Qualitative research has now been widely employed by researchers. Yet, some research work seems to be not satisfactory enough in terms of academic inquiry and requirements though it contains a high degree of rigor.
Lots of interesting answers here…but I think it is really important to clarify exactly:
- what you mean by 'qualitative' research (very broad term)
- what subject area you are researching
- what methodological traditions and epistemological background you are operating from
….without meaning to sound noncommittal or esoteric, it is important to consider. As many have suggested, reaching a point of data saturation may be a sensible end-point, but not necessarily so (or, indeed, possible) with Case Study research. Additionally, various disciplines, such as the various phenomenologies, have particular views on authenticity, and the place for 'generalisability' which often drives decisions around sample size etc.
Then there is the subject you are researching…it may be so niche, that your sample size is dictated by the number of people who meet the eligibility criteria, or who volunteer to participate…
I would be very cautious about opting for a standard sample size and using minimum criteria that seem reminiscent of positivist values… but again, this really depends on the philosophical assumptions and traditions you and your project are operating from… but there is no single best answer really…
Another thing to consider is what is accepted by the wider research community- look at what has been published in this field, you may find a great deal of variety, but it may help you to get a sense of how and why others have justified certain choices…
Dear Giang Van Ngo, Usually a panel of at least members but while presenting the first seminar (about the topic and proposal) audience may comment / suggest. This depends upon the committee members how they consider those suggestions and comments.
With best wishes for your research progress ..continue with sprit .. Good Luck!
Along with what Afaq and Akihiro stated. The number of interviewees depends on the scope of the research. While two or three interviewees may give you data it does not give you the breadth of data that may be required to relate it to your hypothesis.
It is also important that the research can be replicated, which of course builds reliability and validity in your thesis and hypothesis. It is necessary to have an adequate cross section of the given population and it is important to determine which parameters are required for the population you seek based on the requirements from your hypothesis. Are you planning on using face-to-face interviewing or maybe an online survey? This also would determine the number of interviewees that is required and as Akihiro stated the number needs to be manageable. Happy researching!
Ganesh, the quality of research relates to the quantity of the sample from the population. Law of large numbers states the larger the sample size the greater probability that the sample mean will be close to the population mean. This increases the reliability of the study. Please note that a large sample could be as few as 30 participants depending on the research study. In this way, normalizing the distribution of the sample is more likely to represent the whole of the population a researcher is generalizing too. Thus, from my perspective, the quality of the input directly relates to the quantity of participants and the outcome of the study.
Would you explain in greater detail what you mean in #4? How does compromising quantity (to any extent) increase the quality of a study? I may be misunderstanding what is meant.
Asking the question "How many interviews is enough for qualitative research" is in fact a quantitative question and is probably not the type of question to ask when doing qualitative research. Depending on what you are doing one interview may be enough and may be worth 20 surveys. You will not achieve generalisability by one interview - but who is to say you will achieve it my 3 or 6 or 10? The more important question is what are you trying to achieve by the research?
Qualitative research does not subscribe to reliability and generalizability of quantitative study as most commentors here have advocated. This stems from differences in epistemological and ontological orientations that underlie both approaches. Qualitative research hinges towards social construtivist paradigm and therefore there is no reliability and generalizability in the number of people to be interviewed for study. In qualitative studies, the researcher continues to recruit and interview as many as possible until, the researcher felt saturation point has been reached -where nothing new is coming from the participants again. This is the point where he or she will stop. The number of interview could be start from 10. The way qualitative research is evaluated is different from that of quantittaive research, and standard of assessment is different, and this is one of the criteria.
The question is whether you would need interviews at all for qualitative research. If you are interested in qualities of some phenomena, or of some issue, you may as well do data acquistion through observations, interpretations, or ethnography. When still relying on interviews you usually rely on the saturation principle and plausibility than on replicability and statistical tests.
Asked how many interviews would be necessary, my thesis advisor would say, "Enough." That's actually a pretty good answer, though it may sound facetious. When your participants stop telling you something new about your field of inquiry, then you have enough participants.
The nature of your research is also important. If you're studying well-known phenomena as a confirmation or marginal extension of existing research, then a _larger_ number of participants may be required before you acquire something interesting to discuss. On the other hand, a brand-new line of inquiry may allow you a smaller number of participants -- since each participant is more likely to add to the core conceptual categories whilst simultaneously providing new codes and categories.
Practically speaking, I must agree with Ayodeji Ogunrotifa's suggestion of 10 participants as a starting point. In my own research, I had planned on 20-30, and found myself saturated at 19.
Triangulation of methods is the key with qualitative research to provide validity, trustworthiness is gained by special measures (Guba and Lincoln), reliability is provided with a methodical audit trail and unbiased collection, analsys and interpretation. Yin speaks of 6 interviews, but stresses that the quality of the analysis and interpretation is the key as well as triangulation. Mixed-methods is theway forward I think.
Focus on ensuring that the research aim, objective(s) and interview structure are well designed and justified (in context of the problem). In addition, check that data collection and analysis meets the aim! If you do this you are half-way home - and will know when you have enough data. Designing a non-biased solid qualitative interview is not easy!
Thank you all for your wonderful replies. For my thesis, I have conducted 39 interviews. Yet, when it comes to data crunch, so many things have come out and I found it both a headache and an intellectual game to play. Thanks a lot for your input!
