- one of the big factors that a lot reports on polls leave out is the % 'undecided

Women and Ignatieff: What went wrong? , Michael Valpy, From Tuesday's Globe and Mail Monday, Oct. 05, 2009
http://www.theglobeandmail.com/news/politics/women-and-ignatieff-where-did-it-go-wrong/article1313172/ Tab 34

The Comments sound like what they used to say about Harper when he because leader of the Opposition.

Also, one of the big factors that a lot reports on polls leave out is the % 'undecided'.

This is very important in the current dynamics since the Con's have a "die-hard" core of support and they are very unlikely to change from 'decided' to 'undecided'. Further they are very likely to choose anything Con, simply because it is Con, over anything other than Con simply because it is other than Con (that's what 'die-hard' support means).

On the other hand reports that do show the amount of undecided indicate that in the past few months there has been a significant increase in the number of 'undecideds'.

In such circumstances, what that means is that where the 'decided pie' decreases, the % that are made up of Con supporters increases per-centage-wise (if that's a word). The indications are that this is a significant effect. However, unfortunately there is not enough information given in the report to determine to what extent we are seeing this effect as opposed to true shifts in voter preferences. In the last election 'Liberal' voters protested by not voting. It is suggested that we are seeing this phenomenon now in these Polls.

For example, one would expect a high level of 'undecided' (which translates into non-voting in an election) in young people with a gradual decrease in 'undecided' the older the group. For the 18 - 35 group, one might expect little change in the number who express support for one party or the other over time since the number of undecided is so large anyway and the 'decideds' are likely die-hard supporters for one or the other party. As the age group gets older the number of 'undecided' decreases, or another way of putting it, the number of 'decided' increases. So, if the trend is that 'decideds' turn to 'undecideds' you might expect that the total % of decided decreases and where one represents essentially all 'die-hards' that parties % support will seem to increase, since it becomes a larger part of the 'decided pie' and the other party % will decrease. So, the prediction is that as the age group gets older, the total supporting one party or the other decreases with time and the % of decided will increase to the Con's since the % of supporters of the Con's that are die-hards is so high.

Compare this to the trend in the results here:

For 18 - 35, the total for the Con's and the Liberals is lower than the other groups and has the smallest change in time: 59 - 61 (presumably constant within statistical variance)

For 35 - 49, he total for the Con's and the Liberals is in the middle compared to the other groups and the change is decreasing as expected and has the middling change in time: 66 - 58 (change being: -8, presumably statistically significant)

For 50+, the total for the Con's and the Liberals is in the highest compared to the other groups and the change is decreasing as expected and has the highest change in time: 76 - 70 (change being: - 6, that it is lower could be accounted for by statistical variance as well as how the age groups id broken up, but it does not refute the above hypothesis since if the error is +/- 3 it could be - 5 for age group 35 - 49 and -9 for age group 50+, of course it could go the other way as well.)

So, the "trend" could be no more than reflecting that the Con's have a core of die-hard supporters with little more and the Liberals have a core of die-hard supporters along with a large number who support the Liberals but refuse to vote when they don't like what they see. This suggests that the Liberals and Ignatieff make sure these soft supporters understand what they and he is all about. It also suggests that these soft supporters should reconsider their stance if they want to get rid of Harper and the Con's.

Lloyd MacIlquham cicblog.com/comments.html