Page 149 - Balancing between the present and the past

in the mean pre-post change divided by the standard deviation of the untreated population. In our case, this effect size is .72, which is an effect between intermediate and large. This standardized effect of the treatment is significant (p = .001).
Table 23. Students’ mean scores on historical contextualization
Experimental Control
101 11.00 (2.47) 11.53 (2.37) 30 11.07 (1.98) 9.90 (2.43)
Total 131
To examine the intervention effect, we first used a multilevel analysis to explore the
extent to which the differences in student achievement on historical contextualization
can be explained by the differences between classes. We specified classes as a random
factor and the pre-test scores as a fixed factor (-2LL = 539.25). This model showed that
the total variance of student achievement is 4.20 and that 22% of this variance (0.94)
can be explained by the differences between classes. Next, we specified classes as
a random factor and the pre-test scores and condition as fixed factors to examine 6 the extent to which the differences in student achievement between the different
classes can be explained by participating in the experimental condition (-2LL = 535.02, indicating a better fit). This model showed a total variance of 3.71, and 12% of this variance (0.46) can be explained by the differences between classes.
The comparison of the two models showed that the treatment only affected the variance explained by the differences between classes (which decreased from 0.94 to 0.46) and not the residual student variance, which remained the same. The result is that more than half (51%) of the differences between the different classes can be explained by participation in the experimental condition. The effect of the treatment on the differences between the classes was significant (p < .05). We calculated the effect size to examine the amount of variance within the experiment that is explained by the treatment. Our multilevel analyses showed that the treatment was responsible for 11% of the differences in student achievement between students in the experimental condition and those in the control condition, which is considered a medium effect (Cohen, 1988).
A historical contextualization pedagogy
Condition
n
Pre-test mean (SD)
Post-test mean (SD)
147