Dear Betania,

Samling size in qualitative research is always purposive and never has a fixed minimum, maximum or even perfect number of informants. It all depends on your research objectives, your study design and the study area/context your are including as the relevant area to sample from. Representativeness to different contexts or systems is also not a quality criteria in qualitative research - so be carefull to think you will achieve 'representativeness' by just including more informants. The number is not decisive here; it is much more important, whether you think you can cover all relevant viewpoints/health care providers/users etc. to make a comprehensive analysis of the situation.

My immidiate response when I see that you want to analyse croos-program wide and cross-health system wide, is that it sounds like a very small sample to only have 6 informants. I would always recommend to go deeper and more narrow, than going wider and more thinly-spread. Your analysis will not be able to say something profound about all 3 programs and systems based on 6 informants in each.

Consider to focus and narrow in; choose 1 program in three systems, or 3 programs and compare them in 1 system, or even just do 1 program or 1 system, instead of doing a 'all in one study'. And choose your informants strategically, so they represent all relevant care roles/systems levels etc. to make you able to do a comprehensive analysis.

The strenght of the qualitative research is the depth - not the width!

regards,

Thilde

Dear Betania,

Samling size in qualitative research is always purposive and never has a fixed minimum, maximum or even perfect number of informants. It all depends on your research objectives, your study design and the study area/context your are including as the relevant area to sample from. Representativeness to different contexts or systems is also not a quality criteria in qualitative research - so be carefull to think you will achieve 'representativeness' by just including more informants. The number is not decisive here; it is much more important, whether you think you can cover all relevant viewpoints/health care providers/users etc. to make a comprehensive analysis of the situation.

My immidiate response when I see that you want to analyse croos-program wide and cross-health system wide, is that it sounds like a very small sample to only have 6 informants. I would always recommend to go deeper and more narrow, than going wider and more thinly-spread. Your analysis will not be able to say something profound about all 3 programs and systems based on 6 informants in each.

Consider to focus and narrow in; choose 1 program in three systems, or 3 programs and compare them in 1 system, or even just do 1 program or 1 system, instead of doing a 'all in one study'. And choose your informants strategically, so they represent all relevant care roles/systems levels etc. to make you able to do a comprehensive analysis.

The strenght of the qualitative research is the depth - not the width!

regards,

Thilde

Dear Betania, You have chosen a good topic of qualitative research. Congratulations. So far your problem is concerned I agree with all previous posts but would like to add something more as per my knowledge. I think first of all decide clearcut research objectives, specify scope of your work and delimit your work. Thid will provide you the guidelines that what kind of tools will you be using. Even within the interview technique you will haveto decide that what kind of interview technique will you be implementing? Closed, Partially closed or totally open ended? Pilot testing with few selected subjects will help you to decide your curse and time of interview which will ultimately help you to decide that how long time you can spend for each programme in each system. Based on that calculation and prefixed objectives decide that how much sample will be able toprovide you ample data. Like that also qualitative research is subjective. So everything depends. To my opinion in addition you should also consider the time duration of the treatment the subjects were given in all the three programmes which will also help to decide no. of sample subjects. To my mind you should also include participant observation and a check list too to bring out with better redults. All the Best.

Thanks everyone for your comments. The problem I have is that I cannot focus on just one program or just one health system (as per Thilde's very good suggestions, thanks so much), because I am being paid to study all 3 programs in all three health systems. Something that occurred to me after reading the comments (most especially Thilde's but also Dipika's) is to "go deep and narrow" by focusing the interview guide on only one aspect or one element of care for each program. This way, that element of care would become (in a way) a case study of implementation of guidelines, thus permitting (I hope) more in-depth (less superficial) and useful data. For example, it occurred to me to study just cesareans in the maternal health program (as this is a very important issue in our healthcare system, we have a high rate of cesareans in Mexico, where I work; although of course the temptation to study other topics is great: prenatal care or promotion of breast-feeding, for example). This way, although I would not be able to explore in general how maternal care guidelines are implemented, I could get an in-depth look at how guidelines for a single element of healthcare is implemented. Any comments would also be appreciated.

Important to note is that although I do not have time (the agency paying for the study imposes a short timeline) to do observation or shadow studies, this study is part of a larger project which will include quantitative questionnaires applied to providers in all 3 programs, check-lists about infrastructure and supplies available for each program, and a qualitative component focusing on healthcare users' experiences of care in the 3 programs. However, I still feel "my" component (qualitative study on providers) must in some sense stand alone if it is to contribute quality findings.

