Posted on 12/23/2011 12:14:17 AM PST by parksstp
Polls? Screw the polls. If you want any good information you have to find it yourself. So that’s what I set out to do last week. It took about two weeks to produce/run the simulation but I think my product (while nowhere perfect) should be just as good an indicator of what is going to happen in Iowa as a sample poll.
Here is what I did. I went painstakingly through all 99 counties analyzing the numbers on voter turnout and results from both the 2004 General Election and the 2008 Iowa Caucus and General Election to find the most probable 2012 Caucus and General Election totals. The bottom line of those results is that I anticipate about 133,902 people will take part in the Iowa Caucus. I project the following results:
Gingrich, Newt – 34,017/25.40% Romney, Mitt – 33,347/24.90% Paul, Ron – 21,107/15.76% Perry, Rick – 19,906/14.87% Bachmann, Michele – 13,801/10.31% Santorum, Rick – 10,920/8.16% Huntsman, Jon – 804/0.60%
If you open the link to the Google Document Excel Spreadsheet, I will explain how I determined the following results. Again, I want to stress that this method isn’t anywhere near perfect, but I think that most of you will agree that the method is still good and can be somewhat useful.
Okay, when you open the document, the excel workbook opens up to the first tab, called IOWA PROJECTIONS. You’ll notice that I have the party make-up of the state at the top, which is in this case for informational purposes.
I’ve separated all 99 counties into equal blocks containing the same information. You’ll notice that some of these blocks are red, some are blue, and some are purple. RED blocks equate to those counties that were carried by both Bush and McCain in the 2004 and 2008 General Elections. BLUE blocks equate to those counties that were carried by both Kerry and Obama. PURPLE blocks equate to counties that were carried by Bush, but lost by McCain (McCain picked up no counties that Bush lost).
In the left hand corner, each county block will contain in bold the name of the county, the county seat, and it’s approximate geographical location within IA (NW, West, SW, North Central, Central, South Central, NE, East, SE). The blocks also contain the results of the 2008 Republican and Democratic Caucuses. Additionally, the blocks contain the General Election results for the county for 2004 and 2008. As you will notice if you scroll the workbook, you will quickly realize that McCain performed weaker in almost every county as compared to Bush in 2004. Based on the two results, I estimated a percentage of the vote within the specific county that the GOP would need to carry to have a chance at winning IA. I also indicate how strongly partisan the county is (Strongly >70%, Likely >55%, Leans 50-55>). You’ll notice that McCain’s results sometimes downgraded what was a “Likely” county to a “Leans”, or in the worst case, a “Lost Swing”.
The real part of the analysis comes in the far left block. There, I compare the 2004 and 2008 General Election vote totals. A noticeable trend in IA is that many counties, particularly the GOP ones, are declining. This will make it more difficult for a Republican to carry IA in the General. Whether the vote total for the county increased or decreased (rounded to nearest 0.5%) is listed below that. Based on that trend, I use to deduce what the overall 2012 vote totals for that county will be. Obviously, not every county is going to grow or decline at the same constant rate every four years, but I think it’s still a decent number to use. The one exception was the counties with the college towns which had large gains in 2008 that are unlikely to be repeated in 2012, but to keep my method constant, I kept the same format.
Next, based on what I’ve deduced for the 2012 Vote totals for that county, I use my estimate of the percentage I think the GOP needs to be competitive to determine the possible vote total for the GOP candidate in that county. Likewise, I use the reverse percentage to find a vote total that Obama should be held to within the county (example: I say GOP needs 55%, multiply the 2012 vote total by .55, then multiply the 2012 vote total by .45 to show Obama votes). Again, this method isn’t perfect, but barring some strange statistical variation in IA which I don’t see, seems like a good basis to use.
The possible number of GOP voters within the county for 2012 represents the potential pool of GOP and Independents that might choose to take part in the Iowa Caucus. To determine the likely caucus turnout percentage, I added up the total number of votes in each county for the 2008 GOP Caucus candidates and divided that total amount by the number of votes John McCain received in that county in the 2008 General Election. Then, based on whether or not the voter population in the county had increased or decreased, I adjusted the percentage of what I thought the turnout percentage would be for 2012. Usually, voter population gains/decreases of 2% or less kept the percentage stable unless other known factors suggested otherwise. Gains/decreases of 3% -5% usually had a 1% shift, 5-7% around 2% and more than 7%, a 3% shift, though I adjusted for some counties based on the GOP population (especially NW and East IA). Once I decided on a percentage, I multiplied it by the number of 2012 GOP voters within the county, to get a rough estimate of the projected caucus votes in the county for 2012.
Example: Dickinson County. 651 voters took part in the GOP caucus. McCain received 5,162 votes. Based on this, it looks like turnout was 12%. Based on the county population and vote totals from the 2004 and 2008 General Election, the county experienced a 3% gain. Based on that, I determined 10,081 votes were available for the 2012 General Election and that the GOP would need to carry about 54%, which is halfway between Bush and McCain. That equates to 5,444 votes the GOP would be looking for. Since the county grew by 3%, I decided to increase the caucus turnout for the county by 1% to 13%. Then I multiply the 13% by the 5,444 to get what I believe will be the number of caucus goers from the county, in this case 707.
That’s the method I pretty much followed for all 99 counties. Like I said, it isn’t perfect, but it does give a rough estimate of what we can expect voter wise.
Now for the hard part, the distribution of those votes across each county. This is where you guys argue that so-and-so is too high/low in vote counts. One thing I will say is it’s probably accurate that Jon Huntsman’s vote total is too low. I would assume, however, that any gains in votes by him would be at the expense of Mitt Romney, his nearest ideological equivalent candidate.
What we know is that the Huckabee (34%), Thompson (13%), and Hunter (0)% made up the bulk of the 2008 conservative vote. Romney (25%) was the silent moderate (he tried running as a conservative and tried to peg the label on McCain), McCain (13%) was the moderate but did manage pull a little from the other groups, Ron Paul (11%) was the lunatic fringe, and Rudy Giuliani (4%) gave up on IA early.
We know that Huckabee/Thompson voters contained the bulk of the social conservatives, though Paul and McCain probably had a smaller share as well. The question is, where do these social conservatives go in 2012? The challenge is even more trickier when you throw in the Tea Party factor.
AT THIS TIME (notice the disclaimer, at this time), it appears that Newt Gingrich is in line to pick up a plurality (not majority) of the conservative vote. Gingrich will have to share some of the support that gave Huckabee big leads in the various counties among Perry, Santorum, and Bachmann. Each of these three candidates is going to get their fair share of the conservative vote for each of the respective counties. For this reason, Gingrich’s totals and plurality margins in the various counties are much smaller than Huckabee’s.
As the candidate with the deep pockets and the ability to advertise heavily, Rick Perry stands among the three conservatives in the lower tier of making the biggest gains. He’ll outpace Bachmann and Santorum, but the latter two will still have their share of loyal followers, and this will keep him down from challenging Gingrich. Bachmann has slightly more traction at the moment than Santorum, however, their numbers could easily end up merging to 9% each, and based on current polls, that would be seen as a “bounce” for each.
There’s been a great fear about Ron Paul making a mockery of the Iowa caucuses. To be sure, I built into Paul’s numbers a 2012 bounce, especially in the college towns and the counties he ran well in 2008. I think my model is most generous, but even with that, it will be a challenge for Paul to get beyond 16%. statewide. There are only two ways that could happen: (1) A LARGE number of voters from Huckabbe/Romney/Thompson/McCain/Rudy support Paul this time around (Unlikely) or (2) Democrats infiltrate the caucus. Well, if the second option happens, we’ll know it because we can look at the county historical voter totals and find out we’ve been had. For example, in Dickinson County, if instead of 707 total voters, 2,500 or 3,000 total voters showed up and a sizeable chunk voted for Paul, you’d quickly know what was up. I will be watching his numbers closely to see if any Democratic sabatoge is occurring.
Then there’s Romney. He pretty much is about the same. A few gains/losses here and there but overall his numbers are flat. Interesting that both he and Ron Paul run strongest in the liberal eastern portion of IA. While the simulation makes it appear close, I think the margin would be slightly larger for Gingrich, as Huntsman is more likely to take votes from Romney than what I gave him in the simulation. Just taking 1,500 more votes away from Romney and giving them to Huntsman would more than seal the deal for Newt.
So that explains the first tab of the simulation. The second tab is a summary of the vote totals. An additional simulation I ran was a hypothetical 2012 Election matchup between the GOP nominee and Obama in IA. While I listed counties with percentages where the candidate would need to do better than McCain, it still didn’t appear enough to overtake IA. Population decreases in the Republican counties is killing their chances. There appears to be one too many urban centers that gives the Democrats a solid base to work from. IA may soon turn out like PA, where there simply isn’t enough votes to overtake the state. Many of the small suburban and rural counties are 50/50 at best thanks to those farmers that love them some ethanol subsidies.
