make a chart. That seems to be the new policy here at PoliticalStaples. The consensus wisdom is that Liberal support always drops and Conservative support always rises in an election campaign when compared with pre-election polling. What better way to test this than to make another one of my polling charts. As I have said many times already, I trust Ipsos-Reid polls. They seem less prone to wild fluctuations and that makes me believe they are more accurate (have at that assumption as you will).
Ipsos-Reid published many polls right before PM Martin drop the writ on May 23, 2004. I have plotted this data (greyed line) versus the actual result on election night (solid line) to see if there is any truth to the assertion. I put the actual support as a solid line to provide a reference. If there is a better way to do this feel free to let me know.
As you can see, it was not until right before the writ was dropped that Ipsos-Reid began to register a shift in the electorate. Is this evidence the Liberal "parked-vote" began to actually make a choice in whom they would vote for? That would be anecdotal at best. But you can see that the polls never captured what the Conservative support would be on election night. The support was consistently underestimated by at least 4-points. NDP support seemed to relatively well estimated. However, the Bloc Quebecois support was underestimated as well. So in both Quebec and the ROC the Liberals main opposition (excuse the broad generalization) was underestimated or put the other way, Liberal support was overestimated.
Definitely not conclusive but this political shorthand does appear to have some validity.
Considering recent polls have the Liberal-Conservative gap at somewhere between 4-7 points, things may actually be a dead-heat.
