-
Should
we do focus groups first or a telephone study first?
- How many focus
groups do I need to conduct?
-
What size sample do I need for a telephone study to be statistically sound?
This is one of those cart/horse, chicken/egg questions and depends on your
research questions. For example, if you want to explore issues your
customers are having with your service, focus groups can unearth problems and
opportunities you were not aware of. This information can serve to formulate
specific questions for a telephone study that will tell you to what degree these
issues actually exist in the population.
However, a telephone study using fairly standard customer satisfaction
metrics may yield results for which you want a better sense of "why". I call
this putting meat on the bones of the quantitative with qualitative.
Of course, once you embark on your program of qualitative and quantitative
research, it becomes a cycle.
Generally, you want to conduct a minimum of two to be able to point to
consistencies from group to group and account for anything aberrational in any
one group. However, we generally do 4 or more to account for differences between
Men and Women, shoppers of a store versus non-shoppers, your customers versus
your competitor's customers, differences across geographic locations and the
like.
The key is to start with determining what you want to know from
whom then determining how many groups are needed to get the job done. If the
number of groups needed exceeds the budget, then often compromises must be made
(e.g., do one group in the north and one in the south, but combine Men and Women
in each group).
This depends on several factors—
- What is your desired level of confidence? (95% level is common practice)
- Within that level of confidence, how much error are you willing to accept?
(Generally +/-5 Percentage Points)
- But here's the tricky part...do you want that level of confidence for the
whole study or for each sub-cell, such as Men and Women. If for the
study as a whole, that would require a sample size of n=400; for each
cell of Men and Women, you would need n=400 in each cell for a total of
n=800.
- As is usually the case, there are compromises when one considers the cost
of information versus the value of information.
[Have a question? Send it to
Dave Roberts.]
|
|
