I am doing a qualitative research with interview as primary data and using thematic analysis. This is my first time doing investigator triangulation. I plan to code the transcriptions independently then compare the results afterward. However, I predict there will be some differences in results from each of the coders. How I resolve this? and which data should I use? in quantitative research, we can say the validity is for example 0.7. How can we say the validity in qualitative research?
Thank you.
You could to build a framework of main categories, looking for consistency, and apply participant observation. Also yo could to identify some key informants and organize a focus group, and then compare your results with interviews. In qualitative research, the validity is only internal, not external, and it depends of the methodological strength.
If I understand your planned analysis correctly, there will be two main steps in the analysis, namely, the formation of the codes/categories (commonly dummy variables that indicate the presence of a particular idea or theme) that will be used in the content/theme analysis, and secondly the coding of each participant's response against each of these codes. It will be difficult to determine in any systematic manner the level of agreement between coders unless they are using the same set of coding categories.
To derive a commonly agreed upon set of coding categories, a common approach is for coders to use a randomly selected set of responses to separately form a set of coding categories, that are determined, not only by the responses, but also by the purpose of the research or any specific hypotheses. Then the different sets of categories can be inspected and a consensus meeting used to decide on a final set of categories to be used for the second step. When the coders have coded each participant's response against each of the coding variables, then the level of agreement between the raters' can be measured using a variety of well-known statistical procedures (kappa, intra-class correlation, etc.).
I do not think that investigator triangulation and inter-coder reliability are the same thing. First, true investigator triangulation would involve more than one person in the data collection. Second, agreeing on codes is not the same as agreeing on the larger interpretive analysis.
In addition, I'm not sure that inter-coder reliability would add that much to full-scale thematic analysis, since it is usually associated with deductive approaches to manifest content, rather than inductive approaches to latent content. Hence, inter-coder reliability is more often associated with descriptive forms of content analysis, rather than interpretive forms of thematic analysis.
Overall, I would suggest that it is rare to have detailed investigator triangulation unless you are working in a team from the beginning of the research. So, it is unlikely that anyone will hold you to that standard.
- Badges
- Science method
- Similar topics
- Methodology
- Research Methodology
- Qualitative Research