Martine van der Pluijm

144 Chapter 5 We found significant changes on three of the four constructs of the dimension quality of interaction for both activities: child involvement (Activity 1: F (1, 18 = 13.37, p = .002, Activity 2: F (1, 18) = 7.03, p = .01), autonomy (Activity 1: F (1, 18) = 8.308, p = .01, Activity 2: F (1, 18) = 4.80, p = .04) and emotional support (Activity 1: F (1, 18) = 4.57, p = .04, Activity 2: F (1, 18) = 5.20, p =.03). The partial eta squared effect sizes vary between .42 and .28 for child involvement, .32 for autonomy (i.e., strong effect sizes) and between .20 and .22 for emotional support (i.e., medium effect size) (Cohen, 1988). We found no significant differences between pretest and posttest for the other variables in Table 5, although posttest means are higher than pretest means in a large majority of the cases. Table 5.6 shows the results of the factorial repeated measures Anova testing effects of the quality of teacher delivery on changes between pretest and posttest scores for Activity 2. We found significant effects of quality of delivery on change of each of the three constructs (quality of the interaction, quantity of the interaction, and quality of language) for Activity 2 at the posttest. We did not find significant effects for Activity 1. We report results of the Mann-Whitney test when the assumptions of equality of error-variances were not met. TABLE 5.6: Effects of quality of delivery on change of parent-child interaction in Activity 2 F (DF) p η p 2 U Z p r Quality of interaction Child involvement .32 (17) .60 .02 Autonomy .20 (17) .66 .01 Emotional support 1.42 (17) .25 .08 Cognitive support 15.86*** (17) .001 .48 Quantity of interaction Number of words child 10.55** (17) .005 .38 Number of words parent 4.00** -2.60 .006 .60 Turn-taking 4.00** -2.60 .006 .60 Quality of language Contextualized language .26 (17) .61 .02 Decontextualized language 4.88* (17) .04 .22 Interactive language 0.09 (17) .77 .00 Other language 5.50** -2.45 .01 .61 * p = <0.05, ** p = <0.01, *** p = <0.001. η p 2 = partial eta squared, r = Rosenthal (1991)