Michigan: Tie, Foster McCollum White Baydoun (FMW)

[5280]

RCP really did add this already. My god, they're shameless this cycle.

RCP has some cute women working there, it's not all bad.

[Badger]

RCP really did add this already. My god, they're shameless this cycle.

RCP has some cute women working there, it's not all bad.

I have to ask: How do you know this?

[5280]

RCP really did add this already. My god, they're shameless this cycle.

RCP has some cute women working there, it's not all bad.

I have to ask: How do you know this?

Look up Caitlin Huey Burns and Erin McPike

[President von Cat]

Has RCP even added the PPP Virginia poll in? I don't see it on the Virginia display. So freakin' annoying how biased they are, yet the Internet treats their poll averages like gospel.

[Badger]

RCP really did add this already. My god, they're shameless this cycle.

RCP has some cute women working there, it's not all bad.

I have to ask: How do you know this?

Look up Caitlin Huey Burns and Erin McPike

You google hot pollster chicks?

Man, you are a FREAK!

[Simfan34]

I'd like to apologize for that one post, Mr. Foster.

[Antonio the Sixth]

Dear Mr. Foster,

First of all, I would like to thank you in the name of the Atlas Forum community for taking the time to provide the details of your polling methodology. We very much appreciate your attention.

With regard to the merits of the poll in discussion, I think what the Forum community finds problematic is the fact that it seems to sharply differ from the consensus of polling. The most authoritative polling firms have all shown leads for Mr. Obama between 5 and 10 points for several months now. Nate Silver, a renowned polling analyst, calculated an average lead of 6.8 points for Mr. Obama. The results of your poll, thus, differ from the consensus even beyond the statistical margin of error.

Differing from the consensus is not a bad thing in itself. However, when this happens on a consistent basis - your polling firm have always shown results more favorable to Mr. Romney than the consensus - it is the responsibility of the pollster to provide an explanation for this divergence. Rather than a lengthy description of the polling methodology, what the Forum community would expect from your firm is to explain why their methodology is better than that of established pollsters. As your results differ, either your firm is right and the consensus is wrong, or the other way around. The Forum community, as common sense indicates, would lean toward the second option unless a strong case is made for the first.

Regards,

[Ljube]

Dear Mr. Foster,

It seems that when you underpoll minorities and/or some age/gender groups, and then make attempts to offset that by weighting, you may end up misrepresenting the real voting patterns of that minority/age/gender group.

[Holmes]

Good job guys, your constant poll whining has gotten the president of a polling firm to literally join and defend himself.

[EricF]

Hello everyone.

Thank you for taking the time to respond. I appreciate the opportunity to discuss our work and provide the general public with additional insight and clarity as to how we conduct our work. I apologize for the length in my first post. I tend to think in long novel length and tend to talk and write in the same manner. My goal was, as always to provide you and your members with a detailed overview of how our polling methodology is constructed and how we execute our work.

To restate the key reasons why we believe our polling methodology is sound and why our results differ from our competitors rest in three key components:

1.   The use of historical elections data (ballots cast, age of voters who cast ballots, gender of voters who cast ballots, ballots cast in partisan primaries and partisan general election contest, ethnicity statistical data) that is collected by each individual municipal and county clerk’s office for all voters who participated in each election.
2.   The development of a predictive voter behavior model that uses the historical elections data to build turnout profiles of absentee vote totals, election day vote totals, in person vote totals, age, gender and ethnicity clusters of voter participation based on an extensive number of previous elections (minimum 10) and the inclusion of socioeconomic data research on voters to build additional voting impact models that reflect spikes, anomalies and controls based on voters life conditions. 
3.   Using larger sample sizes (minimum 1,000 for Michigan & 1,500 for Florida) to allow for all geographical considerations to be applied to the aggregate universe.