The simulated numbers I have produced the following: GOP Nominee: 739,900/48.31% Obama: 791,813/51.69%
It’s an improvement from McCain, but it still doesn’t seem to be enough. Not sure if there are even 750,000 Republicans/Independents left in IA like what Bush got in 2004.
Anyway, these where my results. Take em, run with them, do whatever. Depending on whether I’m really (in)accurate, I may try to do one for NH and SC.
Ping to the Perry and Newt supporters on the Google Docs link in the thread containing the Iowa simulation.
You guys can argue over the county distribution on Newt and Perry votes. I think what I have is about as accurate as we’re going to get before the actual vote takes place Jan 3. The biggest factor I’m not sure about is Huntsman. He’ll probably get closer to 3,000 votes, most of which would come at the expense of Romney and leave some breathing room at the top.
Thank you for the ping. That was a LOT of effort. It will be interesting to see how you fair with your projections. A lot changes (as you noted) in the political landscape of Iowa and of course the field is different than in 2008 but it is good of you to put your numbers and formula out there for us political junkies to see.
One suggestion. The colors on your chart are so bold that it hurt my eyes. Probably just my old eyes and I don’t know if you can soften that effect.
Agreed that there was a huge effort put into this but when all is said and done it represents a very sophisticated way of allocating guesses. For me, the result does not pass the common sense test - which if I had done the analysis, would have sent me back to the computer.
I’ll say it right now: Newt Gingrich will NOT achieve anywhere near 40%; this estimate is wildly off. Similarly the analysis certainly underestimates Ron Paul, and probably Rick Perry as well.
So despite all the very obvious work, I’m not sure the result was worth it.
There are many, many variables and daily (hourly) events that alter the numbers, and computers and modeling can’t accurately forecast the final outcome.
But being human we try — for fun and persuasion.
I admire your work but it is very difficult to estimate these things. For instance, how can we even guess how many Dems will switch over to vote for Ron Paul?
Yes, the fall election may be difficult. Eastern Iowa has close ties to Chicago. We are going to have to talk issues over geographic loyalties.
I couldn't read your chart at all without my baby blues going ga-ga.
Leni
Gingrich may come back and be the upset winner in Iowa.
Perfect scenario for him.
It will be interesting to see how your projections match up. Hope you'll update us afterwards.
((((applause))))
Did you factor in any census data? i.e. deaths, exodus, relocation & retirement for seniors, relocation for students and job seekers?
Might not be a factor - since Caucus goers are probably Iowan in heart & soul.
thanks for sharing your results.
Thank you for the work you put into this post!
Freakin’ amazing!
Hey, this is good stuff. Seriously, a lot of effort was put into this, and who knows, it might end up being what happens. It’s all mostly guesswork at this stage anyway. I think 25% Gingrich, 24% Romney, 15% Paul is possible, but Paul will probably run better than that. He’s polling better than that, but we’ll see. I think a good number of 2008 ‘conservative’ voters will end up in the Paul column. For example, in 2008, I would have voted for Thompson. He had quit by the time my primary arrived. In 2012, I’m voting for Paul, with Gingrich as my 2nd choice.
If that’s the way the results finish, Paul will be done as a serious candidate. With all the time and effort he’s put into Iowa, he needs to be over 20%. Otherwise, people just aren’t responding to his message. Gingrich, Romney, and Perry will advance in your scenario.
Yes, a 40% estimate is SO wildly off it means just 25.40%!
I see my mistake. But in my own defense I have to say that the data was VERY poorly presented.
Two points:
1. How can you determine how the fact that the primary this year occurs just after the holidays, when many people may still be away, or travelling, and college students ( who can be wild cards in their voting) haven't yet returned to ccampuses? It's just so different from prior years.
2. I hope you'll do an analysis and critique AFTER the Iowa results are known....an "after-action report if you will (g)..to see how you did, and what went right..and/ort wrong?..and kindly ping me to it. Thanks.
Oh yeah..a very Merry Christmas to you and yours..
EXCELLENT JOB!!!!!!!!!! thank you for your hard work and time.
If you hit ctrl A it will soften the chart and highlight. Makes it easy to read
Hit ctrl A and it will highlight and soften the glare
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.