Originally posted by 1905 Giants
View Post
OTOH, If I ask 300 million people a question and that question is phrased in such a way as to be confusing then my results are invalid. Or if I ask 100,000 people and they are not random my results are invalid if I claim that they are random. More counting does not mean more valid.
However if the sample is properly sized and truly random, and the test structured properly, the result can be valid. The only thing you get if you take a larger random sample is a reduced margin of error and perhaps a slightly more accurate estimate of the underlying population. Unless you ask everyone, both measures will still have errors, but if your sampling and testing are done correctly both are valid.
Statistics, again from memory, has no metric that indicates a result is MORE valid. E.g., you may have a 95% CI with a smaller sample or 99% CI with a larger sample. They can both still be equally valid.
Comment