Most of our polling colleagues use exit polling data for one to two election cycles to build their turnout models. These turnout models are the basis of their polling construct and will be weighted to them for any underrepresented groups. We believe using our historical voting data model is more accurate. Exit polling is getting a snapshot of the total voting universe; the historical voting data that the clerks and elections officials maintain is for the total voting universe. If 5,000,000 people vote in the Michigan election or 10 million in the Florida election, we will have the statistics for all 5,000,000 or 10,000,000 voters instead of exit poll surveys. Exit polling is a good methodology and we don’t suggest people abandon it. It is critical for gaining supporting socioeconomic data on the voting universe and their moods. This is the clearest example that I can provide, if 22% of the people who participated in the exit polling survey conducted by CNN, Gallup, Zogby, etc. are ages 18 to 30, you would assume that is the actual voting universe. But when you get the clerk’s statistics for all 5,000 plus voting precincts in Michigan, the actual ballots cast by registered voters between the ages of 18 to 30 is 13% of the total population, which is the accurate reflection of the 18 to 30 year old voting universe. The clerks’ records are the most accurate because they maintain the qualified voter files and they have the voters coded by categories such as age. When we build our model that reflects the actual voting universe of Michigan’s presidential election should be 16% persons between the ages of 18 to 30, we feel confident because of the process of compiling the actual statistics from the clerk’s offices.   

We construct our call files to be reflective of this dynamic, but the human element is always in play with any poll. If your 17% of your call file are people between the ages of 18 to 30 years old and only 4% complete the survey, you will have to make the weighting adjustments to reflect the sentiments of the projected 17% weight that this group should have. We do that with every respondent group that doesn’t meet the proportional goals. It is also something to note that interest among those groups may be waning in the election, which leads to their lower then proportional participation in polling surveys. Additionally, I believe some people make assumptions about voting groups, which don’t fit with the self reported moods of those voters. Two quick examples, President Obama is consistently polling at 28% with Evangelical Christian voters in Michigan and Florida. The assumption is that this voting group would never vote for Obama, but the data suggests they are not totally comfortable with Romney. Additionally, voters ages 31 to 50 have consistently favored Romney in our polling by a range of 10 to 16 points. This voting group has been hit hard by the economic cliff from Bush and the slow recovery under Obama.

In summary, we stand by our methods and the results from our surveys. We have conducted independent or media based public opinion polling since 2010 as well as internal polling for numerous Democratic candidates, non partisan candidates and ballot initiatives. We have a solid track record and believe that our findings are not outliers, but consistent with the historical projected turnout universe for the 2012 general election. In our polling for the Presidential election, in Michigan, Romney only lead once by 3.7 points and Obama’s biggest lead has been 3.45 points. Even if the campaigns haven’t spent significant money, three Super Pac’s supportive of Romney and one Super Pac supportive of Obama have spent heavy here in the past four weeks. My goal is to provide information and clarity on our firm. We respect all views, comments and feedback.   

I also wanted to give you some additional perspective on our previous polling history. Our poll findings have correctly identified the winning candidates, order of placement and geographical, political voting region and congressional district winning candidates with a very high level of statistical accuracy.

FMW & Baydoun Consulting Polling Accuracy – 2010 Michigan Secretary of State Contest

The FMW & Baydoun Consulting polling model was extremely accurate when comparing the aggregate and geographical cross tabular groups.
Our election poll results were:
Total Ruth Johnson               52.89%
Total Jocelyn Benson               46.66%
Another Candidate or undecided:          0.50%    
Johnson’s margin in the FMW & Baydoun Consulting poll was 5.94 points.

The General election was held on November 3, 2010. The final results are listed below:
Total Ruth Johnson               50.68%
Total Jocelyn Benson               45.22%
Other Candidates:                4.10%
Johnson’s margin of victory in the November General election was 5.46 points.

Our polling findings for Ruth Johnson’s victory margin over Jocelyn Benson and their individual percentages of voter support were all within the overall margin of error. We correctly projected that Benson would win the Southeastern Michigan region. We correctly projected Johnson would win the other five geographical regions of Michigan.

