UncleSam made a bold prediction a month or two ago that Abrams would do at least 10 points worse than Warnock. Prophetic?
The fact that Quinnipiac is the pollster backing me up is making me regret my predictions lol
Anyway let’s all agree to a bet: if Warnock does 5+ points better than Abrams in November, we all agree Quinnipiac is the new gold standard. If the two races are within 2 of each other or if Walker wins at all, then we agree to ban Quinnipiac polls from this forum.
Judging pollsters based on a single election is dumb.
Hell there are plenty of crap pollsters that get treated like gold on this forum because they deliberately oversample Republicans and last cycle reputable polls undermeasured Republicans.
Can you name one example of what you consider to be a 'crap pollster' that gets treated like gold on this forum? In my experience the only polls that get treated like gold are ones that have performed particularly well lately, regardless of partisan skew.
Also not for anything but polls that undermeasure support among one of the two parties they are trying to measure support for can't really be considered 'reputable'...at best they're just wrong but correct course after being wrong.
Trafalgar, who pretty much openly just throws in extra Republicans and doesn’t disclose methodology and everyone on here pretends is a real pollster.
Polling is fundamentally predictive modeling, they are taking a snapshot of a small group of people in a moment in time and using that to extrapolate the larger public and importantly, is predicting how an event that isn’t going to happen for some period of time would happen today. Predictive modeling necessarily includes an error bar. And, it’s important to understand that all elections held a on single day are effectively a single event. Any factors effecting one likely effect all elections that day to some degree. So when you error you are almost always going to error in the same direction, when conditions are unusual multiple pollsters will be wrong in the same direction. What is important is that you have plausible methodology that has been tested for multiple events (notice that most pollsters that aren’t jokes don’t exclusively poll elections) and gets you repeatable results within a reasonable error margin.
A guy that tells you ‘no everyone is wrong, here’s what’s really gonna happen’ and doesn’t show his work is a conman. If he happens to be right once he’s a conman that got lucky.