Memo to the polling companies in both the USA and UK. Perhaps if more of your employees were other than Guardian or New York Times reading urban hipsters then you might not constantly overcount – even if only by a few percentage points – whatever the “progressive liberal” choice is in any given election or referendum.
Like the people at the Heterodox Academy say about the same corrosive phenomenon in the academic social sciences – and without any suggestion of actual fraud – if you spend your time in an ideological and social bubble surrounded by people who think as you do then it must at some level warp your perception of reality. If, in the British context, you perhaps know no-one at all who voted Brexit, or who supports the Conservatives let alone UKIP, or who reads the Daily Mail or Sun or Express – and moreover you actively despise such people – then your view of the world and your judgements about the data that you’re collecting are unlikely to be accurate. You’re “finding” what you expect to find.