An Experimental Evaluation of Programmed Learning

Abstract

The programmed text used in the study was a 280 frame (8 lessons, 35 frames) Skinnerian-type learn ing program titled Introduction to Counseling (a revision of Ellison, ref. 2). The content of the program included mainly Porter's categories of counselor responses (8) and parts of Perry and Es tes '(7) article dealing with a resolution of the direc tive-nondirective dilemma in education and therapy. The behavior expected from the learner in relation to this content is made clear in the description of the dependent variable measures. (See description of scores below, and Table II appended.) Structurally, the program is Skinnerian, moving gradually from highly prompted initial frames to un prompted terminal frames which define the criterion behavior. There was no fixed standard for frequen cy of review items, speed of fading, or type of prompts. One criterion of the effectiveness of this program is the percent of correct responding within the program. On a constructed response version of the program, the error rate on lessons rangedfrom seven to 14 percent, with an overall error rate of 11 percent. The error rate on criterion frames a lone (within the program) was 12 percent. However, several criterion frames had error rates ranging from 12 to 32 percent. An inspection of these frames reveals that they involve distinctions between Porter's evaluative'' and interpretive counselor response categories. The current version of the program has cut error rates on these difficult frames by 50 percent, but it appears that future improvements are not likely to be made without a redefinition of the interpretive GABRIEL DELLA-PIANA