"Jon Kyme" <jrk(_at_)merseymail(_dot_)com> wrote:
Hint: You do not need a very large sample to have pretty accurate results
even if the population is very large.
No, but you do need to sample very carefully and have a very good
understanding of the characteristics of the population you're sampling.
... for conclusions with high confidence.
You can always make blanket assumptions about the characteristics of
the population you're sampling. While your level of confidence in the
conclusions goes down, you also have the choice of comparing analyses
made under *different* assumptions.
Pollsters doesn't just interview the first 1000 people that the guy with
the clipboard can detain. The sample they use will often have been selected
from a larger sample based on models of the population demographic.
Exactly. How do they know how to model the population demographic?
Other surveys, samples, and wild-assed guesses.
Even if they did interview the first 1000 people they met on the
street, that data would allow them to set bounds on population
behaviour, based on key assumptions and models.
Alan DeKok.
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg