One week from the caucus vote in Iowa, trend lines are firming up. In this diary, NBC/Marist poll data is used to make a prediction of the Iowa GOP caucus results.
The numbers used in this diary are based on the 2012 NBC/Marist Iowa poll from Dec 27-28 and the poll from Nov 27-29.
Disclaimer: This is not intended to be taken as a rigorous analysis. It is a simple exercise using one set of publicly available data to guess at what the results will be.
The NBC/Marist first and second choice results in Jed Lewison's diary caught my eye (see graphic above). The first choice ranking seemed to diverge significantly from a combined "top two" ranking. For instance, second choice numbers indicate that Rick Perry has much stronger support than first choice numbers show. It also implies that Ron Paul and Rick Santorum are nearly tied in a "top two" ranking. With the amount of candidate switching happening so far, I thought it might give an indication of who caucus voters will settle on come Jan 3rd. To test my theory, I decided to use a weighted sum of first and second choice and extrapolate trends to Jan 3, i.e. a ranking of candidates according to "top two" status. In addition, I distributed undecided respondents according to trends and second choice preferences for each candidate [I use second choice as a proxy for hidden support that may sway an undecided voter]. Below are the results of this analysis, including the data table used. Percentages listed numerically (xx%) below are "top two" trends relative to November numbers for each candidate, not total support.
#1
Romney leads in both first and second choice, with a 22% increase in weighted support since November. Barring a large scale turnout for Ron Paul, Romney likely wins.
#2
Second place is more uncertain. First choice numbers imply Paul, but the weighted sum drops him to tie with Santorum for third, moving Perry to number two. Both Perry and Santorum are surging. Perry increased his weighted support 62% while Santorum had an even stronger surge of 114%. As first choice, Perry is down one point to Santorum, but Perry has stronger second choice numbers.
Hard to call, but a simple straight line extrapolation of the trend has Perry maintaining a lead. The undecided and switchers will determine if Santorum can pull into the lead (see Other/Undecided below).
Based on his historically motivated and consistent base, Paul will likely stay above twenty percent of the vote despite the results of the weighted sum method used here.
#3 and #4
Paul matches Santorum in weighted sum, both current and extrapolated. But with Santorum's surge and poor second choice numbers for Paul, Santorum has a chance to pick up a good chunk of undecideds and switchers. This implies Santorum will pull ahead and finish third, leaving Paul out of the top three (not taking into account the likely strength of Paul's first choice numbers).
#5 and #6
After Santorum and Paul comes a sad story of decline. Gingrich support cratered, dropping 45% from November highs. Will tears for a mother stem his loss and counter negative media portrayals? It seems unlikely that it will move GOP activists in a positive direction, and negative media bodes ill for his campaign. While Bachmann slips down further, her losses are smaller, just above ten percent. Application of the above method puts Gingrich in fifth with Bachmann close behind. But something tells me that Gingrich's fall is accelerating, and that Bachmann's support is resilient. I wouldn't be surprised if their extrapolated support numbers flip, putting Bachmann ahead.
#7
Unfortunately for Huntsman, a steady 2% first choice and 3% second choice implies at most five percent of poll respondents thought of him as a top two GOP candidate. Other/unsure polled almost three times higher than Jon Huntsman. Sorry Jon, not this time. Count yourself lucky you weren't caught in the coming implosion. Hope you didn't say anything that you'll regret the next time.
Other/Undecided
The current other/undecided percentage is similar to final 2008 GOP Iowa caucus polls. Comparing 2008 Iowa caucus results, the top seven candidates received over 99% of the total vote. Assuming 100% of votes will be cast for the current top seven, and distributing other/undecided to each candidate by the product of their trend and second choice numbers, Romney gets a boost and Perry barely keeps second place ahead of Santorum.
To sum up the prediction based on the above method (extrapolated "top two" trend plus distribution of other/undecided), we have:
Romney 24.0%
Perry 21.6%
Santorum 21.2%
Paul 15.1%
Gingrich 9.0%
Bachmann 7.9%
Huntsman 2.5%
Note: trend extrapolation resulted in a total slightly greater than 100%.
A few points I found interesting.
- The NBC/Marist poll released at the beginning of December included GOP matchups with Obama. Only Paul pulls even. Obama trounces all other candidates polled.
- 20% of registered voters said they would attend the Republican caucus, but that implies more than tripling the 2008 caucus attendance (428k vs 119k). Assuming 2008 is a fair reflection of what to expect, results will be driven strongly by which faction is more motivated (the usual case for caucuses and primaries) with Ron Paul the likely beneficiary.
- Tea Party support continues to decline, now at 19% of potential GOP primary electorate. However, likely voters among that group have far higher support for the Tea Party (46%).
Putting it all together, Mitt Romney is the almost certain winner of the Iowa caucus on Jan 3. Ron Paul will likely outperform the numbers predicted above resulting in Rick Perry and Santorum vying for number three and four. Gingrich, Bachmann and Huntsman drift off the radar.
Depending on caucus results, I expect the "top two" method to reflect candidate selection when there is low commitment and lots of switchers, but it will likely undercount motivated factions like Paulites (and Bachmannistas?). Any suggestions or comments are welcome.
DATA TABLE (sorry for the ugly layout)
candidate 1st choice, 2012 12/27-28: 2nd choice, 2012 12/27-28: sum: weighted sum %: sum 1st/2nd 2012 11/27-29: net change in weighted sum: percentage change from 11/27-29: Extrapolated weighted sum on Jan 3rd: Other/unsure distributed by trend and second choice: Extrapolated with other/unsure distribution
Romney 23 21 44 22% 36 8 22% 22.82% 1.21% 24.03%
Perry 14 20 34 17% 21 13 62% 18.42% 3.20% 21.62%
Santorum 15 15 30 15% 14 16 114% 16.78% 4.43% 21.21%
Paul 21 9 30 15% 29 1 3% 15.04% 0.08% 15.12%
Gingrich 13 13 26 13% 47 -21 -45% 10.50% -1.50% 9.00%
Bachmann 6 11 17 8% 19 -2 -11% 8.23% -0.30% 7.93%
Other/unsure 7 8 15 7% 18 -3 -17% 7.11% -0.34%
Huntsman 2 3 5 2% 5 0 0% 2.49% 0.00%