Showing posts with label tracking polls. Show all posts
Showing posts with label tracking polls. Show all posts

Wednesday, September 3, 2008

A Follow Up on the 50% Mark: The View from 2004

Yesterday I made note of Barack Obama clearing the 50% mark in the daily tracking polls from both Rasmussen and Gallup. Now, I mentioned the significance of the 50% barrier and the timing -- September polls are better predictors of November results than late October or November polls -- but let's try and add a bit of context here. What did things look like in the national polls four years ago and how does that mesh with what we have seen and are seeing in this current race?

A quick glance at what Real Clear Politics has to offer from four years ago gives us a good start. By my count, there were 155 national polls that were conducted between March and election day. 76 came prior to the beginning of convention season in late July and the remaining 79 polls were conducted in the period during and after the conventions. Obama became the first candidate during this cycle to hit the 50% mark, just after the beginning of September. By contrast John Kerry hit that mark in early April. He didn't stay there, but the Massachusetts senator broke that barrier first. In fact, he hit it at various other points 10 other times prior to the Democratic convention in late July. President Bush, by contrast, was only over 50% in any of the national polls two times out of those 76.

The post convention story was a bit different. Of those 79 polls during and after the conventions, Bush broke the 50% threshold 27 times (over one-third of the time) while Kerry managed only five polls over that mark in that time. Granted, across the entire 155 polls (before and after the conventions), that leaves 61% that showed a closely contested race with both candidates in the mid- to upper 40s.

In fairness, this is a bit of an apples to oranges comparison (Alright, perhaps a Granny Smith to red delicious comparison.). Yesterday's numbers were from tracking polls, but the above data are from national polls, yes, but not tracking polls. As I mentioned, though, the 50% trend would have to extend to state polls in battleground states and national polls and last for a period of time for this to resonate in any way. The other caveat is that 2004 was a different year with a different set of variables. The presence of an incumbent in the race may have a lot to do with the differences we see. In a campaign without one, voters are still attempting to figure out who the two candidates are in 2008. They got a pretty good idea about Obama last week, and are hearing the rebuttal and McCain pitch this week.

Still, those 2004 numbers speak for themselves. During September, Bush led in 23 of the 25 polls that were conducted (Kerry led in one and was tied in the other.). Once we clear the next week or so (past the conventions), we should get a pretty good idea of where the race stands in 2008.

For a deeper look into the history of the national polls during the course of the presidential campaign, Andrew Gelman had a post up in June that had a figure from his 1993 paper with Gary King charting the polling trends from the 1952-1992 elections.


Recent Posts:
Obama Cracks 50% in the Daily Trackers. What Does It Mean?

And What About the Green Party?

It's Never as Easy as Taking Away Half the Delegates

Tuesday, September 2, 2008

Obama Cracks 50% in the Daily Trackers. What Does It Mean?

You'll notice that I've not really spent all that much talking about national polls much less national tracking polls in our electoral college analyses. They are important, but only to a point. I obviously prefer the state level polls -- though the two are linked in some ways -- due to the way the US presidential election is structured. Having said that, Obama crossed over 50% in both the Gallup and Rasmussen tracking polls today. I find that that to be significant for one main reason: typically September polls are better predictors of the ultimate outcome than those released immediately prior to the election.*

Now Democrats, I don't want to get your hopes up. [It has happened before (see 2000 and 2004).] However, if Obama is able to maintain this level of support through the next couple of weeks (And that is far from a sure thing, folks.), regardless of the fluctuations on McCain's side and thus the margins of the race, he theoretically should be looking pretty good for the general election. Again, this is just one facet of the robust polling regime we have at our disposal. If both the national polls (not the tracking polls) and the polls in certain states are also favorable to Obama over this period, then things will be looking up for the Illinois senator.

Starting tonight, though, the Republicans will start having a bigger say in the matter. Still breaking that 50% mark is worth noting.

*James Campbell has used Labor Day polls in his forecasting models in the past. Speaking of forecasting, following the American Political Science Association meeting in Boston this past weekend, most of those models are out. Here is a link to a synopsis of several of them.

Thanks to Paul Gurian via Del Dunn for the forecasting link.


Recent Posts:
And What About the Green Party?

It's Never as Easy as Taking Away Half the Delegates

The Electoral College Map (8/31/08)