80 Chapter 3 and discussed within the research team. After the coding procedure we re-read all transcripts again to ensure data confirmability, and to confirm ‘meaning saturation,’49 that is we checked if we had harvested all new insights from the data. During the third step, in order to answer our second research question we clustered the individual participants who shared similarities in themes qualitatively into teacher profiles. The theme that each participant emphasised as most relevant was leading for the clustering. To enhance the credibility and trustworthiness of this step in the analysis, the research team frequently gathered to discuss potential clustering, which had to be confirmed by the interview data before being accepted. In the final step we analysed the participants in each profile quantitatively comparing them to all other participants with regard to the coded subthemes, which led to a richer picture of each of the profiles. In order to address the third research question the profiles were related to the contextual and personal factors. For this, we used a Fisher’s exact test to study differences regarding categorical data and a Student’s t-test for continuous data. IBM SPSS Statistics 20 was used. Ethical approval and consent to participate This study was granted an ethics waiver by the Medical School Ethics Committee of the LUMC. The same committee advised that formal written informed consent was not required. All participants were invited by e-mail or telephone by the first author, who emphasised that participation was voluntary and anonymous, and gave oral consent. RESULTS We decided to exclude one SUSM interview for further analysis, because the teacher-educator was responsible for the teach-the-teachers programme which we felt created a potential conflict of interest. Because of his educational expertise (he had obtained a PhD in education), he appeared to answer less from his own teaching experience and more from the theoretical framework that he used in his professional development programme. In the first section of the results we will start with a description of the six themes. All six themes were adequately represented in the data (see number of text fragments per theme, Table 3.1), and data analysis did not generate new
RkJQdWJsaXNoZXIy MTk4NDMw