Hi folks, just reading through the posts it seems to me that the question of numbers is not the really important one - although to some it is. For me the issue is more about understanding how health care workers adapt to change and how changes impact on their work, their lives and their professional identities. Hence mappting the impact of change on the lives of a small sample of workers will help to evoke a deep understading and ensure that the change can be accommodated and implenented effectviely for the benefit of consumers and hopefully for staff too.The constraints of time etc highlight the need to clearly identify the critical questions you wish to ask to achive the level of understadning in this particuar setting. A total of 53 interviews in this context sounds rather ambitious to me Kind regards. Paull Morrison

Postings are great, covering a lot of issues in sampling in qualitative research. Lots more to say, of course, but just want to add that because so many donors (and researchers) used to quantitative research want to support or do qualitative research they demand interview numbers and budgets. We (ethnographers and other qualitative researchers) have been complicit in letting this go uncontested (because basically if we say we don't know how many interviews they think us incompetent, when in fact, it is the reverse). This complicitness makes sense when we're doing rapid appraisals with quantitative and qualitative goals and collecting dozens or more interviews. When we're doing soundings, or pilot research, we can have many fewer, depending on the topic. The more like a language or a specific event the topic is, the fewer informants we need. When it is the experience of different implementation systems as you propose, I cannot imagine how 6 would serve. How could those individuals have had enough experience of all the different operational features... Another way to do this is to explore with some few individuals in each system what the different or interesting parameters are, and then develop a plan to collect information about those parameters with a second and larger sample with more directed questions. With respect to the difficult of analyzing 18 interviews, you pretty much should be able to manage that in your head, especially since you're focusing on some specific bits of each interview, not everything that everyone said. When we think we have found something interesting, of course, we need some confirmation, either from additional interviews or through a preponderance of evidence. As you can gather, my take on this is that, except for budgeting, sampling and sample size are almost always a red herring, it's your findings that count.

I do agree with Steffan Igor Ayora Diaz :the number of informants or participants are not important in qualitative researches but it is important that you reach to data saturation even with 8 participants

Regards

marjan

phd student in nursing education

Patton's book Qualitative Research Methods has good advise on p.244-245.

I think that 6 interviews are to many. You have to get tha facts for your analisys, that is how the programs are applied. Thath´s for you need to focus on interviewing the correct person to get the right answers. I suggest that you can use a SWOT (Strengths, Weaknesses, Opportunities, and Threats) analisys.

That´s a cualitative approach you can use to evaluate a programa status of operation, that is to compare the facs whith the program must be.

Dear Betania, i'm happy to read your topic. i

m presently working on a topic healthcare measures for wards of female prisoners. Shall like to keep in touch with you.

Dear Betania, i would suggest you that many a times interview method proves very vague. A researcher has to go beyond the visible things... try to find fact behind a fact.

Dear Betania, There is a good article published by the National Centre for Research Methods that addresses precisely this issue ('How many qualitative interviews is enough?'). The lead authors are Sarah Baker and Rosalind Edwards.

Thanks to everyone for their thoughtful comments (and for the recommended bibliography). In case anyone is curious, what I have decided to do with the sample size is try to "go deep" or focus on gathering dense material instead of fixating on how many interviews to do. Since the funder requires me to study the three programs and three health institutions or systems, I can't reduce the breadth that way. So I have decided to:

1.) focus only on the implementation of one specific aspect of care for each program (how cesareans are decided upon in maternal healthcare; how condoms are distributed and the counseling that accompanies this for adolescent healthcare; how specialized counseling is provided to women in violent partner relationships (the first session, provided after immediate physical health issues have been resolved).

2.) do case studies of each program, collecting a variety of types of information from a variety of sources for a very small number of cases (a case will be the program as it functions in one specific clinic or hospital), including in each case study: interviews with different personnel involved (health center director, person in charge of the program, a variety of professionals who work in the program), check-lists to register availability of materials, equipment or infrastructure, content analysis of program documents (guidelines, etc.), interviews with national-level program directors, and free-listing and quantitative questionnaires which will be applied to a larger sample of personnel working in the program which will be representative at the national level (I am not carrying out the questionnaire study, but will have access to the free-listing and questionnaire data).

Thanks again for all the comments.

Looks like a great plan, even if four case studies is a lot. Buena suerte.

I concur, we qualitative researchers are not much concerned with numbers. The QUALITY of data matters to us a lot. Check Guba & Lincoln (any edition).

i with agree the comment ofMaditobane Lekganyane ·

I agree with you Maditobane, I can also suggest Denzil & Lincon (2000) and Payze (2004)

One way of widening the numberr of participants without increasing the number of interviws, is to do initial interviews then use the data as a basuis for focus group discussions. This can help validate your data but may also elicit some different perspectives. There is no magical number for interviews it depends on whether everyone is saying the same thing.

Peggy Anne Field

Yes saturation helps determineif one has reached the exit point of interviews, but that also depends on the type of study. For instance if the phenomenon under study specifically is based on a fairly small number of population (say between 1- 10) there is no harm in interviewing all of them.

The question raised here is an administrative and management question. "How can I best achieve the purpose and goals of the research project given the time and resources allocated to the project?

The size of the "sample" is not a statistical question, since qualitative research is not focused on measurement of relative distributions.

Qualitative research is "structural" in nature, or should be. The goal is to "square the circle" as they say. Start with an input (a basic question/problem) and the individual or group that has the problem and then follow them through the system of interactions that result in a outcome or the outcome you are looking for. At each step in the research, that is, each event in the process, you want to know the status and role of each participant in the process and how they connect to the previous and next step until you either come to the end of the process or return to the beginning.

