we have been using a sequential explanatory mixed method research where the findings from the quantitative study are reassessed with qualitative approach such as through interview. The selection of participants is based on purposive sampling but the appropriate number will be guided by data saturation. so when shall we say that there is data saturation? what's the minimum number of interview needed to be conducted?
After having conducted a series of interviews, there is a point when we start receiving very repetitive responses the interviewees. Usually no new themes emerge and there is high rate of duplication or recurrence of responses. This is an indication to stop the interview process and analyze the data that has been collected so far.
There are basically two approaches to saturation. The original one comes from grounded theory and is known as "theoretical saturation." This means that the data you are collecting adds nothing new to your understanding of the research topic (i.e., your data no longer contributes anything new to your ability build theory). Note that in grounded theory, one continually modifies the data collection procedures according to what was learned from previous data, rather than simply repeating the use of one standard data collection instrument
More recently, people have shifted that early definition of saturation into something close to what Ruchi describes: not producing any new codes. This implies that you are using the same data collection instrument and you repeat it until no new codes are produced. I would thus propose that we label this "code saturation" to distinguish it from grounded theory's concept of theoretical saturation.
The best known citation on code saturation is by Guest et al., but the paper by Galvin has a stronger theoretical basis
I think Corbin, Juliet M., and Anselm Strauss. "Grounded theory research: Procedures, canons, and evaluative criteria." Qualitative sociology 13.1 (1990): 3-21. say its when same comments/anecdotes recur and you cease to get much (especially dissenting views) from interviewing more subjects
Agreed with the elaborated information from David. In my experience, after going through the data set after certain gap or to have second look often helps in getting new codes. Before regarding data as saturated, Tesfaye, you may like to try this as well...
Data saturation come along with two effects: to the interviewer, nothing new is added to what has been obtained initially and to the interviewee, it create boring and uninteresting situation.
I also suggest you read up COREC guidelines for qualitative research.
You will need also around 30 interviews minimum to get it published in any decent journa.
Saturation occurs when no new information is got from the interviewees, usually participants are between 17 to 30 maximum. Data analysis is cumbersome if too many participants are involved.
The required number of interviews depends on several factors the first of which is the nature of the topic itself. Consulting the extant literature on the topic (numerous research papers / or very few) will be the very important guide for determining the needed / most appropriate number of interviews. In some disciplines, six interviewees we considered acceptable to provide the needed information. In other disciplines, 20 interviews considered good number to provide good information on the subject. Then, saturation level will enormously vary among disciplines and the researcher needs to refer back to the literature to be safe.
- Badges
- Science topic
- Similar topics
- Medicine
- Counseling Psychology