JRS & B wrote:
BCSnob - I never said anything about the the sample itself being adequate in size.
What I said was the sample was reasonably random and unbiased. Those are two absolute requirements for a valid sample, if you do not want to sample the entire universe of things you are analyzing.
You've totally missed the third requirement for a proper sampling of a population.
A very small sample has very wide confidence intervals; meaning
you really don't have an estimate of the population with the small sample size. Stats indicate we are 95% confident the population rate is within within plus or minus 45 percentage points of your measured 40% rate.
In other words, based upon your 5 dog sample the population rate is between 95% and 0%.