He trabajado con levantamientos grandes de muestra la parte de la información cualitativa, más de corte mixto descriptivo, esto me ha permitido ver los patrones de respuesta y sus diferentes variantes desde la perspectiva nomotética,ponderar la parte de las dimensiones cualitativas considero que ha sido de bastante utilidad, pero también desde la perspectiva ideográfica ha permitido ver situaciones muy específicas con los objetos sujetos de estudio. La información depende mucho desde dónde se está interpretando o sea desde que paradigma interpretativo estás trabajando. Depende efectivamente ¿qué estés investigando?. Pero hacer levantamientos cortos corre el riesgo de estar trabajando con ingenuidad deductiva. Un levantamiento amplio, te permitirá hacer no solo análisis, sino síntesis complejas sobre la subjetividad social.
I always enjoy Kvale's (and later Brinkmann's) response to this sort of question. He/they say with a very straight face (that is how I imagine it whenever I read the comment), with tongues firmly in cheek, that the normal number of interviews in a qualitative research project is 15 +/- 10.
As many have said there is no magic number. Saturation is one point to look for, but some projects may never get there and I have heard enough commentators on this suggest it is its own myth, there is always something else to be found. Resources are an important consideration here. How much time can I afford? How much will it cost to transcribe these interviews, either financially or in terms of my time. Interview as many people as it takes to address your question - adequately.
Definitely agree with Martin Forsey that resources are significant for taking into account of how many interviews are needed. But I am pretty sure that if the researcher has a good master plan for all of this, time (and even money) management for this kind of task will not be too big a problem. Recently, I have come across with the idea of 'earn as you learn' on BBC. So, why not apply this to reduce the financial stress? Cheers.
The answers given are all very helpful. I have also found this paper gives an excellent range of insights.http://eprints.ncrm.ac.uk/2273/4/how_many_interviews.pdf
It is really interesting to read all the responses and I agree that what sample size would be enough to achieve saturation point depends on the subject and scope of research questions. It might not be easy to tell a magical number. However, as we published in the past and the similar way was adopted in BMJ, we can consider 30–35 interviews to have data saturation.
It's very important issue to discuss. In my research I have considered 2 aspects: saturation point and at least 30 interviews. I am advised to have 30 even if I reached to saturation point with less than 30. This means that having 30 interviews is to have adequate number of interviews to show the similarities with quantitative research .
I agree with the responders who felt that sampling should be continued until data saturation has been reached. In my grounded theory study, I felt that I had reached data saturation after interviewing 22 participants, but the number does depend on the type of qualitative research that you are embarking on.
If you find that no new data is coming forth after five interviews, stop it. But if you are still not satisfied with the data, go ahead and conduct more interviews. But for qualitative research, depending on type of research topic, one should not rely on interviews only. The data should be triangulated with other sources/techniques.
In general, surveys are aimed for stakeholders in the form of feedback to support input/s, to get better goals. During this process some times questions are confusing and sometimes the possible answers are not clear. Therefore, atleast more than one interviews are required. Since stakeholders may be at risk therefore, usually, interviews are repeated with considered views of the survey's questionares and possible set answers. As Dr. Mohammad Sajjad Abro pointed in some cases it goes upto 5. Below is an example.
For Census, data collection is carried out by the Interview survey, each household is interviewed every 3 months over 5 calendar quarters that is in total 5 interviews.
Lots of interesting answers here…but I think it is really important to clarify exactly:
- what you mean by 'qualitative' research (very broad term)
- what subject area you are researching
- what methodological traditions and epistemological background you are operating from
….without meaning to sound noncommittal or esoteric, it is important to consider. As many have suggested, reaching a point of data saturation may be a sensible end-point, but not necessarily so (or, indeed, possible) with Case Study research. Additionally, various disciplines, such as the various phenomenologies, have particular views on authenticity, and the place for 'generalisability' which often drives decisions around sample size etc.
Then there is the subject you are researching…it may be so niche, that your sample size is dictated by the number of people who meet the eligibility criteria, or who volunteer to participate…
I would be very cautious about opting for a standard sample size and using minimum criteria that seem reminiscent of positivist values… but again, this really depends on the philosophical assumptions and traditions you and your project are operating from… but there is no single best answer really…
Another thing to consider is what is accepted by the wider research community- look at what has been published in this field, you may find a great deal of variety, but it may help you to get a sense of how and why others have justified certain choices…
Honest and updated knowledge of researches are always get best grades in this world. Depends upon the sample size and the depth of research, the interviewee may be differ. The minimum sample is 29 and the maximum is from 30 to ... is the research grammar in INDIA.
I agree with the comments that illustrate the difficulties of previously established in each study or meet a standard number that directly relates to quality of research with the number interview data collection. One strategy that I find effective is systematically evaluate the findings that address the research question. The findings begin to make sense, they give a consistent answer to the research, are explained the phenomenon under study... Then head to determine the number of interviews will rule: findings (the end) of the study dictate this number so early in the study will not be possible to know. I have developed some studies with a minimum 10 and maximum 20 interviews and I was not able to anticipate this number.
In Portugal, a number of interviews below 10 can not be well accepted by a panel of academic examinations.
I have been struggling with this question also. I came to realize that the sample size in qualitative research depends on a number of issues. Well, 'saturation' depending to what that entails might be one of them. However, from my experience I think it entirely depends on the purpose of the study itself and how it was conceptualized in the first place. The sensitivity of the topic and complexities encountered during fieldwork have a strong impact on the final sample size.
Yes. The number still depends on the analysis of qualitative data follows an inductive or deductive logic. If the logic is inductive may be more difficult to predetermine because we have not established categories.
You might think about the possibility of applying a Delphi technique, especially useful in the case of not having resources to reach a greater number of people. Selection of informants is key (experts). Using this technique, we try to promote agreements between experts.