It seems you are using multiple sources of data, (Triangulation) for your research on students' identity. You will probably be doing thematic or content analysis of the data. Its important to decide what you are trying to find in data from all these sources. I personally believe that identity is more of cognitive nature phenomena so you may use interview as main data source and use the other two to confirm what come out of interview results. Alternatively, you may give equal weight to others data and see students identity with the target appear in observation. Then you compare the results. If the results from all the sources are converging or complementing then the results will be more convincing. Similarly, analysis by multiple researchers will also add to the credibility of your results.
Your method depends on your goals (research objectives, and design): the importance you give to any of the three approaches depends on the relevance you have put to each in your research design, and in the objectives.
Generally speaking, interview gives you the personal aspects of the identity, perhaps the most important ones; focus group gives you the grupal aspects of the issue, in the context of a dialogue; and observation gives you the possibility to triangulation, and to get some aspects which the other 2 sources of data cannot give you.
BUT this is true if and only if the instruments, procedures, and items in the instruments have been correctly prepared and executed. And if they ALL are completely COHERENT with each other.
So, you must apply content analysis (qualitative, and quantitative) to the interviews, in order to discover the aspects of identity you are looking for. And the method must be in accordance to your research objectives.
And it is important to bear in mind that the variables (quantitative), or the categories (qualitative) you explore in the interview be the SAME you explore in the other 2 sources of data.
But this is all generally speaking, it is important to check the whole project in order to get a clear view of.
I agree with Anotonio Velasco that a lot depends on your goals. I identity is a very broad and abstract topic. Can you say more about what you want to do with your data?