@unphurlIf you set the population sample as 100k (i.e. no. of...

@unphurl

If you set the population sample as 100k (i.e. no. of MBS stores) and then took the observation count from 43 to 3x that, e.g. 120 stores, in theory you do get a big improvement in the sampling power, reliability of estimates etc. You would still need a sample of closer to 400 to estimate a proportion of 0.5 within a 5% margin of error. With 120 your margin of error is more like 9%. With just 43, it is 15% so not very useful at all.

But arguably, this is also a very stratified sample, not really a single 'population' of stores. The 4 tiers are presumably differentiated in that way for a reason, because of the different consumer markets at each level. So it helps that they would account for this by going to stores at each tier, but to be representative (to within a 95% confidence interval) at each level they would need to visit a similar amount as above at all four levels (i.e. 400x 4 = 1600 stores).

https://www.abs.gov.au/websitedbs/d3310114.nsf/home/sample+size+calculator