… There is nothing sacrosanct about demographics. These are convenient variables to weight by since there are official statistics (sometimes misleadingly referred to as “the Census”) available on many demographics. However, you can weight on anything, so long as its distribution is known. …
It has long been known that phone samples over-represent people with college or post-graduate degrees and the problem is even worse for most internet panels. This means that samples contain too many people with high levels of news consumption and ideological sophistication. Survey cooperators report significantly higher rates of volunteering and organizational membership. And the list goes on. Some survey methodologists have been in denial about this problem, pretending it doesn’t exist and discouraging efforts to fix it. CONT.
Douglas Rivers, YouGov