The battle has come between two things: deductive versus inductive methods (and methodologies), and how to select methodologies suitable for the type of data and for the purpose of analysis. There now seems like a solution on the horizon, even though I don’t think I’ve reached it yet. I can mix and match, it is allowed.
This is the core of the challenge: I have mixed data – literature and people data. Literature is partially quantitatively (initial selection criteria) and then qualitatively sourced, using a priori deductive categories, then further inductively categorised. All people data is inductively analysed. The two sets need to be combined later. I’m now focused on two methodologies to use: phenomenography and qualitative content analysis.
Data set 1 – Learning experience data, as reported by learners through interviews AND what is gleaned from reviewing their online digital interactions data. This is inductively categorised using a phenomenography approach for ‘outcome spaces’ (which can be areas of focus that are predetermined) which each have ‘variation categories’ (which arise from the data).
Data set 2 – Literature analysis ‘data’ – papers sourced through fixed (a priori) category faceted search in systematic search sets, selecting from highest number of citations. This means a deductive categorisation using an initial partially quantitative method of selection. The papers are then further analysed and categorised inductively for keyword contexts. This is done using qualitative content analysis, first deductive, then leading to an inductive method.
Data set 3 – A final frame work must be generated, consisting of the two sets of data combined and compared, to establish theoretical presence in literature in relation to experience characteristics of ‘learning effectiveness’. I think this is achieved by an inductively arrived at set of categories though again the areas of focus will already exist. But what is in them is inductively produced from the previous sets of analysis.
It is complicated, but actually I don’t see another way of doing this, apart from ditching the whole proposal and just going for simple learning experience effective measurement.
These key texts are what I’m basing these ideas on:
