If you are looking to carry out a qualitative study, then a good rule of thumb is when you reach "theoretical saturation", i.e., when no new codes, themes, or topics appear to emerge from the data after a number of interviews.

Dear, in my opinion, depends on the total universe (or population) involved.

For example: in a city, 150 professional accountants work and we want to know their perception of working at home in these pandemic times. In the social sciences we use a 90% confidence level and, in this example, 97 answers will be needed to achieve this, which will be the sample of the study.

Then, I forward the link to a calculator that we use to identify the required levels.

Hope this helps!

https://praticaclinica.com.br/anexos/ccolaborativa-calculo-amostral/ccolaborativa-calculo-amostral.php#:~:text=Calcule%20a%20amostra%20aleat%C3%B3ria%20simples,%C3%A9%20aleat%C3%B3ria%2C%20pode%20ocorrer%20tendenciosidade.

If you are looking to carry out a qualitative study, then a good rule of thumb is when you reach "theoretical saturation", i.e., when no new codes, themes, or topics appear to emerge from the data after a number of interviews.

Thanks for the question and the answers above, though seems the one from Osmar Antonio Bonzanini is for quantitative study instead of qualitative study.

Dear Xiangdian Emma Chen

It serves both. This calculation was validated in my doctoral thesis, correctly, with 96.5% certainty. The theme dealt with the accountants' perception of professional training, therefore, in applied social sciences!

I am not familiar with this new designation of something as the "Best Answer" for a question, but in my opinion Osmar Antonio Bonzanini's answer is strictly about quantitative rather than qualitative research. In particular, confidence intervals have anything to do with qualitative research. Instead, they define a standard error around a mean, and to derive a decent estimate of such a mean, and to define the more common 95% confidence interval, would typically take over 200 participants.

Instead, Nicolò Zarotti's suggestion that you use theoretical saturation is much more appropriate for qualitative research.

Fortunately, this topic of qualitative sample size has been raised many times here on RG, so you can use the search function at the top of the page to find these previous discussions.

Thanks brothers. I will consider every input to improve and finish my research well.

Dear all,

As I have already mentioned, the calculator that I shared is used for both types of research. Also, the level of confidence varies according to the approach adopted and the standards of the master's / doctoral program in which the study will be presented. As I stressed in my text, this was my experience.

I thank you for all the messages referring to my name.

Health and happiness with an excellent year of 2021, with family members!

If you follow Creswell for a qualitative enquiry, he guides you about the number of interviews according to the type of study you are conducting. The number can be different for ethnography, grounded theory, or phenomenology etc. ranging from 6 to many interviews. Nicolò Zarotti 's comment considering theoretical saturation is also a valid way to determine your number of participants.

See saturation (Campbell & Stanley, 1963)

Osmar Antonio Bonzanini How does the use of a confidence interval level around a mean possible apply to qualitative research? You have now made this assertion three times without any evidence.

Dear David L Morgan

my research dealt with the existing influences in the construction of accounting science curricula in Brazil, despite the law that says that universities have autonomy to do so. The universe of research was the sample is represented by 436 responses (n = 436), obtained from a population of 980 surveys, representing courses in operation in the country. Aiming to confirm the validity of the sample, the formula used to calculate the proportional error in populations finite (Triola, 2013).

Error = 0.035, which allows to affirm, with 95% reliability, that the proportions found in the survey have a maximum error margin of 3.5%.

In other words, of the 980 coordinators of accounting courses in Brazil at the time of the study, 436 responded to the questionnaire that identified 7 different organizations that influence the format of accounting science curricula. Thus, my thesis showed that, although the National Law of Brazilian Education states that universities have autonomy to build undergraduate curricula, in accounting it no longer has such a prerogative, according to my study. I remain available for whatever is necessary.

David L Morgan

In order to better clarify the formula I used in my study, I took the liberty of sending you a private message by the RG, with a summary and the indicators of the study.

Thank you for all your attention.

The formula you supplied was, as I said in my original posting here, based on means and standard errors, neither of which has any thing to do with qualitative research.

So, I again challenge you to apply these quantitative concepts to qualitative interviews.

