What the Media Should Report About the Ames Straw Poll
David Terr
Lead Analyst
Ph.D. Math; Berkeley
8/12/07
Mike Huckabee and Ron Paul Were Clear Winners, Mitt Romney Was a Loser
"Often, however, the media do not actually interpret poll results, but instead present simple, straightforward descriptive statements about them... Measures of association, correlation analysis, and multivariate statistical analyses are virtually nonexistent" - Asher, Ohio State University
I would first like to congratulate Mitt Romney for his impressive win at the Ames Straw Poll before I start criticizing him and tell you what the media is not reporting.
|
When you consider all the money Romney had spent in order to win the Ames Straw Poll, he did not do well at all. I do not know the actual dollar amount spent by the Romney campaign but I think it would be safe to say that it was on the order of hundreds per vote. But Romney is a smart businessman, he understands that winning this event will generate tons of free positive media attention worth far more than the petty millions he spent for Ames. I cannot give the Romney campaign the win here today because his 31% at Ames is already close to the 27% he got in the University of Iowa poll earlier this month. Compare that to Mike Huckabee who got only 3% in that same poll getting 18% here. |
What's even worse for Romney is the fact that Giuliani, McCain, Gingrich, and Fred Thompson were not even vying for votes at the Ames Straw Poll yet they are included in the polls that have Romney in the high-20s. And add to that the fact that 20% of Republican voters are undecided and you should have had a slam dunk by Romney by at least a 5:1 margin over his next closest competitor.
| Actual Results - of the Latest Iowa Polls | ||||
| Candidate | Univ. of Iowa | ABC News | ARG | Research 2000 |
| Mitt Romney | 27% | 26% | 21% | 25% |
| Rudy Giuliani | 11% | 14% | 22% | 13% |
| Fred Thompson | 7% | 13% | 13% | 14% |
| Tom Tancredo | 4% | 5% | 1% | 2% |
| Sam Brownback | 4% | 5% | 1% | 2% |
| John McCain | 3% | 8% | 17% | 10% |
| Mike Huckabee | 3% | 8% | 1% | 2% |
| Tommy Thompson | - | 4% | 2% | 2% |
| Ron Paul | - | 2% | 1% | 1% |
| Duncan Hunter | - | 1% | 2% | 1% |
| Newt Gingrich | - | - | 4% | 6% |
| Unsure | 31% | 7% | 15% | 22% |
| Other | 10% | 1% | - | - |
| None of these | - | 4% | - | - |
So what we did is we took the last 4 poll results and removed all the votes for Giuliani, McCain, F. Thompson, Gingrich, and the Undecideds and divvied up the rest of the votes to see the percentage.
| Normalized Results - Removing Giuliani, McCain, F. Thompson, Gingrich, and Undecideds | ||||
| Candidate | Univ. of Iowa | ABC News | ARG | Research 2000 |
| Mitt Romney | 71% | 51% | 72.4% | 71.4% |
| Tom Tancredo | 10.5% | 9.8% | 3.4% | 5.7% |
| Sam Brownback | 10.5% | 9.8% | 3.4% | 5.7% |
| Mike Huckabee | 7.9% | 15.7% | 3.4% | 5.7% |
| Tommy Thompson | 0% | 7.8% | 6.9% | 5.7% |
| Ron Paul | 0% | 3.9% | 3.4% | 2.9% |
| Duncan Hunter | 0% | 2.0% | 6.9% | 2.9% |
| Percent of Total Poll | 38% | 51% | 29% | 35% |
Flaws. The flaw with this is the error in assuming that voters who favor those candidates we had removed would be distributed across the rest of the candidates in the exact same proportions. The same would be true of the way the votes would spread out with the undecideds and the decideds. The last flaw is that by removing votes you decrease the sample size which results in a grossly high margin of error as much as 9%.
