Appletcollection Vertical Menu java applet, Copyright 2003 GD

The OVERALL ACHIEVEMENT TEST: CUMULATIVE EVALUATION REFLECTING EDUCATIONAL ABILITY LEVELS (sometimes referred to as simply: OAT-CEREAL) was created by Dr. John Willis to illustrate one aspect of the almost total mindlessness of grade equivalents for the reporting of educational test data. The apparent clarity and simplicity of these statistics are deceptive traps for the unwary. At first glance, grade equivalents seem to reflect, with a commendable return to basics, a student's actual functioning level in a given subject area. However, since grade equivalents are simple transformations of raw scores, a student may in fact do no work at all on the grade level reported. The child might do unusually accurate work below the reported level-or the opposite.

Take, for example, Ralph, who tends to be very careless with simple computation and has a slight tendency to confuse the numbers, 6 and 9. Consequently, on the OAT-CEREAL Math test, he failed items 1, 2, 3, 4, 5, 6, 7, and 9, and obtained a raw score of 4 for a grade equivalent of 4.5. This despite the fact that he had no difficulty whatsoever with items 8, 10, 11, 12, all far above a fourth grade curriculum.

Bunny, an otherwise very competent reader, still confuses the letters, t and f. This unfortunate error caused her to hilariously mispronounce eight of the twelve words on the OAT - CEREAL Oral Reading Test and obtain a grade equivalent of 4.5, although she read (and understood) without difficulty the words, cacophony and polyandry.

The point -- obviously enough -- is that, while grade level designations of test items or groups of items may be meaningful and even helpful from time to time, especially with criterion-referenced measures, grade equivalents of raw scores not only fail to present any meaningful information but may in fact serve to convey impressions that are entirely contrary to fact. For normative purposes, it is therefore much wiser to compare student's test performance to that of the student's peers through standard scores, percentiles, stanines, normal curve equivalents, or similar ranking or deviation measures.

"Statistics don't lie, but statisticians do."

For more information, see:

International Reading Association (1982). Misuse of grade equivalents: resolution passed by the Delegates Assembly of the IRA, April 1981. Reading Teacher, January, p. 464.

 

Lyman, H. B. (1991). Test scores and what they mean. Boston: Allyn and Bacon. pp. 112-113.

 

Reynolds, C. R. Conceptual and technical problems in learning disability diagnosis, (Chapter 24) in Handbook of Psychological and Educational Assessment of Children: Intelligence and Achievement (Reynolds & Kamphaus) (1990) Guilford Press

 

Sattler, J. M. (1992). Assessment of children, rev. & updated 3rd ed. San Diego: Jerome M. Sattler, Publisher. pp. 20-21.

 

Willis, J.O. & Dumont, R.P. (1998). Guide to identification of Learning disabilities, 1998 NY State edition. Acton, MA: Copley. pp. 69-70, 222-223.

 

A variation of the OAT CEREAL first appeared in the New Hampshire Personnel and Guidance Journal (1977). page 9.

As you can see, John's been at this for a while!

Content on these pages is copyrighted  by Dumont/Willis © (2001) unless otherwise noted.