How many people you interview will depend on three things: (1) How complex the system is, (2) How many steps there are in the process, and (3) How much time and effort you can put into the project (your budget).

Using Key Informants is the most efficient way to limit the total number of interviews. But you need to know what the structure and process is in order to identify the key informants. To do this, in the context you describe, you can use the formal organizational structure as the starting point. This will achieve three objectives.

(1) You will know who and where the power is in the organization and the system

(2) You can win the support of those in power to support your efforts and provide you with the "official" line.

(3) You will have the "authority" to question and study the organization and system from within the organization and identify the "real" key informants for your particular research goal. [These "real" key informants are those who have a direct hands-on experience with the issue(s), problem(s) and solution(s) affecting the purpose of your study].

I hope this timely and helpful.

La pregunta que surge aquí es una cuestión administrativa y de gestión. "¿Cómo puedo lograr mejor los objetivos y metas del proyecto de investigación, dado el tiempo y los recursos asignados al proyecto?

El tamaño de la "muestra" no es una cuestión estadística, ya que la investigación cualitativa no se centra en la medición de las distribuciones relativas.

La investigación cualitativa es "estructural" en la naturaleza, o debería ser. El objetivo es "cuadratura del círculo", como dicen. Comience con una entrada (una cuestión básica / problema) y la persona o grupo que tiene el problema y luego seguir a través del sistema de interacciones que dan lugar a un resultado o el resultado que busca. En cada etapa de la investigación, es decir, cada evento en el proceso, usted quiere saber la condición y el papel de cada participante en el proceso y cómo se conectan al paso anterior y el siguiente hasta que usted llegue al final del proceso o volver al principio.

¿Cuántas personas que entrevista dependerá de tres cosas: (1) ¿Qué tan complejo es el sistema, (2) ¿Cuántos pasos hay en el proceso, y (3) ¿Cuánto tiempo y esfuerzo que usted puede poner en el proyecto (su presupuesto ).

Partir de informantes clave es la forma más eficaz de limitar el número total de entrevistas. Pero hay que saber cuál es la estructura y el proceso es con el fin de identificar a los informantes clave. Para ello, en el contexto que usted describe, se puede utilizar la estructura formal de la organización como punto de partida. Esto logrará tres objetivos.

(1) Usted sabrá quién y donde el poder está en la organización y el sistema

(2) Usted puede ganar el apoyo de aquellos en el poder para apoyar sus esfuerzos y le proporcionará la línea "oficial".

(3) Usted tiene la "autoridad" a la pregunta y el estudio de la organización y el sistema desde dentro de la organización e identificar los "verdaderos" informantes claves para su objetivo particular de investigación. [Estos "reales" informantes clave son aquellos que tienen una directa experiencia práctica con el tema (s), el problema (s) y la solución (s) que afectan a la finalidad de su estudio].

Espero que esta oportuna y útil.

Saturation of Data matters rather than the sample. If you are getting the same data with the interviews then you end there.

Dear Betania

I agree with Thilde, there is no one fix number of sample for qualitative researchers. however, general tip off point for sample is 'purposive based selection'. Please refer to Silverman (2005) for the sample selection explanation. Silverman (2005) emphasised on special interest as guide for qualitative choice of respondents. Theory of Saturation by Glaser and Strauss also highlighted the on the issue of number of sample for qualitative research. Theory of saturation stated number of sample depends on saturation point of data. Based on my reading, the number of sample is also based on method chosen. For example, if your method is phenomenology, than the number of sampel could be one. And if you chose case study, Yin stated one - four as sample. Otherwise, I would stick to theory of saturation.

Hi Betania,

The guide to sample sizes is the degree of data saturation that is achieved. So you want to have the themes that you wish to comment on continually coming up in your interviews so that your resulting analysis is well rounded and represents a range of dimensions of that experience. How successful that is with a smaller sample size often depends on how homogenous that sample is... so if they come from similar disciplines and work environments and service similar patient populations, you are more likely to have good data saturation. If their work practices and contexts are more diverse, you will probably need to recruit more to be confident that your themes are fully developed.

To me it is a one million dollar question that yet to be answered. In fact this question has been answered one million times but still every one asked about the sample size in qualitative research. My fellow friends, to me, if any one uses the word "Sample" then it is not a qualitative reasearch. The name implies that it is "Quality" not "Quantity". If it is quantity then the "Number" comes into consideration otherwise 'Not". Therefore it all depends on the nature of problem to be invetigated, time, resources and the availability of the specialist qualitative researchers and aslo selection of the "Qualitative Respondents/Participants". Once all set we can get good results.

Peter you may continue interviews other Respondents/Participants if the "quality" and “similarity” of responses from an interview of one individual differs dramatically from others till the responses are alike in a group/category of study participants/respondents of same socio-cultural, socio-economic, religio-political, education and territorial background. These individuals are NOT a "SAMPLE," for they are true or statistically represent a population of respondents. The terms sample, mean, SD, skewed distribution, variability, etc. are used in “Quantitative Research” and NOT in “Qualitative Research” that’s why the results of qualitative research are not taken as the representative of all populations. We study the problem in “Depth” rather in “Breadth” in quantitative research. Nevertheless, in qualitative research, we try to discover something. We seek to uncover the reasons why? May? Or may not be? The data collection strategies used in qualitative research are interactive to discover the natural flow of the events and processes, people's individual and collective social actions, beliefs, thoughts, and perceptions. I am a firm believer that we should not quantify qualitative results.

