The polls have been bouncing all over the place, and pundits are broadly whining that the reason is difference in “turnout models”.
That is, how many more (or less) Democrats will show up to vote for Obama.
To understand the issue, I framed a related – but inverted — analysis by asking the question: by how much does Dem turnout (in swing states particularly) have to exceed GOP turnout for Obama to win?
The answer: Dem turnout has to be more than about 4 percentage points higher than GOP turnout for Obama to win.
Here’s my summary chart … below it are the assumptions and analytical logic.
From the chart: if Dem turnout is about 8 percentage points more than GOP turnout (as it was in 2008), Obama wins by about 4%; if Dem turnout is less than 4 percentage points greater than the GOP’s, Obama loses.
It’s as simple as that … especially on a swing state by swing state basis.
= = = = =
Assumptions & Analysis
While there has been a lot of bounce in the numbers, a couple of things appear to be pretty stable.
First, both Romney and Obama capture over 90% of their party’s votes.
Second, independents are generally about 1/3 of the total voting base … and, independents seem to be breaking towards Romney 55% to 45%.
In a nutshell, that means that Obama has to overcome a 3.3% Romney vote advantage with over-performance in Dem turnout.
- 10 percentage point independent vote differential times 1/3 of the voting population equals 3.3%
Let’s run through a couple of examples:
1) Assume that the turnout is evenly split among Dems, GOP, and independents; that Obama & Romney each get 95% of their party’s votes; and that independents vote Romney 55% to 45%.
Under these assumptions, Dems have no turnout advantage (because that’s what we assumed) … and Obama loses by 3.3%.
= = = = =
2) Same assumptions as example #1, except assume that the Dem turnout is 8 percentage points greater than the GOP’s … roughly comparable to 2008 voting patterns.
Under these assumptions, Obama wins by almost 4%.
= = = = =
3) Same assumptions as example #1, except assume that the Dem turnout is 3.7 percentage points less than the GOP’s.
Under these assumptions, the race is tied … we’ve found the sweet spot … if the Dems turnout advantage is more than 3.7 percentage points, Obama wins; less than that and he loses.
= = = = =
Final Notes
1) It’s simply math magic that the relationship works out to be linear … as displayed on the summary chart.
2) If you don’t like my assumptions, plug in your own … my conclusion: the numbers are pretty robust to changes in the assumptions
3) Nobody seems to be predicting Dem turnout comparable to 2008 … In fact, some are predicting that GOP will have a turnout advantage,
4) You haven’t seen an analysis like this anywhere else, right? Only in the Homa Files …
* * * * * *
The Homa Files 