Sindi - there are no hard and fast rules for sample sizes with qualitative approaches - including ethnography. A bit of 'guess work' is required on your part in terms of personal estimation as to what you think would be 'enough'. That may be on the basis of when you think that data may 'saturate' and/or how easy it will be to gather participants. You also state that this phase will be part of your study - so you also have to ask yourself how many participants are needed to progress to the next phase - especially if this is a mixed methods study. A 'manageable amount of participants will generally be between 10-20.

In fact, qualitative methodology often does not use "sampling." Rather, it may be a convenience or purposeful selection process.

The goal in qualitative methodology is not a "random sample" but rather talking to enough people, and the right people, to give you "data saturation." This means that you talked to enough people that you are not missing something important.

I am currently talking to people about something they were involved in 25 years ago. Only eight people were involved in this event 25 years ago. I interviewed four today, which would be enough, but I will probably talk to one or two more, just in case. That will give good saturation.

Think of qualitative research as talking to the right people, the best people, and enough people, in terms of addressing your research questions, not in terms of a plain random sample.

Sindi, you do not start with a fixed number of 'samples' in mind although you probably would have a rough idea how many would give you enough information. You follow flexible sampling technique; continue collecting data as long as you believe that new information is coming. Once you find that there is no new information forthcoming, i.e. you have arrived at data saturation, you stop.

If you're doing ethnographic research for a product that might be based on different groups, you might find the parallel rationality/Choice Wave approach to be useful in determining whether or not they are truly distinct groups and hence should be treated completely separately in marketing and other research/approaches. If two segments of the data set, for example 2 different ethnicities, can satisfy the Choice Wave criteria for being able to be modeled by two orthogonal and hence statistically independent Choice Waves, then they can and should be treated as two completely separate data sets rather than given a dummy variable or the like within the same data set. That has significant implications for determining marketing strategy, product mix, etc. See more at www.choicewaves.com and some of the references mentioned there.

There are three factors that determine the size of the confidence interval for a given confidence level:

• Sample size
• Percentage
• Population size but doubling the sample size does not halve the confidence intervals so if the FMCG products around 100000 so the sample size is 500 and that depend on the percentage of population that suitable for these products