Originally Posted by
hummingbear
If you’ll read my post I referenced the qualities of a “truly random sample”- didn’t say our phone polling system was bias free. Still, if you think APC is more representative than even a poorly randomized phone survey I’ll respectfully disagree.
No, you're still not understanding. A 600-ct, random sample of pilots at United Airlines will not represent the "average" will of the 15000-ish. The only way to eliminate sample bias would be to make it not random, but targeted. And you'd need more than 600 people to do it. The NC isn't designing the poll, but you also have to know what you are asking for the poll to do to get accurate results.