In order to determine the required samples, researcher also need to establish the research scope, then sampling frame, the criteria of samples just to get to the purposive samples and the core and rooted data. So determining numbers of samples is not the purpose of a qualitative research because generalization of data to the population is not a qualitative characteristics. Pilot testing your question guides for interviews to a few samples determined as mentioned earlier is essential to evaluate these questions effectiveness before going ahead with the interviews. This will ensure you the real data, not really the number of sample. Thus probing to obtain in-depth data is the technique that will get you to deep rooted real data that can become findings to your reseach.

I would like to share my view in regards to the use of the word 'SAMPLE'. I agree to some degree with Shujaat. Based on rheortical assumption, qualitative research should use qualitative wording.In qualitative study, philosophical assumptions are core to support the validity and reliability of data. As rhetorical assumption is part of the philosophical assumptions of research, it is important to use the 'qualitative words'. SAMPLE is common in quantitative research as it is considered 'n', which means a certain number out of population based on quantified means. In Qualitative perspective, we do not select 'n' as we do in quan study. There is no quantified criteria requires but we do select our 'sample' based on certain criteria, which commonly based on purposive sampling (pls refer Silverman, 2005 for detail of criteria selection of respondent). As qualitative has its own ownerhsip qualities and also to support the validity and reliability, it is best if we use the word such as 'respondent', interviewee, subject etc, which is more persona, hence support the rhetorical assumption quality of doing qualitative research. Although it is not wrong to use the word 'sample'.Beauty of qualitative study ...flexibility yet systematic as said by Charmaz and if I am not mistaken Glaser (THE GURU OF GT)

I never get hung up on sample size as long as your methodology explains what you did and why. The small sample only means that you cannot perhaps generalise the results. In these cases I prefer to use Action Research as the umbrella which I feel fits your research.

In-depth interviews with targeted respondents will provide the rich information I need. As long as I'm able to identify the right group of people, perhaps through snow ball method, and able to gather rich sets of information from 'right' sources that supports my research questions, I believe sample size is not an issue. The issue is using the right methodology (and method), which is of course influenced or guided by your research questions and when I am in the field as well. Hope this helps.

Dear Betina,

I do agree with the comment made by Thilde, Zaleha Othman and Zaidi. More importantly in any study, you must be very sure of what you want your study to answer in assisting you to improve the implementation of the guideline (study objective). I would approach the study by interview the key informant in each of the program (qualitative) and combine the finding from the quantitative study (finding from those using the guideline). Triangulation approach will increase the strenght of your study. Further, if I not wrong you would want to compare the effectivness of the implementation. Comparative study is better by quantitative.

Agree triangulation is useful and again fits the AR methodology and is research in practice. This approach gives some more repeatability and validity to your findings

Already, there have been useful suggestions. I think it willmalso be important to have some informal discussioms with a few of the personalities involved in the programmes. This will give you a sense of how the programmes operate, and of course the various facets. Additionally, it will give you first hand insight into key persons you will have to interview eventually. One other key point here is that, this could give a clue on the data collection instruments to be deployed.

On the substantive issue on sample, I believe the idea of 'continuous data gathering till saturation is reached' should be followed. This way, you will be sure you have captured every bit of relevant information you need. Also, in order not to waste resources, make concious attempt to get experienced persons so that youndo not have to intervierw a thousand and one people to get what one experienced person could have provided..

I hope these comments will help in your decisions.

The researcher actively selects the most productive sample to answer the research question. This can involve developing a framework of the variables that might influence an individual's contribution and will be based on the researcher's practical knowledge of the research area, the available literature and evidence from the study itself. This is a more intellectual strategy than the simple demographic stratification of epidemiological studies, though age, gender and social class might be important variables. If the subjects are known to the researcher, they may be stratified according to known public attitudes or beliefs. It may be advantageous to study a broad range

of subjects (maximum variation sample), outliers (deviant sample), subjects who have specific experiences (critical case sample) or subjects with special expertise (key informant sample). Subjects may be able to recommend useful potential candidates for study (snowball sample).

The iterative process of qualitative study design means that samples are usually theory driven to a greater or lesser extent. Theoretical sampling necessitates building interpretative theories from the emerging data and selecting a new sample to examine and elaborate on this theory. In practice, qualitative sampling usually requires a flexible, pragmatic approach. This may be illustrated by my own study of the professional relationship between GPs and specialists (in progress).

