Maria Knobelsdorf, University of Dortmund, Germany - Didaktik der ...
Maria Knobelsdorf, University of Dortmund, Germany - Didaktik der ...
Maria Knobelsdorf, University of Dortmund, Germany - Didaktik der ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
The assumption <strong>of</strong> sphericity was not met for either the content<br />
concepts (W = 0.003, χ 2 104 = 178.35, p < 0.001) or the process<br />
concepts (W < 0.001, χ 2 119 = 248.60, p < 0.001) at the α level <strong>of</strong><br />
0.05. In the further analyses, we therefore applied the ε correction<br />
<strong>of</strong> degrees <strong>of</strong> freedom proposed by [23], as presented in table 1.<br />
Table 1. Results <strong>of</strong> the ANOVA with Huynh–Feldt ε correction<br />
<strong>of</strong> degrees <strong>of</strong> freedom<br />
The main effect A (BW vs. BY teachers) was not significant at the<br />
α level <strong>of</strong> 0.05 (F1, 36 = 0.24, p < 0.63). The corresponding H0 was<br />
therefore not rejected: The teachers from BW respectively BY did<br />
not differ in their global evaluations <strong>of</strong> the content concepts.<br />
The interaction effect A × B (group × content concept) was significant<br />
at the α level <strong>of</strong> 0.05 (F10, 360 = 2.26, p < 0.02). The corresponding<br />
H0 was therefore rejected: The teachers from BW respectively<br />
BY differed significantly in their evaluations <strong>of</strong> individual<br />
content concepts.<br />
The interaction effect A × B × C (group × content concept ×<br />
process concept) was not significant at the α level <strong>of</strong> 0.05 (F78, 2809<br />
= 1.09, p < 0.29). The corresponding H0 was therefore not rejected:<br />
The teachers from BW respectively BY did not differ in their<br />
evaluations <strong>of</strong> the relationships between individual content concepts<br />
and individual process concepts.<br />
7.4 Individual Comparisons for the A × B Interactions<br />
The global test <strong>of</strong> the A × B interaction revealed a significant<br />
overall effect <strong>of</strong> group × content concept. Therefore we evaluated,<br />
which concepts were rated differently by comparing the mean<br />
values, applying t-tests in or<strong>der</strong> to test simple AB effects for<br />
p•q×r split-plot designs (see [46], pp. 535–536), concerning the ε<br />
correction <strong>of</strong> the degrees <strong>of</strong> freedom (see section 7.3). Fig. 4<br />
visualizes the means <strong>of</strong> the A × B interaction. As the figure shows,<br />
the content concept model was rated significantly different by the<br />
two groups <strong>of</strong> teachers (a1, a2) at the α level <strong>of</strong> 0.05 (t396 = 2.23, p<br />
< 0.027). The differences for system, computer, and information<br />
are remarkable, but not significant.<br />
7.5 Individual Comparisons for the A × B × C<br />
Interaction<br />
The global test <strong>of</strong> the A × B × C interaction did not reveal a significant<br />
overall effect <strong>of</strong> group × content concept × process concept.<br />
Taking into account the significant difference regarding the<br />
content concept model (see section 7.4), it makes sense to compare<br />
only this concept with respect to the process concepts a<br />
posteriori. Fig. 5 displays the comparisons <strong>of</strong> the means on the<br />
concept model regarding the different process concepts.<br />
71<br />
These were calculated by applying 16 t-tests to analyze simple<br />
AC effects for SPF-p•q×r experimental designs ([46] pp. 535-<br />
536); An ε correction <strong>of</strong> degrees <strong>of</strong> freedom was taken into account<br />
again (see above). The t-tests were calculated at an adjusted<br />
α level <strong>of</strong> 0.05/16 = 0.0031. It turned out that the ratings from BW<br />
teachers (a 1) respectively from BY teachers (a 2) differed significantly<br />
regarding the content concept model related to the following<br />
process concepts: classifying, finding relationships, generalizing,<br />
comparing, questioning, and or<strong>der</strong>ing.<br />
Content concepts<br />
b1 = problem<br />
b2 = information<br />
b3 = model<br />
b4 = algorithm<br />
b5 = data<br />
b6 = structure<br />
b7 = system<br />
b8 = computation<br />
b9 = process<br />
b10 = s<strong>of</strong>tware<br />
b11 = program<br />
b12 = test<br />
b13 = communication<br />
b14 = language<br />
b15 = computer<br />
0.0<br />
1.75<br />
Rating<br />
2.0 2.25 2.5 2.75 3.0 3.25 3.5<br />
Groups<br />
a1= BW teachers <strong>of</strong> computer science<br />
a2= BY teachers <strong>of</strong> computer science<br />
a1<br />
a2<br />
__<br />
simple AB effects<br />
0.85-<br />
0.63-0.84<br />
0.53-0.62<br />
0.31-0.52<br />
0.16-0.30<br />
0.00-0.15<br />
Figure 4. Comparisons for the factor level combinations A × B<br />
Rating<br />
4.0 4.5<br />
4.25<br />
3.75<br />
2.0 2.25 2.5 2.75 3.0 3.25 3.5<br />
1.75<br />
c1 = analyzing<br />
c2 = classifying<br />
c3 = problem solving and posing<br />
c4 = categorizing<br />
c6 = finding relationships<br />
c5 = investigating<br />
c7 = generalizing<br />
c8 = creating and inventing<br />
c9 = comparing<br />
c10 = finding cause-and-effect r.<br />
c11 = questioning<br />
c12 = transferring<br />
c13 = communicating<br />
c14 = presenting<br />
c15 = collaborating<br />
c16 = or<strong>der</strong>ing<br />
p < .0006<br />
p < .003<br />
p < .006<br />
1.05-<br />
0.91-1.04<br />
0.67-0.90<br />
0.45-0.66<br />
0.23-0.44<br />
0.00-0.22<br />
p < .01<br />
p < .05<br />
p < .10<br />
Figure 5. Comparisons for the content concept model<br />
8. DISCUSSION<br />
The results <strong>of</strong> the performed evaluations support the research<br />
hypothesis that computer science teachers from Baden-<br />
Württemberg differ from computer science teachers from Bavaria<br />
in the assessment <strong>of</strong> key content concepts <strong>of</strong> computer science<br />
related to central process concepts <strong>of</strong> computer science. Already<br />
from the descriptive evaluation, it has become clear that there are<br />
differences in the assessment <strong>of</strong> content concepts by the two<br />
groups <strong>of</strong> computer science teachers. There were differences for<br />
the concepts <strong>of</strong> content model, system, computer, and information.<br />
The analysis <strong>of</strong> variance and the individual comparisons showed<br />
that the two groups rated the individual process concepts classifying,<br />
finding relationships, generalizing, comparing, questioning,<br />
and or<strong>der</strong>ing differently with respect to the content concept model.<br />
As the regular teacher education programs in the German states<br />
are standardized by the KMK (see section 5), it is not likely that<br />
those differences would be caused by the courses <strong>of</strong> lessons in CS<br />
that the teachers had attended at their universities. Therefore, the