skip to main content skip to footer

Notes on the Nature of Gamble in Test Score Scaling

Author(s):
Tucker, Ledyard R
Publication Year:
1951
Report Number:
RB-51-27
Source:
ETS Research Bulletin
Document Type:
Report
Page Count:
10
Subject/Key Words:
Equated Scores, Error of Measurement, Scaling, Statistical Bias

Abstract

The purpose of this set of notes is to establish a formal structure for the nature of errors in test score scaling. These errors result from the various methods of scaling and various sizes of groups examined in establishing the conversions from raw scores to scaled scores. It is common practice in testing programs to establish a standard type scale for all scores on all tests and forms in the program. Several experimental and quantitative methods exist for obtaining the requisite conversions from the raw test scores to scaled scores. Each of these methods may be used with results from samples of candidates who have been examined. Sometimes special testing materials are required. A common feature of all conversions by any of the methods is that they may miss the mark of the theoretical scaled scores. Errors in scaling then result. It is of practical importance to have a formal system within which to consider these scaling errors.

Read More