Page 92 - The SpeakTeach method - Esther de Vrind

Chapter 4. Perspective of the students - adaptivity
found support and feedback adaptive, linear multilevel analyses were applied, with teacher added as random variable. Linear Mixed models (SPSS version 25, using the Satterwaithe’s approximation to calculate denominator degrees of freedom) were carried out in order to investigate whether there were differences in the degree of adaptivity between the experimental and the control group (factor between subjects) and whether there were differences between the pre- and post-test (factor between subjects), in a first analysis for activities for improvement and in a second analysis for feedback. The same students participated in the pre- and post-tests. Because student-ID was not recorded in the questionnaires, the pre- and post-test scores could not be linked to individual students and therefore, this factor was treated in the analyses as a between-subjects factor.
For both analyses, we tested whether adding teacher (as random intercept) as well as adding a random slope for time per teacher contributed significantly to the model. In order to do this we compared the simpler with the more complex models by comparing the difference in Log Likelihood and chi-squares. In this way, we tested whether some teachers elicited higher scores than others in general (teacher as random intercept) and whether the difference between pre-test and post-test would be different for different teachers (by adding the random slope). Analyses revealed that teacher contributed to both models, and that a random slope for time also contributed to the model for adaptivity of activities for improvement. These models are reported on below. Whenever significant interactions were found between time and group, we carried out post-hoc analyses (data split by group), to interpret this interaction.
Adaptivity of activities for improvement and feedback after a cycle of the self-evaluation procedure – intermediate tests
The number of times the students found the activities for improvement to be adaptive were compared to what would be expected by chance (50%) using the binomial probability function. This analysis was also carried out for feedback.
Research question C
To answer research question C – To what extent did students experience the self-evaluation procedure as motivating and did their speaking anxiety change during the course of iterations of the self-evaluation procedure? – we examined:
90
89