A sample of 10 national figures in positions of leadership and responsibility within the profession were chosen. Since the total population of possible key informants is small, this was necessarily a convenience sample, though there was an element of a judgement approach, since efforts were made to ensure that participants came from a range of clinical, academic, managerial and political backgrounds. New themes stopped emerging after about 15 interviews and an acceptable interpretative framework was constructed after 24 interviews—the stage of thematic and theoretical

saturation.

the sample size is not the most relevant in qualitative research. The most important is to target your participants according to your research objectives. You stop to add news participants once you think you have enough data to explain what you were looking for

Generally I agree with the comments posted. One of the common statements about this type of problem is that one interviews to the point where redundancies occur - that is, latter respondents are telling you what earlier respondents have stated - and in many studies this can emerge after about 20 such interviews. However, outliers are not without interest and long tails can occur that are also are not without interest.

I always feel also that one interview is often not enough - that the ethnographic work referred to by Carl earlier really requires immersion in the lives of the respondents.

The type of research to which you refer requires time, and I am wondering if it is a commissioned piece of research that needs to be completed within a short period of time. If this is coupled with a restricted research budget you have real problems of research design and at best I think you could only complete an initial exploratory study - which may not please those commissioning the research (health authorities?).

In addition to interviews I would suggest you spend time observing at the places of service provision as (informed) observation is part of the armoury of the qualitative researcher - coupled with note taking, diaries and statements of both the affective as well as the 'objective'/cognitive.

Try to make sense of the following questions and ask yourself whether 6 individuals are sufficient to explain the study results, interpretations and type of knowledge generated from this study

• Why do you think the participants selected are the most appropriate to provide access to the type of knowledge sought by the study

• How do you justify the selection criteria for selecting the study sample ( N = 6)

• Is the sampling strategy based on a particular theory (emergent or preconceived)?

• How can you explain and justify the issue of selection bias?

• The results of this study can lead to at the most information perceived by 6 individuals. No patterns, no relationships, less chance for a new theory. Is the type of analysis appropriate for the type of study?

• How do we explain the strengths and limitations of this study?

• How can we check the trustworthiness of the data and interpretations obtained from just 6 people?

To me, none of the above issues or questions can be answered if you select just 6 individuals as sample. Therefore, I suggest, even in qualitative study, with little careful selection strategy (meeting the requirements of representativeness) one can answer all the above questions (which are also relevant to quantitative studies). To me, a mix of 50 individuals (representing proportionate gender, age and geography or income characteristics), would be sufficient to answer all the questions above.

Hope this helps.

Vilas

To me the main distinction between qualitative and quantitative research is numbers. I had similar problems when I was doing qualitative research for the first time. Coming from an engineering background and doing qualitative research in construction I just couldn't help keep numbers out of my mind. But once I mastered the trick of not thinking about numbers and just jumped in straight into asking questions and looking at the respondents responses made me realise that in qualitative research numbers really are insignificant. My advise to you would be to please try not to think of numbers but rather keep your interviews going till you reach a point of saturation i.e. when you feel that there is no new more information coming out from the interviews you have conducted. Once you reach the point of saturation that is your final number of interviews. You may be able to reach your point of saturation with 5 interviews or maybe more or even less. I hope this information is of use to you please do let me know if it was helpful. Thanks

I agree with the comments of Thilde Rheinländer, Samling size in qualitative research is always purposive and never has a fixed minimum, maximum or even perfect number of informants. It all depends on your research objectives, your study design and the study area/context your are including as the relevant area to sample from.

Even one person's life history and case study can explain lots of social phenomena. It all depends on the objective or hypothesis of research. What one is trying to learn.

Not sure if this is still relevant seeing the date of your post. Even if it is important to study the program within a unit in your facilities I imagine 6 providers would still be small. In this case they would be key informants in your study who will be speaking to the issues where you are posing your questions.

Dear Colleague

There are different kinds of qualitative interviews depending on their structure. If you take your health care providers who implement the national guidelines as your key informants, then you need to conduct key informants interviews. What I understand the ideal sample size for key informants' interviews is between 15-35 informants.

An online publication where the experts answer to the question of N in qualitative research: http://eprints.ncrm.ac.uk/2273/4/how_many_interviews.pdf

Hi, many great responses here, and I was going to say, if its a good one you would only need one interview. I wondered why you had chosen interview as the design? If you chose it - ethnography would help tremendously here. A couple of scoping interviews, then some detailed observation, and then some more post obs interviews building an enquiry from what you have seen. Healthcare providers have many reasons to agree to interviews, to promote, rationalise perceived failings, encourage support, and possibly get info from lower down on reality, so confidentiality/anonymity is important. I'm guessing here - but maybe you want to know how things are actually delivered ? If so - go to where this happens- include the givers, the receivers and the 'official' view... hope this helps

Hi, folks. I found many interesting views here. But, as someone noted above I think interview combined with observation would be a good approach for this kind of study with health care providers. This combination gives an opportunity for data triangulation and analyze the issues from different different perspectives.

It's quantitative but look up Kimball Romeny and William Batchelder on Consesus Analysis, which statistically establishes that if you have about 30 informants, they will be a pretty good representation of shared cultural knowledge. The trick is in selecting the informants in a stratified way. Good luck!

If you have three different sample levels, that would work out as well. You have only 50 samples. That will work well for qualitative method but you need to work a lot on the literature that carry all the ingredients you want to cover from the 50 samples. You can also get secondary data to back up your findings.

