One of the biggest lessons of Trump's candidacy - and there are a lot of lessons - is that people tend to predict what they want to happen anyway. Especially when their predictions are public.
Part of that is because a prediction is often packaged with persuasion. 'The outcome will be X because actions/desires A, B and C are Bad Things which we should not do or desire.' Opposing illegal immigration is undesirable, therefore we cannot argue that it will go over well with the electorate. Whether or not that's actually, you know - true - is entirely a secondary concern. To predict is to persuade, and who the hell wants to persuade people in a way that goes against their interests?
You'd think analysts would be aware of this and avoid this kind of thinking like the plague. But that works under the assumption that both the analysts and the people who hire them are hiring them for their accurate predictions to begin with. You'd think so, right? I mean, when they talk up Nate Silver, they point at his string of successful predictions, don't they? (Well, not anymore - ha ha.)
But when's the last time you heard about an analyst, particularly a political analyst, being fired for their bad predictions? You hear about them being hired, being hot commodities, for successfully predicting this or that. But, at least where media outfits are concerned, heads never tend to roll just because people screwed up forecast after forecast.
It's almost as if the whole point of boosting 'look at their accurate predictions!' is to make them seem more persuasive when they predict, eh?