Well, I'd say you can rather fool people into believing that because they don't understand that you were actually proving something completely different.

But that's why checks and balances are so important in science (like everywhere). When we publish something, it'll have to go through a peer-review process, ie. a paper will be reviewed by a group of selected referees. Whilst for e.g. empirical papers that cannot imply an actual check of the estimations, it implies a plausibility check. But once a paper is published, it is open to scientific debate. And believe me, there's a lot of it. Often it involves actual replications of the empirical estimation to verify or falsify results. And often, wrong results are actually getting detected and dismissed that way. A famous example was Reinhart/ Rogoff (2010), a paper arguing for an unsustainability threshold of the debt-to-GDP ratio of an economy, which was dismissed as it couldn't be replicated. Long story short: Economics isn't a perfect science and we can't predict everything and sometimes we're wrong (so are engineers, medical scientists and physicists BTW) but at least in academia there are rigorous quality checks. The problem is only that those making political or business decisions are too often ignoring the academic debate, because it is slow and often doesn't deliver directly applicable results. That this fundamental research is yet useful, unfortunately not too many people appreciate.


They actually are fairly consistent, but obviously vary across the different samples used. The message stays the same: Young, urban, left-wing and more highly educated people are more likely to vote remain. That's not really too surprising, given that the out campaign is mainly driven by an anti-establishment sentiment. But I will have to add that the poll of polls might be a bit outdated: it's from last year. But generally the main result didn't change too much. That is: There tends to be a marginal lead for the in campaign, but it is likely to be within statistical errors. So it probably won't tell us much.

On a more general point on survey data, as empiricists tend to say: "Never trust survey data". It tends to be very biased as subtle changes in wording can have big impacts and also suffers from bias as what people say tends to differ from what they vote for. One can correct for some of the biases but there are obviously limits: If my data is a bit rubbish, I won't get great results. That's why I think there's a point in the criticism of these polls, and particularly the polls being over-quoted. I accept they are the only measure of public opinion we have, but I don't see how having another marginal poll published every other day should enhance anything.

Apologies for another lengthy post and remember

Goodnight!