hi ali,
my understanding of mixed methods designs is that they bring together approaches to research that are perhaps informed by different epistemologies, or notions of what constitutes knowledge. when you say that you are proposing the use of “coded classroom observations of frequency and type of pupil teacher interaction” it sounds to me like you are choosing a structured observation method which will generate primarily quantitative findings. this is quite a different to what your semi-structured interviews will generate – i.e.textual transcripts which are well-suited to thematic analysis.
i would encourage you to identify your epistemology, your research questions and your methodology *before* you nail down the methods, so that you are clear what you are hoping to find out about this concept of ‘belongingness’ and why and also how you will know that your data provides an answer to your question(s). it will also be helpful to look at what the literature in the field suggests for future research, as quite often that will provide you with a particular niche to explore and also some ideas about an appropriate methodology e.g. if there’s been hundreds of studies of a particular type done in the area and there’s a huge gap in terms of content or process, there’s your opportunity to contribute to the field.
cheers,
sam