A PRELIMINARY INVESTIGATION OF THREE COMPROMISE METHODS FOR ESTABLISHING CUT‐OFF SCORES

The study compares three methods for establishing cut-off scoreS that effect a compromise between absolute cut-offs based on item difficulty and relative cut-ffs based on expected passing rates. Each method coordinates these two_types of information. differently. The Beuk method_obtains judges' estimates of an absolute cut-off and an expected passing rate, and constructs a cutting line whose slope is the ratio of the absolute and relative standard deviations and which passes through the point of_ mean absolute/relative cut-off. The judges can be either test-oriented or -examinee-oriented depending on whether they show greater agreement (small standard deviations) on the absolute or relative cut-offs. The Hofstee method draws a cutting line_through two extreme points: (1) maximum cut-off, minimum failure point; and (2) minimum cut-off, maximum failure point. The DeGruijter method is similr to the Beuk method, but uses confidence estimates for the absolute and relative cut-offs to define a criterion ellipse. These methods_were applied to two,tests from a certification program. Judges rated item difficulty by the Angoff_method and estimated a desirable passing rate. All three compromise_methods brought the cut-off two points below the absolute level, in line with an_acceptable passing_ rate. This study suggests that further research into all three of the compromise methods is needed. (LPG)