The process described above involves developing categories from the data. A variation on this process is to sort the information into previously determined categories. For example, you may be trying to find out about reading strategies used by learners at different stages. You may decide to use categories from other research, such as "print-based" and "meaning- based." A drawback to using predetermined categories is that they will influence how you see your data and you may miss new ways of seeing things. Whether you use predetermined categories will depend on your research purposes and questions.
Once data has been sorted into categories, you can begin to construct themes or broader categories that address your research questions. In developing themes, you'll likely draw from the data, from your own experiences and knowledge, and from the relevant literature. For example, in my research about sharing power, one of the themes I found was "devolution of power." This theme came from recognizing how, in a number of instances, I tried to hand over power to the group. But it also came from reading about power and approaches to keeping or sharing power.
In my research about using arts-based approaches in literacy education (Norton, 2008), I identified categories, such as "making room for emotions," "self-esteem and confidence" and "drawing on and discovering strengths." I then looked at how these categories related to larger themes related to holistic learning, such as emotional learning and spiritual learning.