Comparing Tracker Trends
Charles Franklin | November 3, 2008
A look at the trends for the different trackers. Interesting difference in those that are recently rising vs those flat. There is a notable difference between some rising and others nearly flat. Eight are rising at least a little with two falling (DKos and Rasmussen) but three or four are of the rises are quite small.
The range: +3 to +11.5 for Obama.
Comments
Huh, why does this look so different than 538's poll of polls super tracker?
http://www.fivethirtyeight.com/
Posted on November 3, 2008 3:17 AM
"Huh, why does this look so different than 538's poll of polls super tracker?"
538 weights all these polls differently.
Posted on November 3, 2008 3:46 AM
I don't know which 538 poll you are talking about, but the vertical axis is labeled as "Estimated Obama minus McCain Margin". For example, if Obama was leading 50 to 45, then the graph will have a value of +5. Over the past 2 months this graph is showing the trend of all these polls by only showing the difference in percentage points between Obama and McCain.
Posted on November 3, 2008 6:30 AM
538's Supertracker includes data from state polls as well as national. The recent decline in margin is due to a bit of narrowing in the former, while the average for national polls, as Pollster shows above, has been fairly even.
The interesting thing, as Mark points out, is an apparent divergence into two groups in the last day or two. Given the small number of polls, however, and the lack of an adherence to longer-term house effect patters, I suspect this is noise.
Posted on November 3, 2008 7:01 AM
It occurs to me that one thing Hotline, R2K, and Rasmussen have in common is a fixed partisan ID methodology (with weekly adjustments in Rasmussen's case). A lot of ink has been spilled over the merits of such an approach, but in this case I wonder if that methodology is in effect causing those polls to miss important information coming out of early voting.
Posted on November 3, 2008 8:26 AM
Post a comment