Hi. I agree with Thilde and Paul, specially the advices given by Thilde. You have a sort of "quantitative" focus over a qualitative objective. In my opinion, you could explore, firstly, with qualitative focus (few interviews), to delimitate the question and, secondly, to make a quantitative survey... ¿?

The single case could give you much of what you are looking for. I suggest you investigate Q methodology developed by William Stephenson - see http://qmethod.org/about

Hello Betania,

In qualitative research the number of respondents to be interviewed depends on your research objectives and hypotheses already formulated for your research. Qualitative research has to do with the quality of the data obtained and not quantity. In quantitative research a researcher might be inteterested in sending questionaires to hundreds of respondents. This is unlike qualitative research. You need to ask yourself about the quality of the information the respondents will give you. The stakeholders in your field of research will be most important. For instance if you want to conduct research in the area of crime prevention then the stakeholders or role players in the criminal justice will be critical. Also, in the area of health resaerch, it will be difficult to justify interviewing of mechanics of fashion designers for the purpose of data gathering.... Wishing you all the best in your research

According to Albrecht et al. (1993) and Wilkinson (2003), you can consider focus groups of qualitative research ranging from 6-12 participants.

I think 8...check out Creswell (2009).

It all depends on your research design. But as you have already described and seeing that you are pooling about 54 persons into the sample, then a focus group discussion (FGD) approach as suggested by Bijay Kandel above may be used for it to be manageable. Yet you may also discover that some informants may be important than others. Therefore, you may need to identify significant key informants and conduct interviews with them, while you apply FGD for the rest either in uniform or mixed groups.

if u want to look at health care provision more deeply, it doesn't matter whether you have 6 or more. but more is impractical for qualitative research unless you have assistants to help you with the in-depth interviews. but even if you have, you need to train them, too. i agree with Thilde--choose a program for each system to focus on. all three programs seem interesting to look at, but quite huge if you have to study all three . qualitative research is very time consuming.

I agree with much that has already been said. The issues are very clearly identified in "Qualitative Case Study Methodology: Study Design and

Implementation for Novice Researchers", Baxter & Jack (2008). I have attached a copy , in case you wish to follow this reference up.

I will just add a little more with the opinions expressed here by different experts. Considering the situation, I think you can consider about deploying key-informant interview (KII) technique only to understand the phenomena. As I have understood, you will mostly require expert opinions regarding health systems (from key informants), contrasted to lived/personal/individual experiences (in-depth interviews). Purposeful (critical case sampling strategy) can be utilized, which many scholars advocate as one of the most useful purposeful sampling strategies in situations where time are resource are limited (Given, 2008; Marshall, 1996; Patton, 2002). This will also allow you to develop logical generalizations from the rich evidences through studying a few cases in depth. For the proposed sampling strategy (and if the key-informants are chosen carefully), you will possibly reach the point of ‘saturation’ and ‘redundancy’ with a very limited sample size.

The recommended number is about 6. See for instance

Holloway, I., & Wheeler, S. (2010). Qualitative research in nursing and healthcare (3rd ed.). Oxford: Wiley-Blackwell.

Morgan, D. L. (1998). Planning focus groups (The Focus Group Kit). Vol. 2. Thousand Oaks, CA: Sage.

Sandelowski, M. (1995). Sample size in qualitative research. Research in Nursing and Health, 18, 179-183.

Querida Betania, em pesquisa qualitativa, não há exatamente um número certo. Acredito que 6 seja um bom número. O mais importante é ir atrás dos sujeitos mais representativos dentro do seu grupo, em acordo com o que você busca, sua hipótese, suas perguntas para a questão a ser verificada. A profundidade será dada nas entrevistas. Particularmente trabalho com pequisa qualitativa, e método etnográfico, onde a vivência no grupo, com o grupo, dá profundidade. Há também a intuição do pesquisador. Você tem que ser e ter um bom filtro, boas leituras na área de antropologia e saúde, antes e durante o trabalho de campo. Caso contrário você pode correr o risco de se perder. Quando se vai a campo, questões novas vão aparecendo. Quando se está na observação participante, você sente um pouco as questões "na pele", e isso faz toda a diferença. Mas, mesmo não sendo uma etnografia, poderá fazer entrevistas abertas, com roteiro memorizado (para não se perder), e depois ir filtrando o que é de interesse para sua questão - análise de dados. Eu sou uma entusiasta da pesquisa qualitativa em saúde! Boa pesquisa! Bom trabalho, muitas inspirações! Um abraço.

Para acrescentar: trabalho, em geral, com cerca de 20 entrevistas. Mas tudo depende do seu objetivo, seu contexto. Concordo com a maioria das respostas dos colegas e as sugestões são boas.

I will make a comment here about saturation. In qualitative research data collection and data analysis can occur simulataneously, and perhaps this may be the best method of ensuring you have 'enough' people or information to inform your research question(s).

