Originally Posted by
orbitmic
Actually sampling error increases disproportionally when moving to smaller n and larger number of values, so there is a much bigger impact between 20 and 40 (definitely not negligible) than between 2000 and 4000 (especially if we add that randomness of the sampling is actually an untenable assumption here), and even bigger 10 and 20 and massive between 2 and 4....
The sampling error (assume no biases) is a function of the sqrt of the sample size. From 2 to 4 or 20 to 40, the difference in both cases is a multiple of sqrt(2). I think we only disagree on the qualitative meaning of negligible here. For 20 people, the error is about 20% and for 40 it is about 15%. For me that is a small difference.
For your F cabin example, other errors such as rounding becomes dominant since you can't have 1.75 chicken, 3.5 pork and 0.75 fish..
For the CW cabin, if 35/40 pre-orders, then you can overload the remaining meals by 200% and still come out ahead of the expected 115% error mark for the entire cabin.