skip to main content skip to footer

Construct Validity and Construct Similarity: Methods for Use in Test Development and Test Validation MCAT

Frederiksen, Norman O.
Publication Year:
Report Number:
ETS Research Report
Document Type:
Page Count:
Subject/Key Words:
Medical College Admission Test (MCAT), Medical Education, Problem Solving, Research Methodology, Test Construction, Test Validity


The first concern in test construction should be the development of a sound theory regarding the abilities to be measured. Methods like those used by cognitive psychologists in developing process models are recommended. Test development might then begin with construction of a criterion test for use in verifying or revising the theory and in subsequent validation studies. The criterion test might take the form of realistic simulations of situations that elicit the behavior to be predicted and that yield scores reflecting all the important aspects of performance. Construct validation may involve correlational methods or other procedures (such as comparing response latencies of experts and novices) to test the process model. Construct validity is supported by relationships of the criterion scores to other measures that are consistent with the process theory. Tests that are more efficient and economical may be required for operational use. The validity of such tests can be investigated by examining construct similarity—the extent to which relationships to process variables are similar for the operational and the criterion measures for a sample of subjects. Construct similarity is also useful for comparing test formats (such as free-response and multiple-choice) to see if they measure the same construct and for comparing groups (such as trained and untrained candidates) to see if a given test measures the same construct for different groups. Studies of construct similarity might similarly be useful in cross-cultural research. Application of these ideas on construct validity and construct similarity in a study of the problem solving behavior of medical students will be described. (36pp.)

Read More