Well, as a Finnish methodologist, professor Pertti Töttö, writes in his new book. it's rather hard to define this "saturation point". What if there would be some new ideas when collecting extra data after the "saturation point". According to him, many qualitative researchers use only "conveniance sampling" instead. - As Popper asked: Will there also exist black swans? :-)

Here come an intriguing comment. What is meant by "collecting extra data" after the saturation point? A saturation point is basically a point at which the data become redundant or start to repeat itself. In another word, it is a point in the data collection process when new data is not emerging except those having similar ideas with the previously collected ones. In fact, it is difficult to figure out the data saturation point unless both data collection and data analysis processes proceed concomitantly.

How do we know the "point in the data collection process when new data is not emerging except those having similar ideas with the previously collected"? E. g. in interviews, how can I be absolutely sure that after a point in data collection there won't be any new ideas? I think I can't be sure unless somebody defines the saturation point adequately.

One will never be sure if the saturation point is already reached. How will one know that there is no "new" data that will be discovered if one will go on gathering data. But the data gathering has to be predefined subject to statistical and logistical considerations. These will define when to end data gathering whether saturation point is reached or not.

The smallest possible is unboubtedly one.

It depends upon the question and the purpose of the study. You only needed one in A L Kroeber's study of Ishi who was the last of his Indian tribe, the Yahi of California in (1916). For a political study of Republicans and Democrats at least two. Etc. For a study of 911 you might want at least 100 drawn from people who played the various roles during and after the event.There is no rule except those imposed by the questions you are asking and the number and quality of responses you need to be comfortable with the answer (outcome).

Of course, it's sometimes possible to collect a total sampling; at least when studying small groups or organizations in intend to make only cautious generalizations beyond the object being studied. As David Silverman once wrote: the results of qualitative inquiries can at minimum generate testable hypotheses.

After re reading your description I think you are confusing methodology with the research objective. You want to make the simplest sample size. But you want to use the program as your unit of observation. You would be asking such questions as: How large, complex, and different are the 3 programs you want to study? Are these the same or different programs based on the problem they are addressing? Who would be your informants? What is their status and role within the organizations? What role will the interview play in your research -- is it the only data or is the interview to confirm information and hypothesis derived from other sources before or after the interview? What are the similarities and differences between the programs? Is this difference caused by the problem they are treating? The similarities due to the funding or organizational sources supporting the programs?

As I pointed out earlier -- the question dictates the method and the sample size. Otherwise all you are doing is a journalistic exercise two sources from each agency (program) is a journalistic methodology.which tells a story but is not science. History yes, science no!. Qualitative research is designed to produce as full a picture of the subject as is practical -- the structure and function of the programs. The qualitative study produces the picture or map for later quantitative studies because it identifies what there is and can be measured.

As for the number you ask for -- Again I say -- It depends on the question(s) you want to answer.

Dear Betania,

the interesting question you posed is deeply rooted to the notion of the philosophical paradigm. If this project is a part of a phd thesis, then you will need to consider your philosophical positioning. If this is an interpretive or phenomenological study numbers are not an issue... it's about offering an understanding and reaching depth through thick description and polysemy. If your approach is more structure -what Prasad & Prasad, 2005- label "qualitative positivism" you may consider the issue of saturation, i.e. stopping when you identify emerging patterns in your data.

The charm of qualitative research is that it is dynamic and flexible and I would try to avoid deciding the number of my interviews a priori. In addition, sampling is an evolving process that includes multiple decisions. It also has a temporal aspect - you can stick to the same respondents and collect interviews at different points of time.

In any case, I would go through the Handbook of Qualitative Interviewing published by Sage. Good luck!

dear friends

sampling size in quantitative research is not fixed. the smallest sample size may be depends on the population. If it is homogenity in nature less than 30 can be taken and we can analyse with the help of t test.

Dear Betania

I think 6 would be a good number if you pick your six informants randomly- and not from one sort of population only. To do justice to those six can take several months of work in a qualitative study. So do not worry about huge figures like 54 because you are not looking at numbers but individuals here. I am currently working with a sample size of ten and it is a huge thing-

You would be able to distill the outcomes of the healthcare if your interviews are careful in covering all issues that you want to look at. The best part of qualitative research is that it is not just a one-time interaction you have with the informants. You can go back to them a couple of months later and check the validity or progress of your earlier recordings.

We have to remember it is the depth we are focusing on and not the breadth. Also because people are always changing, what is an affirmative today can become a negative tomorrow. But since your area is maternal health, it is a somewhat more stable sort of area than certain other areas. But since you are doing a form of assessment of a healthcare system, it makes sense to pick up your six informants relatively randomly and from different income groups too. Best wishes

Dear Betania,

sampling methods in qualitative research are not as critical as in quantitative research- in any case, if you have to analyze 54 interviews, the greatest you could do is to analyse them all- if it is impossible, and you want to select a sample, you could consider a stratified sampling method (stratification should be done based on basic demographic characteristics of your survey population)- Good luck.

Hi I believe qualitative research is based on sampling a percentage representing the broadest view of the unrepresented in a study. regardless of whatever used to collect data to support the purpose, my view is that you must deal subject by subject because if you want to use 6 people for breastfeeding, same people with represent the one in maternal care. In brief, since number informant is purposive and unlimiting in total for balanced results.