Despite all of that, 3 out of the 4 polls suggest that Romney would be getting 71-72%. And by averaging all of those normalized averages, we correctly predict the Top 4 finishers in exactly the same order they came in.
| Candidate | Combined Normalized Average | Straw Poll Average | Percent Gain | Net Gain |
| Mitt Romney | 66% | 31.5% | NEGATIVE | -34.5 |
| Mike Huckabee | 8.2% | 18.0% | 220& | +9.8 |
| Sam Brownback | 7.4% | 15.3% | 207% | +7.9 |
| Tom Tancredo | 7.4% | 13.7% | 185% | +6.3 |
| Tommy Thompson | 5.1% | 7.2% | 141% | +2.1 |
| Ron Paul | 2.6% | 9.1% | 350% | +6.5 |
| Duncan Hunter | 3.9% | 1.2% | NEGATIVE | -2.7 |
According to this estimate, Romney should have had an 8:1 margin of victory over his next closest competitor. In fact, I heard some analysts that made this exact conclusion.. they probably did something very similar to me to get their prediction.
I think the pundits that claimed Romney's win would be very comfortable hurt him in this Straw Poll. Much like when Clinton went for reelection in 1996 and all the pollsters claimed an easy Clinton victory, many voters were apathetic and just stayed home on election day. The same could have been true of Romney's supporters expecting a very comfortable win.
There is also the the likelihood that the Unsure voters flocked towards the other candidates (not Romney) in a larger proportion than the decideds did. That is very likely because whenever there is a race between a very well known candidate and someone less known, a voter being Undecided is usually a No vote to the well known candidate. The other unknown is how the Giuliani, McCain, Thompson, and Gingrich voters distributed across the other candidates. They may have supported Huckabee, Brownback, and Tancredo at a higher rate than they did Romney.
In looking at these results, Romney did more than a factor of 2 worse than expected as per the Combined Normalized Average (CNA). He was a big loser.
Mike Huckabee more than doubled his expected CNA and is obviously the true winner at the Straw Poll. Brownback and Tancredo had similar jumps but coming in third and fourth is about as good as kissing your sister.
Tommy Thompson barely moved up compared to the CNA thus he was clearly a loser. He has continuously said that unless he gets second place finish at the Straw Poll, he will not continue on with the campaign. Well sayonara Tommy because not only did you not do well but you finished sixth place.
Ron Paul showed a 350% improvement over his CNA at the straw poll and is clearly the second winner at the event. It is easy to show such a marked up win when you are only averaging about 2-3% in the polls but his polling average has been 50% lower than Duncan Hunter yet he wiped the floor of Hunter by a factor of 8. He had a raw net gain in percentage more than Tom Tancredo.
Ron Paul supporters should be very proud of his performance here. It may be reported as a horrendous performance because he came in fifth place but when you consider he was not even included in many polls just a few months ago and his poor polling numbers thus far have been keeping him down, 10% here is great.
Herbert Asher, a Political Science professor at Ohio State University, said that "continuing coverage of long-term trends, and background news is often neglected" by the media. The background news in this story is simple:
- Romney did not do well, his supporters may have stayed home expecting an easy win.
- Huckabee and Ron Paul soared above all expectations and are on the way up.
- Duncan Hunter and Tommy Thompson should drop out of the race.
I think Ron Paul's above expectations performance is a result of the underdog theory in reporting of polls. The concept that when supporters see their candidate down in the polls, they support him with greater angst.
With Mike Huckabee, I think the above expectations performance will have far greater impact on the presidential race than anything else. Look at the front runners: Romney is a mormon, Giuliani has been divorced and dressed in drag, McCain is old and dropping... Huckabee has set himself as the clear alternative choice. He had polled as high as 8% in Iowa (ABC News Poll) and will likely see himself in double digits in the next poll. In the next few months, it is going to be interesting to see if Huckabee can oust Romney. My perception is that once voters believe Huckabee can win, some of them will drop the Romney campaign for Huckabee. Because essentially a vote for Romney is a vote for traditional christian conservative values that Huckabee brings to the table.
Call this spin or whatever else you want but this is my interpretation of the data. Leave your comments below.