Karteikarten
PSR Lecture 5 - Qualitative data analysis
Karten in diesem Stapel (11)
- 1Analysis starts from the very beginning and we generate an iterative relationship with our data as we generate it as we go through the study. Therefore the idea is to keep memos, to keep a log of those interpretive thoughts as you engage with the data, as you proceed through the study in an intuitive fashion.
- 2Analyse from ‘above’ and ‘below’: o working ‘down’ pre-existing ideas, generated before the study inform key themes of the study analysis of content to see if, where, and how they appear refine them on the basis of what is found (‘working up’) o working ‘up’ ideas found in the data, generated through the study categories not specifically asked about but were raised – directly or indirectly – by the data generated analysis of content leads to discovery ensures key themes of the analysis are ‘grounded’So working down is more of a top down approach where the key themes, framing the study, might be generated before the study begins and then apply to the content which is generated by the study and this is quite a familiar way of working in policy applied versions of qualitative research where the key themes of interest are known before the study begins.
A working up is more of a grounded approach whereby your themes or key categories which you want to be working with, are generated through the data and this leads to more of a discovery of new themes being generated which weren't pre-existing or identified before the study. - 3social imagination going beyond description, as understanding the ‘meaning’ of data can involve drawing upon broader perspectives to situate them in context (history, social structure, comparative cases) this is why analysis can involve working iteratively between ideas generated from ‘above’ (beyond the data) and ‘below’ (from the data)
- 4—Framework analysis is better adapted to research that has specific questions, a limited time frame, a pre-designed sample and a priori issues. In the analysis, data is sifted, charted and sorted in accordance with key issues and themes using five steps: familiarization; identifying a thematic framework; indexing; charting; and mapping and interpretation. Framework analysis provides an excellent tool to assess policies and procedures from the very people that they affect.
http://research.apc.org/images/a/ad/Framework_analysis.pdf
- 5One distinction between a framework analysis and most examples of thematic analysis is that in a framework, when charting the data according to the key themes you're working with, you tend to summarise the content rather than work with the kind of actual words and extracts of participants. Whereas with thematic analysis, especially those that are more interpretive and those which remain closer to the language of participants themselves, you tend to work with the actual words and extracts of the participants and how they and what they've actually said in interviews.Thematic analysis provides a purely qualitative, detailed, and nuanced account of data (Braun & Clarke, 2006).
Conversely, content analysis uses a descriptive approach in both coding of the data and its interpretation of quantitative counts of the codes (Downe-Wamboldt, 1992; Morgan, 1993). - 6Steps in a thematic analysis: o familiarisation with the data o identification of themes, and a coding framework (while refining the coding frame through code and case comparison) o annotating interpretations, generating ‘case’ or ‘theme files’, charting the data (by case or theme) o exploring code relationships, building interpretationsTwo different ways today to coding.
1) Top down approach
o ‘first-level’ (root) codes descriptive and broad as a means of ‘chopping’ dataset to generate a coding scheme
o ‘second-level’ (sub) codes break data into smaller units, useful for making comparisons within and across cases. Many of the second level codes might be more likely to be themes which are emerging out of the data in a more bottom up fashion.
o emphasises ‘working down’ especially in policy research where key themes are known
2) Bottom up approach
o ‘line by line’ coding (“open”) on a sub-set of data, look for detail, especially in relation to participant meaning/language, and stimulate ideas
o identify core codes (“axial”) through comparison, explore relations between codes, make analytic memos as you go, and identify core and overarching categories which link to clusters of codes
o emphasises ‘working up’ more commonly linked with ‘grounded’ and ‘theoretical’ analyses - 7Number of codes: Key is to have not too many codes! There's a tendency in computer assisted coding software, such as NVivo to make it very easy to code extracts in very specific ways and to generate many, many codes and this can actually hamper analysis because it becomes difficult to look at the relationships between codes, which then hampers some of the interpretive thinking about what you think might be going on. So I like to work with, as I say, no more than 15 really key parent codes and it's often, I find, that the smaller chunks, the second level codes within them help me do most of the interpretive work in terms of analysing or working out what's going on.By exploring and asking questions of the relationships between themes and codes which helps you build a hypothesis as to what you think is going on in the data.
- 8In qualitative research, analysis is usually said to be inductive in orientation, rather than deductive. In induction, theory or hypothesis emerge after data generation. Your understandings of what is going on will be emerging iteratively through your study and through your data generation, and this is why qualitative studies tend to be quite exploratory as well as hypothesis generating. However in deductive approaches, you start out with a theory or hypothesis before you generate data. Obviously, there are in fact elements of induction and deduction across many forms of research.
- 9In qualitative research we often talk about purposive sampling. Purposive sampling is a way of building into your sample. The dimensions which you theorise will make a difference in relation to your topic and the study. So if, for instance, you think issues of gender are going to be important and will make a difference in relation to what you're studying then you might purposefully sample minimum quotas of women or men, for instance, or it could relate to age. So the idea here with purposive sampling is that you are generating a sample which is sufficiently diverse.Theoretical sampling is a well-known method in grounded theory studies to seek additional data based on concepts developed from initial data analysis. This method involves following where the data have led to expand and refine the evolving theory during the analytical process.
In other words, you are sampling cases in order to challenge the interpretations or understandings that you are building through your data generation. - 10Quality: Maximising ‘validity’ o negative case analysis. The idea here is that within your data set and across your transcripts you seek out the negative case to the rule. The negative case which might challenge the interpretation or understanding which you are building through your data. Sampling new cases is also a way of doing this, as we have discussed. The main reasons for doing this is that it might, in fact, give further confirmation or support to the interpretation you are building or it might need or show you that it needs to be refined or modified. o member validation o triangulation o providing sufficient context for interpretations Maximising ‘reliability’ o cross-checking codes with another researcher o low-inference description – sufficient detail Other o transparent account of methods and interpretation o compare data between/within cases and to other studies o account of the role of the researcher in the researchConclusion 1: Tips when coding
o be open, let the data do the talking—data invokes or provokes ideas without preconceptions
o idea initially is to stimulate ideas, rather than document evidence; be prepared to change. This is especially important at the beginning of a study when you need to stimulate ideas rather than close them down when building up a provisional coding frame.
o stick closely to the data, and look for in-vivo codes—ideas and language of participants.
o code for actions (eg. ‘using drugs behind closed doors’) and insider perceptions (eg. ‘keeping it unaware, protects’), not only topics (eg. ‘secrecy’)
o code for meaning, as well as for concepts
o compare codes, and data to codes, note exceptions, and code rules - 11Conclusion 2: Summary o interpretation and analysis starts as soon as data is generated o move between ideas grounded in, and generated by, the data and surrounding ideas which situate the analysis ‘in context’ - So there's an iterative relationship between our ideas, our interpretations and the data and what that is telling us and they move in sync recursively together. o stay close to your data, allow your data to do the ‘talking’, and build categories from the ‘ground up’. In practice qualitative analysis include combinations of both top down and bottom up approaches, but the more interpretive approaches tend to lean towards more grounded, more bottom up approaches, rather than dealing with mere description. o be systematic in your comparisons within and across cases and categories, pushing the boundaries of your interpretations o allow your systematic comparisons to inform how you sample, as well as generate and analyse data, in an iterative fashion o stay close to your data in your write-up, account for negative cases, and be reflexive as well as transparent
Informationen
- Beschreibung
- public health Karteikarten am PSR Lecture 5 - Qualitative data analysis, erstellt von -- -- am 21/12/2021.
- Erstellt von
- -- --
- Aufrufe
- 0
- Bewertung