Dear Betania,

I would be glad to know whether the healthcare systems in your study are different healthcare systems within a specific country or whether they are different national (countries') health systems. This, I think is an important piece of information in understanding and clarifying your research processes and in giving an informed opinion concerning the questions you have posed.

The main purpose of qualitative research is not to facilitate making of general statements and discover universal laws, although this is not precluded. On the contrary, qualitative research, among other things, seeks to provide in-depth understanding of social actions, institutions, events, experiences and opinions. As other commentators have said, there are no fast rules regarding the sample size and no requirement for representativeness. The decisions concerning sample size are best guided by the research questions, the resource capacity and time considerations.

In your case, every healthcare system and health care provider in your study most likely operate under different circumstances or conditions (available infrastructure, organizational climate and training status of personnel). These circumstantial differences will in all likelihood limit the extent to which you can make comprehensive or general systemic assessment on how the programs work. For a researcher to make a general statement concerning how things are, the sample should not only be randomly selected but should also be representative so that the heterogeneity of a population is taken into account. Representativeness and random selection of a sample guarantee valid, accurate and reliable generalizations.

Arising from the foregoing, with the results you generate, you will be able to describe how the specific programs implemented by specific providers work although certain common trends that cut across the providers will likely emerge.

The insights thus obtained and the common patterns that emerge in your analysis may have some implications for what the case in other facilities may be. This, however, depends on the extent to which the circumstances under which the other providers operate are similar. In other words, if the operating environment is similar, the findings may be applicable to these other cases as well.

7 respondents is fulfill your objectives. Researcher has to take in depth informations from respondents.

wonderful comments here and all true. i must admit that there is no minimum or maximum in qualitative research but you will definitely need Representativeness. whats ur total sample size? regional, national, international, global? depends also on the country and context that you are collecting your data. in Ghana for instance, all eg. midwives should have similar training to be midwives so u have a homogeneous environment to sample from. also the length of service also can be a criteria to select so if i take only midwives with 5 years of working experience at the position that i want to interview, you will have reduced your total sample population. your selection criteria must take into consideration alot to reduce your total sample population. the result is that, u shorten your sample population before u take ur sample size. in addition, the length of the questionnaire can help you decide because if u make it short and in-depth for specific categories of people, you will be able to collect all indepth information that u need. its all dependent on the available information you have on your sample population. in the developing worlds, you may not have all that information so ideally,you would choose a district or region as a case study and then further trim down your sample population. the reality is that, conditions on the field are not always as you plan it. you may get there and realize that you only have six key informants on each of these programs or systems you wanna analyse so you take the available number and justify during your write up or analyses. lastly, if 54 interviews are to be analysed manually, it could take you months so if you use QRS Nvivo or Atlas Ti, you could really analyse all the interviews are a good sample considering key characteristics. good luck

There seems to be some confusion here between what is Qualitative and Quantitative research. Qualitative means quality: What are we talking about and how does it look? Quantitative means quantity: How large is the thing we are measuring? How do the parts relate to one another in terms of size?

Think of the Qualitative study defining the universe while the Quantitative study measures the universe through statistical measurements. If you don't know or define the universe your statistics are worthless and meaningless.

Also think about data "saturation" - this is something I have heard discussed amongst researchers conducting studies amongst healthcare providers..

6

Do not understand how the question has an answer! With the information provided it is impossible to respond!

I do not think you can set a sample size in advance. It depends on what you find in primary interviews. Is there agreement or disagreement among your respondents. You may want to think about focus groups to increase your "sample" based on your findings in the first interviews. Peggy Anne field

In qualitative research methods you cannot pre-set sample size, but the sample size is determined by reaching theoretical saturation; whereby the interviews are no longer generating any new information

The size of your sample can be as small as one if you are doing phenomenology. Creswell provides a good overview of the different approaches to qualitative research, and explains the differences between them in terms of how much data is usually collected, how it is analysed etc across the different approaches. This book is invaluable. http://www.amazon.com/Qualitative-Inquiry-Research-Design-Approaches/dp/1412916070

Like any other in-depth interview sample: As much as your information is get saturated. You better stop If you see that the late two or three respondents do not give any new information.

I would suggest you try focus group discussions too and remember that qualitative data is not about how big the sample is but rather how in-depth you try to understand a phenomenon.

I agree with you Job. It is about the understanding of the phenomenon in question much more than the sample size considered quantitatively. A focus group discussion may help you assess this.

Even in qualitative research, one need to specify the "sample" size (convenience or purposeful sampling) and justify the reasons for the same. However, in order to adjust for "saturation of information" one may need to give a range (example 4-6 interviews) . In your case, if you wish to interview one male and female health provider from each healthcare system x three programs, the sample size you have proposed is adequate enough. I do not see much problem in analyzing the qualitative interview data from 54 providers (18 per program per health system) provided you are able to obtain richer and in-depth information on health providers experiences.

The number that can make you reach the semantic saturation

Paul - useful paper in your link - will point my students to it.

Thanks Paul. Quite some useful information. Will share this with my students.

@Jenine. Agreed Creswell's book is a must read!