An “expert panel” claims it has figured out the polling errors the industry has made over the last several election cycles—all the way back to the Eisenhower election cycles, in fact.
Pollsters oversampled one party or the other.
NSS.
The “experts” missed the real problem, though, or they chose not to mention it (which elision, if that’s what they’re doing, would be consistent with polling behavior).
Regardless of the cause of their oversampling (and this panel suggested a number of them), the pollsters—in every poll—knew they were oversampling one party or the other by the time they’d closed each poll and begun going over their data. The pollsters then made the decision to make no effort to correct for that oversampling, or to say in their published results that they were oversampling and had no reliable method of correcting for that.
They chose, instead, to masquerade their results as valid without caveat of any sort.
That’s the dishonesty of the pollsters.
And this: the number of polls examined may be too small to indicate whether the apparent bias toward Democrats/Progressive-Democrats is valid, but the appearance should be investigated in its own right.