Dear Mohammed

Mixed methods studies combine qualitative and quantitative components to produce a whole that is superior to the parts. As for the size of the samples, you should be guided by the guidelines that exist for qualitative and quantitative methods. I suggest the following:

Focus group: From 6 to 12 participants per group. At least 2 groups. The ideal is to continue performing groups until saturation, ie until no new interesting information emerges.

Qualitative interviews: Plan at least 12 or 15 interviews. Conduct more interviews until reaching saturation point, ie until new themes or categories no longer appear.

Questionnaire survey: If you want your sample to be representative of the study population (employees), you should calculate the size with the corresponding formula and perhaps readjust it by finite population. Then make a random selection of individuals. Additionally, you should consider the statistical analysis to be performed. Multivariate models require larger samples. A large number of studies use samples between 300 and 500 individuals.

It is important that you be clear about your design of mixed methods. If it is exploratory, sequential, mixed methods approach, you must begin with the qualitative phase and in the quantitative stage test the hypotheses.

Good luck in your research.

Jorge

you are saying that you will use

1) Questionnaire survey

2) Qualitative interviews

3) Focus groups

each one of these methods has its different purpose and method. Researcher should not mix between qualitative and quantitative sample. Also, within, lets say quantitative approach, sample calculation depends on many factors such as the research questions, analysis method, level of measurement for the varialbes, the assumptions for the statistical tests..etc.

There certain rules developed by scholars in determining sample size. For quantitative studies, Hair et all., (2010) suggest a rule of thumb of 10 samples per every measurement variable. For example, if your questionnaire has 25 measurement variables, then the sample size should be 250. This book is an excellent reference for quantitative research.

Good response by Jorge and Ismail for a start. When to stop, Jorge said it clearly and is dealt with in other threads, reaching saturation, where no new ideas or data is generated.

I was part of a similar research using mix methods. We started first with Focus groups (8 members each), we used 4 groups. Outcomes were used to generate a survey questionnaire which was tested for  validation and changes were performed, the sample was University students of 470. The questionnaire had 32 variables (so Ismail's point is taken). However, interviews were added later (10 of them) with a group of influential persons. At the end triangulation was used among the three methods and later on validated y reported research.

For Focus groups and qualitative interviews a smaller sample as per the purpose would be fine, may be around 10, but for questionnaire survey either go by the thumb rules or determine by using formula as mentioned above.

The fact that you are using mixed methods arguably does not directly affect the sample size per se. You should follow the methodological guidelines for the specific methods you use. 

I suggest you look at a research by Marie Luis Small, especially these papers:

Small, Mario L. 2011. “How to Conduct a Mixed Method Study: Recent Trends in a Rapidly Growing Literature.” Annual Review of Sociology 37:55-84

Small, Mario L. 2009. “‘How Many Cases Do I Need?’ On Science and the Logic of Case Selection in Field Based Research.” Ethnography 10(1): 5-38

You can access these here:

https://scholar.harvard.edu/mariosmall/methods

Ekaterina

These are all excellent remarks. I would add my pound to the pot - eventually research is about identifying a phenomena. As we are missing the part about what your research is all about, it might be that instead of numbers, perhaps focus on your research question, and what are you trying to identify. If your research is quite innovative it might be that qualitative method is the key. 

Querido Jorge, 

Interesante pregunta, hago un par de aportes.

Por lo general una investigación basada en el método mixto consta de dos fases realizadas de manera concurrente o secuencial, en donde se debe tener los preceptos epistemológicos de cada uno de los enfoques, sea cualtiativo o cuantitativo. 

En tal sentido, para la fase cualitativa, el concepto cálculo de la muestra pierde sentido, ya que el cálculo del tamaño muestral para lograr una potencia estadística adecuada en el estudio es propio de la epistemología positivista, y la investigación cualitativa emerge desde una epistemología constructivista, en donde la realidad es tan relativa, que con la subjetividad de una persona se puede construir una teoría sustantiva del fenómeno en cuestión, sin embargo, meses después esta realidad puede cambiar, porque para el constructivismo lo existente es dialéctico.

Entonces, el tamaño muestral en la investigación cualitativa no es relevante, lo que sí se debe tener presente, es el concepto saturación de las categorías, que implica la carga subjetiva que aporta cada participante para construir el fenómeno de estudio, lo cual si determina el tamaño de la muestra, pero no previamente al planificar el diseño de investigación como es característico en el enfoque cuantitativo, sino, al construirse la investigación en el interaccionismo simbólico entre investigador y sujeto investigador. De manera que, con tres entrevistas podría ya saturar el fenómeno de estudio o podrían ser más, todo depende de la profundidad subjetiva que busque el investigador. 

En cambio, en la fase cuantitativa de un diseño mixto, el cálculo del tamaño muestral es importante, ya que de éste dependerá que no se cometa errores de tipo I o II de la investigación. Además, en los estudios de comparación de medias mediante t de Student o ANOVA se debe considerar los conceptos tamaño del efecto de las variables, alfa de error que se tolerará en la investigación y la potencia estadística del estudio. Este análisis del tamaño muestral en la comparación de medias se lo puede realizar en el programa gratuito G Power http://www.gpower.hhu.de/  , que permite calcular de forma a priori la potencia estadística del estudio, en base a un cálculo del tamaño de la muestra que debería tener la investigación, ya que en el diseño cuantitativo, a diferencia del cualitativo, el interés es extrapolar los resultados a toda la población. 

Por otro lado, en estudios que busquen análisis multivariantes como los modelos testeados por ecuaciones estructurales o análisis factorial, se debe considerar al menos, a 10-15 personas por cada ítem que posea la escala. 

Un abrazo, 

Carlos

http://www.gpower.hhu.de/

Saludos Carlos. Gracias y un abrazo.

Thank you all for this really useful thread. May I ask Ismail for the reference for Hair et al 2010. I am unable to find a reference supporting the use of 10 samples per measurement variable.

Thanks again

Lizzie

in case sir explanatory sequential design, what should be sample size for 2nd phase? Suppose, we had n=400 in first phase, so in second phase what would be the logically accepted sample size?@ Jorge Cruz-Cárdenas

I am quite interested to read the book referred by @ Ismail Nizam . Could you please give me the link of book or the complete book name.

https://digitalcommons.kennesaw.edu/facpubs/2925/

I believe this is the book cited as Hair et al., 2010

Thank you Chinchu

Hoping you can please guide me.

We have two trial arms: one with continuous chemo treatment and one with intermittent. On the intermittent arm, patients receive chemo until the cancer reduces by 50%. Then it pauses until clinical progression. Outcomes are a quality of life survey (which can be analysed as either a single total or individual questions, say X in number) and survival time. Four-weekly QoL surveys are proposed. I'm assuming analysing this is would take a mixed methods approach (as death would create missing data). I'm trying to calculate the required sample size but getting stuck. Could 10X be a useful start?

in one of my research, a comprehensive statistical analysis was carried out for the HydroGIS tool development. I could prove that the confidence of the 10 samples to detect 87% of the known defects in a system. read at

Conference Paper Selecting a Usability Evaluation User Group -A Case Study th...

large and varies

This is a great thread. Thanks for sharing.

Just that I find it difficult to agree with the idea of continuing with interviews until saturation is achieved. Practically, you don't get to know anything about saturation until the data are analyzed. Was Jorge Cruz-Cardenas suggesting that you first interview a set of interviewees, carry out analysis and then continue with interviews when there is a deadline to be met?

In a mixed method approach how and when do we conduct qualitative research after doing quantitative one?

400 respondents