section 2(a)(3), namely "admissions and placement," attempts to predict career success. The only role played by the considered tests in a person's professional life is in selection for admission, and it is on this basis that they should be judged.

Section 3(a)(4)(A) would require it to be determined to what extent test scores improve prediction of grade point average over and above all other information used. But the indicators used by various institutions differ, and with varying success. It seems to me that it is sufficient that a test producer convince a given institution of the value of its product, and I have no objection to this information being made public.

However, justifying the use of any element in a selection process should be the responsibility of the institution using it, not of the supplier of some aspect of the procedure. No test producer can speak for the varied procedures used at all institutions, and surely it cannot be the intention of the bill to impose a mandatory, uniform selection process on all institutions.

In the same section, it is specified that the extent of improvement be expressed as a percentage. This is ambiguous and has been the source of some abuse of statistics. The basic question is: A percentage of what?

The most widely accepted measure of prediction accuracy is a squared correlation coefficient or some variation thereof. Assuming the use of multiple linear regression for prediction, there are two relevant square correlations: With and without the test as a predictor, the former being the larger. Test critics prefer to use the difference between these, which can be interpreted as a percentage. Test supporters tend to divide the difference by the square correla

tion without the test, which can also be interpreted as a percentage.

I would recommend the latter method, since it reflects the ability of the other indicators to predict grade point average as well as the value of the test, itself. As a minimum, more specific direction is needed. However, I am rather skeptical of the value of either of these indices for a public untrained in statistical methods.

Section 3(a)(4)(B) calls for an examination of the relationship between test score and income level. This invites interpretation of a casual relationship between these variables which is based on questionable assumptions. Moreover, such data are available now to competent researchers who are skilled in its interpretation.

Section 3(a)(4)(C) would require an analysis of the value of test preparation course. Since in most cases the tests we are discussing measure learned abilities, it is reasonable to expect that such programs can improve test scores, though available evidence indicates that short-term experiences do not do so appreciably. It seems to me that it is more the responsibility of these programs to advertise themselves than it is the responsibility of test producers to do so. Test preparation courses are the province of numerous private tutors, schools, and even books, very few of which report to testing agencies. Separation of the effects of these programs would be a monumental task, but necessary for meaningful interpretation of the data. I should also mention in passing that a test score average is not suitable as the base for a percentage, as called for in this section.

Section 4(a)(1)(A) would require making public studies about or using these tests. This seems to me to carry the potential of discouraging research unless test agencies were also required to provide test data under certain conditions. This area needs more thought, which should be undertaken before enacting a provision such as this one.

I also note that, while subject anonymity is guaranteed, institutional anonymity is not. This would also tend to discourage valuable research.

I would like to close by pointing out that I am not at all against scrutiny of the testing industry by the public. Actually, I feel the testing industry, as a whole, is turning in a creditable performance. But I find H.R. 1662 to be, in general, both harmful and misdirected. I would prefer to see, instead, inquiry into the uses of tests and other indicators of success, as well, made by institutions.

Many of the suggestions in the Association for Measurement and Evaluation in Guidance statement appended here are pointed in this direction.

Thank you.

Mr. WEISS. Thank you, Dr. Schafer.
Dr. Casteen?

STATEMENT OF JOHN T. CASTEEN III, DEAN OF ADMISSIONS, UNIVERSITY OF VIRGINIA

Mr. CASTEEN. I am John Casteen, the dean of admissions and associate professor of English at the University of Virginia.

I have come today as a representative of the National Association of College Admissions Counselors, with a mission of sharing with you a description of a research project and planned national forum on a publication having to do with college admissions tests that my association will undertake.

The prepared materials given to you include a prepared statement, which will be a somewhat longer version, as well as a copy of our proposal and, finally, an appended article to which I will refer to in a moment.

Mr. WEISS. Your entire statement will be entered into the record in full.

Mr. CASTEEN. I am a member of the American Personnel and Guidance Association, American Association of Collegiate Registrars and Admissions Office, a trustee to the College Board. The NACAC plan calls for three tests, and the defenses made with regard to it as it relates to the college process, the college admissions process, and not other aspects of testing.

We intend, first of all, after having secured proper financial support, probably from foundations, to commission a group of four investigative papers having to do with specific aspects of the controversy about college admissions tests. In my prepared materials I have described the likely contents of these papers as well as likely authors for them as a way of indicating the scope and the quality of investigation.