FMW & Baydoun Consulting Polling Accuracy – 2012 Michigan Republican Presidential Primary

The FMW & Baydoun Consulting polling model was extremely accurate when comparing the aggregate and geographical cross tabular groups.
Our election poll results were:

Mitt Romney:                          37.90%
Rick Santorum:                         35.76%
Ron Paul:                             9.12%
Newt Gingrich:                         8.31%
Undecided:                            8.90%
Romney’s margin in the FMW & Baydoun Consulting poll was 2.14 points.

The primary was held on February 28, 2012. The final results are listed below:

Mitt Romney:                          41.10%
Rick Santorum:                         37.87%
Ron Paul:                             11.63%
Newt Gingrich:                         6.53%
Other candidates:                         2.88%
Romney’s margin of victory in the February Primary was 3.23 points.

We correctly projected the following election winning occurrences:
•   Eleven (11) of the 14 Congressional Districts.
•   Southeastern Michigan, Southwestern Michigan and Thumb Region of Michigan.
•   The major 17 county cluster and each sub group (Major Democratic 6, Republican 7 and Swing 4 counties).
•   The Next 7 County and Other 59 County cluster.
•   Wayne, Oakland and Macomb Counties

Our polling model correctly identified two of the four Congressional districts that swung based on Election Day turnout, the 1st and 13th Congressional Districts. Both of those districts swung from Mitt Romney to Rick Santorum. We also correctly identified the battleground important of the 5th Congressional district. Our final poll found that the district was tied at 41.77% for Romney and Santorum. Romney won the 5th Congressional District by 452 votes. That margin ultimately prevented Rick Santorum from winning the majority of delegates in Michigan by having won 8 of the 14 Congressional districts.


Michigan 14th Congressional District Democratic Primary Election Exclusive Polling Study for Fox 2 News Detroit July 23, 2012
The (FMW)B polling model was extremely accurate when comparing the aggregate and geographical cross tabular groups.
Our election poll results were:

Current Democratic Congressman Gary Peters:          45.11%
Current Democratic Congressman Hansen Clarke:           26.72%
Southfield Mayor Brenda Lawrence:               9.64%
Former State Rep. Mary Waters:                 1.17%
Retired Magistrate Bob Costello:                0.29%
Undecided:                         17.08%
Congressman Peters’ margin in the (FMW)B poll was 18.39 points.

The primary election was held on August 7, 2012. The final results are listed below:
Current Democratic Congressman Gary Peters:          47.03%
Current Democratic Congressman Hansen Clarke:           35.18%
Southfield Mayor Brenda Lawrence:               13.29%
Former State Rep. Mary Waters:                 3.33%
Retired Magistrate Bob Costello:                1.17%
Congressman Peters’ margin of victory in the August Primary election was 11.85 points.

Our polling findings were very accurate for Congressman Peters’ aggregate support percentage and key geographical cross tab groups. Our poll findings found that Congressman Peters’ had the following support:
•   Southwestern Oakland County region – 73.98% poll support to 77.77% actual election voter support
•   Southeastern Oakland County region – 46.78% poll support to 46.47% actual election voter support
•   Near Detroit Suburbs communities – 52.78% poll support to 50.82% actual election voter support
•   Wayne County region – 33.06% poll support to 33.53% actual election voter support
•   Detroit region – 28.42% poll support to 29.19% actual election voter support

[Likely Voter]

The simple fact is that these FMW polls are clearly different than other polls so they are either on to something, or wrong. 20 paragraphs about the methodology isn't going to change that.


I hope that the FMW people (if genuine) return to this forum after the election and take their victory lap when it was shown that their model accurately portrayed the general election voters of 2012 in MI and FL.   Or vice versa. 

[Ljube]

Mr. Foster,

It would be great if you could do another poll in FL and another one in MI during the next week.

[minionofmidas]

I certainly approve of the general idea of the owners of second-rate polling firms posting on the Atlas Forum. I approve of it very much.
For that reason alone, thumbs up.

(As to the poll, lol. As to the post, tldr.)