Quote:
Originally Posted by MarchHare
Gotta be an outlier, IMO. Remember that Nanos uses a three-day rolling average (1200 sample size total, 400 per day for three days), so to see that much movement from the previous day suggests to me that the 400 people they randomly interviewed Friday are more anti-Conservative than their usual sample.
That said, I certainly hope this is not just an aberration but a refection of an actual decline in CPC support.
|
Not necessarily. In a rolling poll, the movement each day is caused by two numbers: the new number that gets added to the poll and the old one that gets taken off. Look at the last four days for the Tories:
32.1 - 31.6 - 31 - 28.6.
That 32.1 was their high-water mark since the beginning of the month. So this swing is probably the combination of a bad polling result for the Tories (but not necessarily much worse than the previous two days), plus their best day of the month getting removed from the 3 day average. So the swing merely tells us that the new poll was significantly worse for the Tories than the one three days earlier, which itself might have been a slight outlier in the Tories' favour.
It's certainly possible to interpret the CPC Nanos numbers this month as the Tories being in steady, gradual decline with an outlier in their favour on the Oct 6th result. We'll see how much of a bounce-back there is 3 days from now, which will tell us whether this most recent one was truly an outlier in the other direction.