-61
Comparisons Symbols
02<01 05<04 X treatment (module) O observation system
02<03 02<04 R random assignment
05<01
Using this design, teachers in groups A, B, and C were observed and given the module on a four week cycle. A univariate F ratio was used to test comparisons 02<01, 05<04, 02<04, 05<01. The T Test for Related Measures was used to test comparison 02<03. Table VIII shows the F and t ratios for the overall observation instrument. While only seven were statistically significant, forty-nineout of fifty were in the same direction. If the module made no difference, then the 95 percent confidence interval would range from seventeen to thirty-two differences favoring the module. Forty-nine differences is approximately seven standard deviations above the expected number favoring the module -twentyfive- if only chance were operating.
Nature of the situation variable indicated the best predictor of change since it distinguished three out of the five comparisons as significant. There was evidence to support that comparison 05<01 showed the most significant change in behavior between the experimental and control