The Relationship between Criterion, Domain-Referenced, and Norm-Referenced Testing

Professor George Cunningham, ULouisville, feels they are separate process-driven frameworks and not at all able to be combined in one instrument. Victor Steinbok makes a compelling case against this claim.

I offer this brief treatment based upon some other experts' views in the field as an alternative argument for refuting Professor Cunningham's claim..

• Context - the issues as summarized by Victor Steinbok...

(The FairTest, ARN list correspondence from 04Mar02):
Subject: Criterion-referenced tests and arguments against state testing

George Cunningham's claim... (As summarized by Victor Steinbok): There is no way that criterion- and norm-referenced assessments can be combined.  If they are combined, you don't have criterion-referenced assessment. The main difference between standards based and norm-referenced assessment is how the cut-scores are set.  If they are set without any consideration of the resulting pass-rates they are standards-based.  If the cut-scores are set based on their expected impact on pass rates they are norm referenced.

•   Anastasi & Urbina (1997) differ with Prof. Cunningham on "definitions"...

(From their Part Two - Technical and Methodologcal Principles; Chapter 3 - Norms and Meaning of Test Scores)
Section - Domain-Referenced Test Interpretation: Nature and Uses

These authors of a prominent text on psychological testing in its 7th edition since 1954 claim in reply: content-, domain-, and objective-referenced as terms are sometimes employed as synonyms for “criterion-referenced” and sometimes with slightly different connotations.  gradually, the more precisely descriptive terms have been replacing the earlier “criterion-referenced” designation.

• Anastasi & Urbina (1997) on how "domain- and norm-referenced" tests are related...

(From their Part Two - Technical and Methodologcal Principles; Chapter 3 - Norms and Meaning of Test Scores)
Section - Domain-Referenced Test Interpretation: Relation to Norm-Referenced Testing

The Anastasi & Urbina argument at its heart: It should be noted that domain-referenced testing is neither as new nor as clearly divorced from norm-referenced testing as some of its proponents implied. Evaluating an individual’s test performance in absolute terms, such as by letter grades or percentage of correct items, is certainly much older than normative interpretations. More precise attempts to describe test performance in terms of content meaning also antedate the introduction of the term “criterion-referenced testing” (Ebel, 1962; J. C. Flanagan, 1962; Nitko, 1984, pp. 14—16).

•   The argument will certainly be hashed out in much further detail on the ARN list that you can join @ :

Archives of ARN-L@INTERVERSITY.NET :: Assessment Reform Network Mailing List Look in the archives within the first and second weeks of March, 2002...see you there.