Funny thing about the Kyodo polling reports whose numbers I posted yesterday. If you actually add up the numbers in the article, the various shares of the vote add up to 109.5%.
Someone somewhere is unclear on the concept of "share of the total vote" -- and it is probably me.
Now if the percentage of the electorate that is undecided is 30.3% -- then the numbers add up.
Otherwise, not so much.
(Yes, I know the other column is wrong too).
Later - The Lawyer guesses that the figure is an epic typographical error on Kyodo's part, that the number of undecided voters is not 39.8% but 29.8%.
The Lawyer is almost certainly correct.
Now how to fix the second column, which still adds up to 103.5%...
Is it possible that what is being reported are the percentages of those who expressed a preference?
ReplyDeleteNot percentages of the total?
So as an example, 55% who supported a candidate supported "A", and 45% supported "B". Eight percent were undecided.
It really means that of all the poll respondents 55% * 92% = 50.6% want "A", and 45% * 92% = 41.4% want "B".
The assumption is that the people who say yes to polls are reflecting the preferences of those who say no. Also, it assumes that the people who say they haven't decided are even going to vote at all.
Plus, the second column looks like it adds up to 113.5%, not 103.5%
ReplyDeleteSo it may also be that this is a poll that works like this:
1) Have you decided which party to vote for?
2) If you have decided, please tell us which party it will be?
The number that is not being reported is the people who decline to say which party they will vote for, although they HAVE decided to vote for one.
In the first polling, that would be 32.1%; in the second, 30.3%.