From a purely Christian standpoint, we know we can't predict the future. But with the Enlightenment, and the elevation of science from a tool to increase our understanding of the world around us to Deity, we've gone down the path of convincing ourselves that Science, and its handmaiden, Mathematics, have the capacity to tell us what they manifestly cannot: the future. In public policy, this phenomenon infects EPA rule-making that predicts the numbers of lives saved by regulations (really? Care to share some names of lives you've saved this year?) and, most notoriously, it is the soul of the anthropogenic global warming nonsense. In politics, it is reveals itself in our slavish devotion to polling.
After Super Tuesday I did an overview of polls so it is only fitting that after the March 5 primaries we do it again.
I don't even know what you call a poll that misses the winner by 20-30 points. A wild-assed guess? A brain-fart?
There was only one poll conducted in Kentucky since June. If Western Kentucky had been smart it would have made it zero. The Trump number was good but they overstated Rubio's support by six points and understated Cruz's by 17 points. They got the winner right but the order of finish and actual results grossly wrong. This rates as an uninspired guess.
This is one of those cases where averaging bad polls makes the bad polls look not quite as bad. The average overstates Trump at the upward bounds of the margin of error and misses Ted Cruz by 11 points. This is not vaguely accurate and unless it is at least vaguely accurate it is useless. The polls that make up the average are worse.
There was no polling in Maine. This is probably one of the best polling decisions made this election cycle.
I expect all kinds of kvetching on debate performance, closed primaries, etc., etc. None of this explains the polls. Most of the Super Tuesday primaries were open primaries and they also followed a bad debate performance by Trump. Trump's strength was overstated by about four points and Cruz was understated outside the margin of error. Yesterday were closed primaries that followed two bad debate performances by Trump. I'm not convinced that debates have a large impact outside the relatively small group of people who obsess on politics. If they did, though, Kansas and Louisiana had polls that would have captured the impact of both events. They didn't. What the problem points to is that the polling companies don't have the vaguest notion of who composes the GOP primary electorate this year and are constructing models that favor Trump. While those models are merely bad in open primary states they are ludicrous in closed primary states.