Robert Bidinotto isn’t impressed with polls or pollsters – not even the tracking polls I’m fond of:
The whole point of multi-day tracking polls is that they’re supposed to be more accurate than single-day polls having smaller samples of interviews. Tracking polls are supposed to smooth out quirky single-day fluctuations, determining broad trends more accurately. But when we see even these three-day averages, which incorporate thousands of interviews, jump around this much–and in opposite directions, to boot–what are we to think of the reliability of any of these numbers?
You might reply that at least these two polls are moving closer together: today both show a nearly tied race. But is that true?
Read the whole thing.