A week or so ago,
ManfromMiddletown posted a story on the potential for a
Democratic Landslide...in Indiana. In it, he outlined the reasons why Indiana, who is known nationally as one of the redder of reds, is not really the foregone conclusion that everyone thinks it is. Rather, the potential exists in the state for Democrats to retake the state legislature, the Governor's mansion, and the Congressional delegation...all within the next two years.
------->
But the most important part of the story came at the end, when ManfromMiddletown discussed some informal polling done by the Barry Welsh campaign in Indiana's 6th district. Barry posts on here regularly, and has embraced the blogosphere and the spirit of "People Powered Politics". Done completely by volunteers, the poll revealed that incumbent Mike Pence is much more vulnerable than thought. Despite an error in their sample data, (which ManfromMiddletown explains would help Pence), the potential for these homegrown, grassroots polls to provide a buzz for second-tier campaigns is obvious.
Now this is where you come in...
ATTENTION
I have been so impressed with the feedback I have received from people interested in helping with this project.
For the record, ANYONE with interest should email me at denymyfreedom at gmail.com and get added to the mailing list. Specifically, if any of you have polling experience, website development experience, or computer systems management experience, I would love to hear from you.
I am going to be taking all of the information in this 'brainstorming' thread, and turning it into a formal proposal, which I will be sending out to DFA and other national organizations. So keep the ideas coming!
Thanks!
-------------------------------------------------
(Originally posted at
Deny My Freedom)
This was said toward the end of the aforementioned story:
For its flaws, Barry has hit on something with his effort to do a poll. Barry was behind the 50 state strategy website.
Unless they are considered first tier race, most congressional races will never be polled, and potential weak seats will be ignored because of this. With the 50 state strategy we need "open-source polling," where templates for simple polls designed to be clear, precise, and scientifically valid are run by volunteers at the local level.
Kos, and Chris Bowers point to polls from 2nd tier races show that democrats are competitive in districts that aren't target races. In Indiana, Congressional districts have about 675,000 people (over 80,000 population size is not very important), in order to get a margin of error of +- 5%, 384 voters would have to be contacted in each district. At 5 calls an hour, that's 64 hours(or Woman hours, ladies?).
Put another way that's 8 people volunteering for an hour an half a day for a week. That's doable. And even if you give volunteers a little something (this draws in the college students) say $50 for the week, that's still only $400 for a poll. A benchmark poll from one of the national's will set back a campaign $15,000 to $25,000 at least. Most campaigns can't afford that. This needs another diary.
(emphasis added)
I think this is a great idea. It is a quick, easy, and completely people-powered solution to a barrier that has kept low-budget candidates out of the national spotlight for too long. By no means am I saying that this sort of system would make large polling firms obsolete, but rather that it could provide a foot in the door for second-tier candidates to get their name a little buzz.
Let's use the Barry Welsh survey as a case study:
We contacted 17 Non-primary Voters from all 19 Counties taken at random from the voter registration rolls. There were 323 Successful calls.
There were only two questions asked. Here is the script.
"Regarding the 2006 Congressional Race in your district,
Do you plan to vote in November?"
If No, call was ended with a thank you, if Yes, second question.
"Do you plan to vote for Mike Pence, Barry Welsh, or are you undecided?"
The call was ended with a Thank you.
There was no identification of party or campaign. Each call was less than 20 seconds with few exceptions, and it took over 6 hours to get 323 successful calls. (Obviously this was not done by a paid service, but by volunteers.)
The results of the Non primary or Independent voter Initial Survey:
239 Said they planned to vote. 84 Said they did not plan to vote.
91 Said they planned to vote for Mike Pence.
72 Said they planned to vote for our campaign.
76 Said they had not decided.
Easy. Quick. Effective.
