Page 91 - The SpeakTeach method - Esther de Vrind

self-evaluations during the period of the intervention. The pace was tailored to the students, the curriculum and other differences in the context.
In order to evaluate changes within learners, we included the same 142 students for the first and fourth cycles in the analyses and used the McNemar-Bowker test for matched pairs. In this way, we examined whether the evaluations in the first cycle were focused on certain categories of errors and positive points in particular by counting the frequencies of areas for improvement and the frequencies of positive points the students had noticed in their first cycle for the categories (message, vocabulary, grammar, pronunciation and fluency) and compared these frequencies with frequencies in the fourth cycle for each category.
In the same way, we used the McNemar-Bowker test to investigate to what extent shifts in focus in the plans for improvement occurred.
Development in independence
To answer the question to what extent the students went through the cycle of self-regulation more independently (research question A.2) we used the McNemar-Bowker test for matched pairs to analyse the extent to which students requested their teacher’s assistance in the fourth cycle compared to the first cycle.
Research question B
To answer research question B - To what extent did the students consider feedback and activities for improvement provided after their diagnosis to be adaptive to their needs? – we performed two different analyses:
Adaptivity of activities for improvement and feedback from the learners’ perspective – pre- and post-test
We first calculated the reliability (Cronbach's alpha) of the items in the questionnaire that should measure the same constructs, namely: the construct adaptive activities for improvement and the construct adaptive feedback. Both proved to be reasonably reliable (adaptive activities for improvement α = .68; adaptive feedback α = .76 after deleting one item).
With participants in this study taught by different teachers, the data were structured hierarchically. Since teachers could influence differences in the extent to which students
88
89
4