EFFECTS OF ITEM SEQUENCE, STEP SIZE AND INTELLIGENCE ON A TEACHING MACHINE PROGRAM.

Abstract

Recent studies comparing ordered vs random presentation of a teaching machine program have indicated that retention tests do not reveal significant differences between these two types of sequence (1, 2 ) . The present study was designed to examine the effects of the sequence variable on programs which varied in step size, using Ss who varied in intelligence. Three step-size program sequences of 103, 150, and 199 items designed by Shay ( 4 ) , to reach Roman numerals were used. For the ordered sequence the programs were presented in their original logical order. For random presentation items were presented in a sequence determined from a table of random numbers. Ss were 144 children 9 yr. of age, half with Oris IQs of 110+ and half with Otis IQs of 90-. The 144 Ss were divided into rwo groups, one receiving the ordered form of the programs, the other the random form. Each group was then divided into 3 subgroups each of which was presented with a different step-size form of the program. Finally, each of these 6 groups was divided into a highand a low-IQ group. The design was therefore a 2 X 2 X 3 factorial with 12 Ss per cell and with program sequence, step size and intelligence as rhe variables. Four learning measures were recorded: errors during acquisition, time per item during acquisition, an immediate posttest and a 3-wk. retention tesr. Results of analyses of variance and t tests on these data were as follows: ( a ) The only significant difference for the sequence variable was in errors during acquisition ( P < .01) . This was anricipared as the original sequences were designed on the basis o f error rate and thus altering the sequence by randomizing rhe items should increase the error rate. Neither the immediate posctest nor the 3-wk. retention test indicated significant differences berween the two sequences. ( b ) The 3 step-size forms produced no significant differences in the immediate test, but Ss receiving the 199-item program scored higher on the 3-wk. retention test than Ss given the 103-step program ( P < .01) . (c ) Ss trained on the 199-item program made significantly fewer acquisition errors than Ss trained on the 150-item program ( P < . 0 5 ) . ( d ) Ss trained on the large step-size program spent a significantly longer time per item than Ss given the two smaller step-size programs ( P < .01). ( e ) Ss in the high-IQ group performed significantly better than the low-IQ group on all 4 measures. N o interactions or other significant differences were found. These results support the studies to date [with the exception of Roe ( 3 ) ] on ordered vs random presentation of a program and indicate some of the difficulties underlying program construction. If the end result of a program is to produce a change in behavior as measured by a posttest, the present results suggest little confidence in designing thc sequence of a program on error rate alone.