I am reviewing the methodology for sample sizes for surveys that are mailed out to credit union members on a monthly basis. Please forgive me for not using proper formulas. Any advice is appreciated.
Total population is 57,200 members, but not all members perform transactions each month. There are 25 branches, some are very small.
Current methodology to figure out sample sizes for each branch:
# transactions at each branch / total # transactions for all branches * 1,200 for each branch * 15% = # surveys sent out to each branch on a monthly basis
Using this method, the number of survey returns meets +/- 5 margin or error within 95% confidence level for the total population.
Is this method acceptable to determine the statistically valid sample sizes for each branch based on number of transactions?
Some branches are very small, and many members are moving toward online banking. As few as 7 surveys might be mailed to represent a smaller branch.
Any advice is appreciated! Thanks, Lorie