Substantial early (absentee) voting has occured in Iowa. Returns are reported by registration. Thepre-election PPP poll can be compared and contrasted with the reported returns to validate and extrapolate results.
PPP asks respondents if they have voted early, and presents cross-tabs for their selection.
The PPP poll reports 47% of respondents have voted. Current total returns for absentee ballots (672,255) are 45% of the Iowa 2008 Presidential total. A voter turnout model using current registrations and historic (2004, 2008) turnout figures of 71-72% are coordinate with the polling report.
This implies the PPP poll is accurately reflecting likely (as opposed to registered) voters.
The PPP poll cross-tab reports 61% of Early Voters selected Obama. This is difficult to reconcile with the party registration breakdown of the actual returned ballots, which is 42% Dem, 32% Rep, and 26% Ind. Balancing cross-over voting, the independent vote would have to go 19% to 7% (or 73% of the subcategory) to Obama to record the total position on 61%. Since the overall Independent split reported by PPP is 51% Obama, this high Independent Obama support implied among early voters seems improbable. Cross-over voting is reported in the tabs, and is 13% Dem>Republican and 10%Republican> Dem. This cross over fraction is similar to the Nevada, and other tabs I have inspected.
This high Obama early vote fraction reported in the PPP poll implies that for some reason the PPP is oversampling Obama supporters in early voting. I made the same observation on the Nevada-PPP cross tab comparison.
Conversely, the PPP poll seems to reflect a day-of bias for Republicans. The reported split for "late" voters is 41%O-56%R. This is also similar to the over-weight in late voters in Nevada I observed in the cross tabs. In terms of age breakdown, the same paradox that I observed in Nevada reappears: Older voters are more likely to have already voted, but older voters are more likely to be conservative.
The age-candidate cross tab implies that there is a cohort of late-voting conservative 30-45 year olds in Iowa -- this is similar to Nevada where the young age and conservative selection would need to be combined to get the reported late voting conservative slant.
It appears there are a poll-responsive pattern where conservative young voters claim to plan on participating, but are perhaps unlikely to appear. I have no good explanation for the apparent overweight towards O in the early voting crosstabs in both NV and IA, and the apparent underweight towards O in the projected late voting crosstabs. Poll participation (hang-ups) may be occuring at different rates in these two groups.
Presently, late-voting projected turnout ( in aggregate 73% of registered) and the PPP recorded preference split (51% of Independents for Obama for all respondents) implies an extremely narrow race in IA. This split is modeling a 49.4% vote total for Obama with returned ballots and projected turnout.