ManfromMiddletown was quick to point out that the Welsh campaign would have been better served by weighting the amount of voters from each county based on population, but that the lack of weighting may actually have helped Pence, rather than hurt him. Regardless, this sort of informal, internal polling has provided the Welsh campaign with a spark that has energized his campaign, and the people within his district. After his speech at the Indiana Democratic Party State Convention yesterday, where he outlined the many reasons why Pence was not only beatable in November, he had to be beat, there were spontaneous calls for district-wide donations, and loud cheers of support.
This is what people-powered politics looks like.
So what can we do to help these proud Democrats across our nation? What can we do to assist Howard Dean in his mission to fight a 50 State Strategy?
Let's make this concept a reality.
This is what ManfromMiddletown has in mind:
We need a poll in every district, and open source polling for every district. If we are going to have a Fifty State Strategy, we have to be honest where we stand. Polling isn't rocket science, and I think that we can both dramatically reduce the cost of polls, and greatly expand the number of districts polled by moving to an open source polling model. Polls of this type are not going to be given the same credence as polls by national professional pollsters, but they can demonstrate where we stand in each district.
The open source poll conducted by MyDD earlier this years demonstrated that open source polling, where the methodology is openly available on the internet is viable. However, the use of a professional polling firm greatly increases the cost of running a poll. I think that the time has come to take this a step further, and conduct opens source polls in all 435 congressional districts conforming to a national template.
The Universe
The average congressional district has 630,070 residents. This is not our universe.
Most states have some kind of publicly available voter registration database, and this should be the basis of any open source poll. This is the group of voters that we want to sample.
The Sample
Lets say that in an average congressional district there will something like 500,000 registered voters. In order to get a margin of error of +- 5% you need a randomly selected sample of 384 voters. As I said earlier, most states have some sort of voter registration database, and most state parties have access to what's called a voter file that contains all the information that you provide when you register to vote. The key is finding a way to randomly select a sample of 384 from registered voters in the district.
The easiest method would be to find a way to assign a unique, sequential, integer to each registered voter in a congressional district, and then to use a random number generator to produce a list of 1000 numbers that produce a sample from the voter database. The reason why the sample list created has to larger than the final sample is so that if a voter is not available, or there is no phone number, etc you can move on to the next call to get your sample. It's very important, that those making the calls speak only to the person they have the name for. Just speaking to
whoever answered the phone can introduce bias (in male dominated households, that can create a bias towards the choice for males, etc.)
The Template
Perhaps the greatest value of doing open source polling in every congressional district is that it provides an objective way to compare races that reflects the political situation of the present rather than the previous election cycle. In order for this to be workable, there has to be a relatively small number of questions that assess the state of race, without pushing those polled one way or another.
The call begins with a short one or two sentence introduction that should be careful to identify party.
Hello, I'm calling from the Open Source Polling Project. Would
you be willing to answer a few questions about the Congressional race
in your area? (Yes, No, No answer)
The first question assesses the poll participant's general mood about the state of the nation.
In general, do you feel that the nation is on the right track, or wrong track? (Right Track, Wrong Track, Don't Know, No Answer)
The second question asks whether they plan to vote.
Do you plan to vote in this year's election? (Yes, No, Don't Know, No Answer)
The third question measures voter identification of both candidates, no party identification is given.
Do you know who (Candidate name, rotate, repeat with other
candidate's name) is? (Correct answer, Wrong Answer/ Don't know/No
answer)
The fourth question asks who they plan to vote for in November. Party
identification is only give if asked for. People who ask for
party identification vote straight ticket.
Do you plan to vote for (Democratic candidate) or (Republican candidate) this year's election? (Rotate candidates names)
The call ends with a thank you.
Thank you for participating today in our poll.
The Cost
By having a national template that has been run in each district we can get a state of the nation at the district level, (which is after all how the House is won). By using people at the local level, this can greatly reduce the cost of running the poll. In order to produce scientifically valid results its very important that the workers doing the calls understand that there is no right or wrong answer (so as not to introduce bias, or argue with the polled voter), and that it is essential that they stick to the script, and talk only to the voter they have the name for. In order to maintain confidentiality, while the raw results would be available at no charge on the internet, anything that reveals the identity of polled should be deleted.
