Europa / Europe and the EU + EEA

There is an equal chance that all 800 were statistical outliers as there is a chance 0 were

This does not hold. Most people are not statistical outliers, so it is highly unlikely that 100% are outliers. Likewise, more than 1 in 800 are outliers, so if you had none of them in your sample it would not be representative.

As for the 1 in 20 critique: welcome to social sciences. This is why survey studies need to be pre-registered, hypotheses need to be clearly formalized before being tested, and everything needs to be taken with a grain of salt. For opinion polls, it means that we need to take error margins very seriously, and make all effort we can to ensure representative samples. And it's still hard. We know all that. But your criticism is still nonsensical.