Um, I'm sorry, but you're wrong. The chance of a random sample of 1,300 being representative of a population of 800,000, within an acceptable RSME, are so small as to be infinitesimal. Even if the 25,000 were demographically representative, the changes of 1,300 RANDOM samples being representative of the whole, within an acceptable RSME, are extremely low.
Add to that the fact that, by definition, the responses are NOT random, but in fact are motivated by factors that skew the results, and the chances of any useful information coming out of this pool are nil. For reference, look up 'selection bias' and 'self selection'. |