The titles proposed at the moment include:

What is the "Truth About Testing?", a paper that would explore many of the issues that have been referred to earlier by members of this panel, a paper having to do with admissions tests and the admissions process, trying to describe with some accuracy the number of different processes used to indicate the rating given, different credentials in the process and experiences students and institutions have had with the process.

Third, a paper having to do with alternative ways to match up students and colleges, a paper exploring specifically other kinds of admissions tests that might be used to provide the kind of ongoing and comparative data that we now get from standardized tests, as well as ways to match up students in a larger sense, to see to it that they do not incur the wasted experience that some students have when they begin.

Fourth, a paper that will embody some recommendations to your committee, entitled "College Admissions Tests and the Public Interest," trying to define on the basis of the first three papers what in large terms ought to be the Federal or congressional stake in college admissions tests, and what national policies perhaps including a document with legislation like that proposed today that might serve college admissions.

After having secured production of those papers, the National ACAC intends to sponsor a national forum on college admissions tests and the admissions process to invite representatives of the major professional and academic organizations that serve this area, representatives of the public, and we would hope to have representatives of your committee attend and take part in a forum that will deliberate the findings of the four papers, make further recommendations that we might sponsor concerning revisions to be made

to the papers and forward a document to be the NACAC report to the public in the form of a published book.

We envision having the papers completed by early fall and expect to convene the conference in late fall or early winter and publish the report as soon as possible thereafter with an eye toward providing a timely document for your use.

We want to achieve what we hope will be a balanced analysis of the facts.

In my prepared statement, I said we are apprehensive that the bulk of the material currently available to the public and to your committee is political and partisan in nature, and that the organizations have not devoted the kind of time, energy, money, and concern to the project that we think it deserves, and we intend to do that.

We want to inform the admissions profession about the controversy in admissions testing and to see to it that our organization has practitioners that have the best possible advice.

Third, we do want to pursue alternative ways of matching up students and colleges. We believe other methods are possible. We are conscious that in Great Britain there is a very different kind of admissions test, the discursive test that is represented by level A and O examinations serves the system, and that students progress to college in other parts of the world and, finally, we hope to offer reasoned recommendations to the public and to the Congress concerning the proper public stake that exists in this area of admission to college.

In concluding this overview of my prepared materials, I need to express some areas of personal concern that I recall having first expressed to your committee in 1979 when I came to you as a teacher and dean to express my personal concerns about legislation proposed to you at that time.

Education's problems are at this point quite large. We have problems of funding, serious problems of teacher morale and concerns about the quality of teacher preparation. Ample evidence that the curriculum currently existing in high schools is less than optimal in its service to our students, ample evidence that we are less successful than we ought to be in serving nonmainstream students with access to higher education.

No serious critic of the educational scene at this point blames either declining achievement throughout the system or the continuing inequity of opportunity and achievement that separates nonmainstream, especially racial minority students, from mainstream students or students from relatively poor economic backgrounds from relatively rich economic backgrounds or what appears to be evidence of the diminished overall effectiveness of education during the last 15 years or so, the kind of evidence that turns up in the report of the President's Commission on Foreign Language Study.

I am concerned to note that while your committee is deliberating standardized tests, no serious critic who has looked at these problems blames the test for these conditions. We do have initiatives within education to try to see to it that we face squarely some of the problems that have been evidence.

You have heard one described earlier by my colleague from Guilford County in attempting to decide what schools ought to do. I have attached a newspaper article describing the College Board's Project Equality, which is one of several efforts to address the question of what goes on in the academic courses that students take in high school, whether to prepare for college or life in general.

It seems to me Congress may or may not find sufficient cause to enact H.R. 1662 or analogous legislation, but that it ought to do so only in light of a broader examination of American schools and schooling than has yet occurred, and my concern in the largest sense is not so much the presence of this proposed bill or of earlier bills of the same kind, as it is the absence of initiatives undertaken by the Congress in the direction of solving other problems of which I suspect the testing controversy is merely one symptom.

It seems to me we need to look into the whole question of how schooling works in this country and to separate the core issues having to do with what students learn in schools from what I see as procedural issues having to do with how we measure and report and compare those results in student to student.

Thank you.

[The prepared statement of John Casteen follows:]