Assuming that it takes 10 minutes for a worker to complete one call after time after disconnected numbers, no answers, and refusal, that means it would take 64 man hours to complete a poll. Rounding that up to 70 man hours to account for the unforseen that means with 6 callers that would take about 12 hours grand total. Ran in 4 hour shifts 4PM-8PM that would take 3 days (for example Tuesday through Thursday.)
While volunteers could be used for this task, I think that paid staff making the calls inspires confidence, and allows you to stay on script better. If each caller is given $150 for their efforts, that translates out to roughly $10/hour if you include a short session introduction to polling to ensure sticking with the script, and proper polling techniques. So for an individual campaign, that translates out to $900 for a poll that gives a state of the district.
Spread over 435 districts that translates out to just under $400,000. While the framework. I've given is for implementation at the district level, there's nothing preventing this from being done at the national level (DNC). And there are bound to be economies of scale that result from a centralized approach. Either way, this provides a basic, but scientifically valid benchmark from which to run campaigns, and prioritize the allocation of campaign resources.
If we're serious about running a Fifty State Strategy, then the expense of conducting polls in every district is worth the cost. And if the Democratic Party publishes the results of polls it conducts, we can call for Republicans to do the same. And the chimera that Republicans have constructed with carefully crafted media messages and outright lies can be taken down once and for all. They can't say that they're winning when they're to scared to release the results from their own polls.
Regardless about how you feel about the idea of doing this on a national basis, the possibility of these sort of polls helping second-tier, underfunded candidates is truly exciting. This is something that I believe the DKos community can bring into a reality, and that is why this diary is a call to action.
I would love to hear some feedback on all of this. There are plenty of you out there who are experts on polling, website creation, grassroots organizing, and political campaigning. In the least, I think this idea deserves its own website, where second tier candidates can utilize templates and tips to organize their own open-source polling, but I know that many of you might have your own ideas about this.
So let's hear it...what can we do to make this a reality, and help support Dean's Dream of a national party once more?
(special thanks to ManfromMiddletown for these exciting ideas.)
**UPDATE**
Downthread, grapes mentions the idea of possibly contacting the Pew Research Center in regards to sponsoring this as a truly open-source public opinion project. TeresainPA mentions DFA as another possibility.
Does anyone know anyone inside of these organizations, or have information on how possible this may be? Any thoughts?
Thanks.
**UPDATE II**
I just want to summarize what we are talking about here, as the length of the post doesn't lend itself to casual browsing.
The idea of this project would be to create an open-source, completely free system by which campaigns and candidates can organize localized polling operations within their districts. This will allow local candidates, as well as second-tier national candidates, the opportunity to have polling data where they might otherwise not have access due to financial limitations. This could help remove the stranglehold the established polling firms have, by providing small candidates with packaged, open-source system they could use to organize their own informal polls. The idea would be to have it be as scientific as possible, providing both good polling data, and the ability to release 'internal polling numbers' to the media.
This would empower local candidates to use the people-powered politics that are supported on this site, and turn their volunteers into a fully functional polling center, for fractions of what it would cost nationally.
It wouldn't be a flawless poll, but it would infinitely better than nothing. Who knows how many districts could be competitive right now, but we just don't know because of a lack of polling data?
There are three things you can do right now to help get this project started:
- 1) If you are a candidate, or involved in a campaign, consider running one of these polls yourself. This will help to provide more concrete evidence that the concept works. My email address is provided below, so if you are going to do this, please let me know.
- 2) If you are an individual who is interested in helping with this project, please email me at the address below, and I will compile a list of interested parties.
- 3) If you think this concept is worth its weight in snot, feel free to give this diary a Rec, so we can keep it up throughout the day to gauge interest and gather ideas!
Email: denymyfreedom AT gmail.com
Thanks Again!
p.s. If anyone has ManfromMiddletown's email address, can you give it to me? I have been looking for it everywhere, and I just can't seem to find it.