If a 300 people well-conducted poll say:
- 35% Candidate A
- 5% Candidate B
- 60% Undecided
Can something be said about the winning chances of Candidate A? Is there a way to factor in the undecided statistically?
If a 300 people well-conducted poll say:
Can something be said about the winning chances of Candidate A? Is there a way to factor in the undecided statistically?
As long as there's no reason to assume that the 'undecided' group will show a different behaviour in the future, you could divide the votes in Undecided pro rata between options A and B. The estimated totals would then be:
Note that this is a very naive, numbers-only approach. If, for example, this poll was done before an election, and between the poll and the election one of the candidates rescues a three-legged puppy from a burning building, your mileage may vary.