spelled out in detail in Campbell and Stanley (1963). Yet it continues to plague researchers and readers of research in education and related fields. Major, funded research, supposedly selected and monitored, is no guarantee that matching of individuals from intact groups will not be used as the design. (See Mazurkiewicz, 1966; Asher, 1968.) In fact, in one of the most important educational studies ever done, the federal government's Office of Economic Opportunity (OEO) evaluation of Head Start, used a matched individual design. The investigator in this case managed to match in such a way that it is likely some positive results of Head Start were obliterated by the regression-toward-themean phenomenon! Even more disheartening is the fact that the responsible federal officials in OEO's Evaluation Division, who prepared the general design for the Head Start evaluation, seem unaware of the real problem. (For an excellent discussion of these points, see Chapters 10, 10b, and 10c by Campbell and Erlebacher; Evans and Schiller; and Campbell and Erlebacher in Disadvantaged Child, Hellmuth, ed. 1970.) The behavioral sciences' literature in addition to education's also has this same error from time to time. (See Croxen and Lytton, 1971; Vockell and Asher, 1973.)

Another type of design problem, and related data interpretation, also involves regression. Occasionally investigators will test a group of students and select those who score lowest on the test for remedial treatment. After the treatment, they will then retest the students and report that there are gains on the test scores. They ascribe these gains to the quality of the treatment invoked. The treatment might indeed be good, but regression effects are also taking place and causing gains. The direction of the regression gain on the second testing for students selected from the bottom of the tested group will be upward toward the mean of the total group.

A typical example again is selecting children for special classes on the basis of low intelligence or school achievement score. If these children are given a remedial curriculum, or even if they are not, a second testing will show gains as the result of the regression effect. It is interesting to note that many reports of the effectiveness of remedial and special classes appear in the educational literature, but few present before-and-after test data on classes of gifted students, selected on the basis of their high-tested achievement. A second testing likely would show a decrease in scores because of a regression-toward-the-mean (unless maturation and learning effects were sufficient to overcome the regression effect). These results would suggest that when the outcome appears favorable and complimentary to the curriculum developer and

THE UNIVERSITY OF THE STATE OF NEW YORK

THE STATE EDUCATION DEPARTMENT

OFFICE OF THE PRESIDENT OF THE UNIVERSITY

AND COMMISSIONER OF EDUCATION

ALBANY, NEW YORK 12234

November 2, 1977

Dear Congressman Perkins:

I received a carbon copy of a letter sent to you on October 10, 1977 by a William Asher, Professor of Education and Psychological Science at Purdue University. Mr. Asher states that my testimony on the Title I is in error because of a faulty research design.

It is unfortunate that Mr. Asher did not take the time, nor did he have the courtesy, to contact the New York State Education Department concerning my testimony. We are fully aware of the problems he raises in his letter, particularly the phenomenon referred to as "regression toward the mean. The study which I cited in my testimony, in fact, extensively discusses this issue.

If I understand Mr. Asher correctly, his primary objection is that the New York State Education Department accepted a faulty conclusion that "... children who did not participate in compensatory programs scored higher at the end of the year than the Title I children. " The study results did indicate that non-participants scored higher than Title I students. Both the study I cited and my testimony clearly indicated that students were not randomly assigned to either the Title I group or the non-participant group, as would be called for in a true experimental design. Had the evaluation of compensatory programs been conducted with a true experimental design, we would be justified in concluding from the data that the overall program treatments were simply not effective in improving students' reading achievement. We did not conclude, however, that the Title I programs were ineffective. Given the uncertainty of the magnitude and direction of regression effects, our conclusion which I cited in my testimony was that (T)hese results would seem to contain an implicit challenge to the effectiveness of the compensatory education programs involved. " Notwithstanding Mr. Asher's objections, we believe the results of our study do raise questions which should be addressed.

I hope that this letter clarifies any misunderstandings which may have arisen as a result of the letter from Mr. Asher. If you require any further clarification on this matter, please contact me and I will be glad to provide whatever additional information that may be necessary.

PART 12: NATIONAL ORGANIZATIONS

TUESDAY, SEPTEMBER 27, 1977

HOUSE OF REPRESENTATIVES,

SUBCOMMITTEE ON ELEMENTARY, SECONDARY

AND VOCATIONAL EDUCATION, COMMITTEE ON EDUCATION AND LABOR, Washington, D.C. The subcommittee met, pursuant to notice, at 8:30 a.m., in room 2175, Rayburn House Office Building, the Hon. Carl D. Perkins (chairman of the subcommittee), presiding.

Members present: Representatives Perkins, Mottl, Miller, Quie, Jeffords, Goodling, and Pettis.

Staff present: John F. Jennings, majority counsel; Christopher Cross, minority senior education consultant; and Nancy L. Kober, staff assistant.

Chairman PERKINS. The hearing will come order.

Today we have a panel consisting of Dr. Wilson Riles, Superintendent of Public Instruction, California State Department of Education; Mr. Albert Shanker, President of the American Federation of Teachers, AFL-CIO; Dr. Will Davis, President of the National School Boards Association; and Honorable Jeannette Reibman, Pennsylvania State Senator, Chairman of the Policy Committee of the Education Commission of the States.

STATEMENT OF WILSON RILES, SUPERINTENDENT OF PUBLIC INSTRUCTION, CALIFORNIA STATE DEPARTMENT OF EDUCATION, ACCOMPANIED BY WILLIAM WHITENECK, DEPUTY SUPERINTENDENT OF PUBLIC INSTRUCTION, CALIFORNIA STATE DEPARTMENT OF EDUCATION: ALBERT SHANKER, PRESIDENT, AMERICAN FEDERATION OF TEACHERS, AFL-CIO: WILL DAVIS, PRESIDENT, NATIONAL SCHOOL BOARDS ASSOCIATION: HON. JEANNETTE REIBMAN, PENNSYLVANIA STATE SENATOR, CHAIRMAN, POLICY COMMITTEE, EDUCATION COMMISSION OF THE STATES

Chairman PERKINS. Our first witness will be Dr. Riles. Dr. Riles, it is a great pleasure to welcome you here this morning from the great State of California. You have made a wonderful record out there. The record speaks for itself throughout the United States of America. Perhaps you have done as much for the disadvantaged youngster as any school administrator living.

(417)