re: #361 jaunte

One problem with polling data is the alacrity with which the results are extended by assumption to cover the opinions of millions of unpolled individuals. The tendency for those who self-select out of the process to have a different pov that those who remain can skew the results quite a bit, too.

As noted, it’s a question of sampling. The 2% accuracy of the poll deduced from the 2000+ respondents is entirely valid - IF the sampling was truly random and drawn from a representative population. Issues raised yesterday regarding apparent biases held by the polling company call into question whether both of these assumptions are valid. If either is wrong, then it is simply impossible to evaluate whether the results can be extended to the population at large, but it’s a safe bet they cannot be.