A comparison of control groups and

A comparison of control groups and experimental groups indicates that an accounting change did induce changes in decisions by memhers of the experimental groups. In the increase condition, the average difference score across all subjects in the control group was 4.26 percent, while the score averaged across all subjects in the corresponding experimental groups was 27.23 percent. The one-tailed t value (df = 51) is 3.096 and is significant at the .0016 level. Corresponding values in the decrease condition are 6.91 and 14.66 percent, resulting in a t value of 3.246, significant at the .0010 level. The analysis-of-variance conducted on the four experimental groups
supports my hypotheses concerning the increase/decrease condition and the interaction, but not the information condition. As stated earlier, I hypothesized that (a) the increase condition (changing from full cost data to variable cost data) would result in a larger change in decision models (less fixation) than would the decrease condition, {b) the high information condition would result in less fixation than the low informa-
tion condition, and (c) the interaction between the two experimental treatments would not be significant. Comparison of increase {x = 27.23%) and decrease ix = 14.66%) conditions yielded anF value {df = 1,76) of 8.20 and &p value of .005. Comparison of low information (a": = 21.70%) and high information (x = 20.19%) conditions yielded axxF {df = 1,76) of 0.12 and a p of .731. Interaction resulted in a n F {df = 1,76) of
0.38 and a p of .538. The interaction graph in figtire 2 indicates that the effect of one experimental treatment did not depend on the effect of the other treatment.
DOLLAR AVERAGE DIFFERENCE SCORES
Analysis in terms of dollar average difference scores yielded statistically nonsignificant results for both treatments and the interaction. For the experimental groups, the increase/decrease condition yielded a n F = 1.13, p = .291; the information condition, F = 0.20, p = .657; and the interaction, F = 0.75, p - .388. A comparison of control groups with experimental giroups indicated that the accounting change induced
changes in decisions hy subjects, just as the analysis in terms of percentages had indicated. The t value for the increase condition was 1.954, p = .0281, while corresponding values for the decrease condition were t = 3.389, p = .0007. These results were unchanged by the exclusion of an apparent outlier (the second subject listed in the increase-low information group in table 2). It should be emphasized, however, that the analysis in termts of dollar amounts is likely to be much less valid than the analysis in terms of percentages because percentage deviations from

Hasil (Bahasa Indonesia) 1:

A comparison of control groups and experimental groups indicates that an accounting change did induce changes in decisions by memhers of the experimental groups. In the increase condition, the average difference score across all subjects in the control group was 4.26 percent, while the score averaged across all subjects in the corresponding experimental groups was 27.23 percent. The one-tailed t value (df = 51) is 3.096 and is significant at the .0016 level. Corresponding values in the decrease condition are 6.91 and 14.66 percent, resulting in a t value of 3.246, significant at the .0010 level. The analysis-of-variance conducted on the four experimental groupssupports my hypotheses concerning the increase/decrease condition and the interaction, but not the information condition. As stated earlier, I hypothesized that (a) the increase condition (changing from full cost data to variable cost data) would result in a larger change in decision models (less fixation) than would the decrease condition, {b) the high information condition would result in less fixation than the low informa-tion condition, and (c) the interaction between the two experimental treatments would not be significant. Comparison of increase {x = 27.23%) and decrease ix = 14.66%) conditions yielded anF value {df = 1,76) of 8.20 and &p value of .005. Comparison of low information (a": = 21.70%) and high information (x = 20.19%) conditions yielded axxF {df = 1,76) of 0.12 and a p of .731. Interaction resulted in a n F {df = 1,76) of0.38 dan p.538. Grafik interaksi dalam figtire 2 menunjukkan bahwa efek pengobatan eksperimental yang satu tidak tergantung pada efek perawatan lainnya.DOLLAR NILAI RATA-RATA PERBEDAAN Analisis dalam hal nilai rata-rata perbedaan dolar menghasilkan statistik nonsignificant hasil untuk perawatan dan interaksi. Untuk kelompok eksperimental, mengurangi kondisi menghasilkan n F = 1.13, p =. 291; informasi kondisi, F = 0,20, p =. 657; dan interaksi, F = 0.75, p -.388. Perbandingan kelompok kontrol dengan eksperimental giroups menunjukkan bahwa perubahan akuntansi diinduksiperubahan dalam keputusan hy mata pelajaran, sama seperti analisis dalam hal persentase telah menunjukkan. Nilai t untuk meningkatkan kondisi adalah 1.954, p =.0281, ketika nilai-nilai yang sesuai untuk kondisi penurunan t = 3.389, p =.0007. Hasil ini yang tidak berubah dengan pengecualian outlier jelas (subyek kedua tercantum berkelompok kenaikan-rendah informasi dalam tabel 2). Harus ditekankan, namun, analisa termts jumlah dolar mungkin menjadi jauh lebih sah daripada analisa persyaratan persentase karena persentase penyimpangan dari