Dear David L Morgan

In that case, I may not have had the right understanding and I have put my opinion based on my study and I will not continue the discussion.

My apologies.

Osmar - feel free to continue the discussion. ResearchGate should be a venue to openly discuss research-related issues. However, as David has correctly highlighted, you have appeared to 'method slur' between qualitative and quantitative concepts. That is understandable. Many do - but the paradigm 'rules' have been established for some time now.

In addition to above answers concerning data saturation also follow Guest, Bunce and Johnson (2006). I believe they made a thorough work to clarify this issue. They state that most interviews in qualitative researches reach a saturation point with about 6 to 12 informants or participants as a starting point.

It depends on the target population that your sample is representing.

It varies! Reliable resources suggest though that anywhere from 5 to 50 interviews/participants is adequate and good enough.

Dear Dean Whitehead

I thank you for your words. Occasionally, because we do not completely master the English language, we have an incomplete interpretation. Let's keep learning from everyone!

Cheers.

A minimum of six interviews should be fine

Any quanlitive study must to be approximted by quantitative indicators and later on verified in practice, so I think the both Osmar and Nicolo approaches are valid. The final concytuions come from the practical verification.

The suggestion is that you try and get as many informants as possible. You can also look at the numbers used in studies of similiar nature, when reviewing literature. This can act as a guide as well.

I agree with Dean Whitehead, and I want to say that David L. Morgan is one of the most helpful scholar and researcher in researchgate. Whenever I see a a question (basic or complex) asked by researcher he lends a hand to all researcher, so I think you can continue discussing contentedly .

Osmar - great to see you still contributing. If RG is a community of scholars, then open debate/discussion is exactly what should drive it. We all 'live and learn' and, hopefully, grow. Abdullah is correct. David has always contributed in a selfless manner. We are all capable of being 'direct' - but I see nothing wrong with that.

Mochamad Muslih, it depends on a number of factors. You may read this paper for more insight:

DOI: 10.1177/1525822X05279903

Dean Whitehead

I thank you for your motivating words. Yes, we will continue to follow the topic, I see the excellent contributions!

Cheers.

15

Minimum Sample Size

To ensure that you have saturation you have to go beyond the point of saturation to make sure no new major concepts emerge in the next few interviews or observations. Consequently, 15 as a minimum for most qualitative interview studies works very well when the participants are homogeneous.

This depends on the research theme you are involved in. Other themes may have more people involved in the subject matter and vice-versa. At minimum six could be ideal......

Thank very much for the answers. This is the first time I do qualitative research by myself.

Thank very much to all of your answers. I hope this qualitative research will result in representative conclusion, right conclusion. I am happy to become research gate member.

Qualitative research does not require any sort of statistical sample. Rather, you need to interview enough people until you reach "data saturation." In other words, interview enough so that you would not learn more by interviewing more people. i.e. you have identified all of the major themes.

How many this is really depends on the context in which you are researching, and also on selecting the "right" interviewees who will give you insightful answers.

Mochamad Muslih, em relação a sua pergunta, espero ainda poder contribuir a partir da ideia de que em pesquisa qualitativa, não há preocupação com o número de participantes. Para afirmar isso é preciso identificar se você já conseguiu atender ao seu objetivo de pesquisa com o número de contribuições que obteve desses participantes. Pode ser que um pequeno número apresente contribuições que satisfazem a sua pesquisa.O aspecto a ser considerado é a qualidade dessas contribuições. Por isso, amostra e população, bem como quantidades específicas, não são preocupações próprias de pesquisas qualitativas. O que pode contribuir para a sua pesquisa em termos qualitativos é a escolha de uma metodologia de análises de dados qualitativos adequada ao seu problema de pesquisa.

Thank very much for your answer. I have concluded for the answers I got that I should interview enough people until I achieve data saturation.

Mochamad Muslih excelente conclusão!

Saudações,

A minimum of 15 respondents is required to understand and attain the saturation of responses and purposive sampling can be used to reach 15 informants.

Nivedita Roy where does the idea of a minimum of 15 participants come from? What if the data becomes repetitive after 10 interviews, do you collect five more because of this arbitrary "rule"?