GOTV - Part 5 - Need for a Longer Time Horizon

by: Grebner

Fri Nov 26, 2010 at 05:49:59 AM EST

In order to encourage someone to vote, we stick a leaflet in her door, if she lives in a particular class of precincts. If her name appears on a particular list, maybe she gets a call from a volunteer phonebank.  If she says something in response, unless it fits into the handful of specific categories the volunteer has been told to record, the voter's comments are simply ignored. 

Without thinking, we have constructed a GOTV system that deals with individual voters almost solely in brief stand-alone contacts that are each nearly anonymous.  Where did we lose sight of the fact that voting, or not voting, is a decision made by each voter in the complicated context of their entire life?  Why do we believe the most cost-effective interventions can be broken into tiny, atomic, fragments?

[Note - My GOTV essays, plus many more, are collected under Technical Politics]  

GOTV - Part 4 - Marginal Costs

by: Grebner

Sun Nov 21, 2010 at 21:49:32 PM EST

You can skip today's lecture if either 1) you already understand the difference between marginal and average costs, or 2) don't feel any need to understand the arithmetic behind my arguments.  For anybody in the middle, I promise this post will be the low point of the series.

[Note - My GOTV essays, plus many more, are collected under Technical Politics] 

GOTV - Part 3 - Resources

by: Grebner

Tue Nov 16, 2010 at 04:00:42 AM EST

If GOTV is an enterprise, in need of optimizing, we should search for methods which transform our limited resources into the greatest possible amount of output.

But volunteers make up a major part of our "resources", and they stubbornly resist being treated like "inputs" and insist on being treated like human beings.  Unlike a railroad carload of potatoes, if we treat our volunteer "resources" badly, they walk away.

The other major resource - money - comes with a web of restrictions on how and where we can use it.  Money provided by an organization almost always comes with legal or political strings attached.

You can think of my entire thesis here as an argument for breaking those strings (to the extent we can) in order to liberate the activity from excessive control by history and tradition.  We need to convince the people who control the money that they will never get their money's worth as long as all the details are dictated by people committed to preserving the old ways and afraid of change.  It's as if we need to fight the industrial revolution all over again. 

GOTV - Part 2 - Ceremony versus Enterprise

by: Grebner

Sun Nov 14, 2010 at 19:13:51 PM EST

Human activities can be classified in many different ways, depending on what you want to analyze.  If we think of GOTV as primarily a "ceremony", it's easy to see the roots of many failures and inefficiencies.

[Note - My GOTV essays, plus many more, are collected under Technical Politics]

GOTV - Part 1 - Why we do it this way.

by: Grebner

Thu Nov 11, 2010 at 22:29:24 PM EST

I plan to write a series of posts, exploring GOTV practices and arguing for change.  First, let's consider conventional wisdom and how it arose.

This first post is somewhat polemical, intended to frame the problem, but not really quantify it.  If I oversimplify and caricature, please forgive me.  Subsequent writings will be more focused and factual.

Comparison of Democratic and Republican absentee programs

by: Grebner

Sun Oct 31, 2010 at 22:50:12 PM EDT

The odds are stacked against seeing a perceptive comment in a Detroit News story.  For one thing, the News doesn't WANT anything that distracts the reporter from filling in the stereotyped story assigned to them, so if a source offers a genuine insight, it's treated as irrelevant.  Second, they generally assign reporters who don't understand the topics they supposedly cover, so real ideas are likely to go over their heads.  And anything that makes it past these two hazards is sure to come to the attention of a copy-editor who will garble it to fit their party line.  

So when you do spot something interesting in a News story, the most likely explanation is that the words happened to be randomly rearranged into an order that spuriously suggests actual intelligence.

Which explains a paragraph they printed Thursday :

Snyder's huge lead among absentee voters is significant because the Bernero campaign appeared to have no mailing campaign to assist and encourage absentee voters, said Eric Foster, president of the political consulting firm Foster McCollum White and Associates.  

Faux Tea, Chapter Sixteen, "Grand Jury Days".

by: Grebner

Wed Sep 08, 2010 at 20:12:10 PM EDT

There may never be any answers, but it won't be for lack of questions.  The Oakland Circuit bench has established their one-person grand jury, to kick the fallen Faux-Tea folks while they're down:

 http://detnews.com/article/20100908/METRO02/9080425/One-person-grand-jury-to-probe-alleged-Oakland-election-fraud 

Circuit Judge Ed Sosnick is  69 years old, so he probably doesn't have any political ambitions beyond retiring and then being allowed to cash his pension checks in peace.  He's a Republican-leaning independent, judging from his record of campaign contributions.

AV ballot return rates in the primary

by: Grebner

Mon Aug 30, 2010 at 01:50:13 AM EDT

I have clients who hyperventilate ten weeks before each election, calling me with the rumor that somewhere, some County Clerk has passed at least some AV ballots to the local Clerk, ready to be mailed.  "WE'VE GOT TO GET OUR MAIL OUT!" they scream at me.

I've never been able to convince anybody that - even if the rumor was true, and applied to the Clerks within the correct district (which is never the case) - that most people don't vote their ballots right away.

I've learned not to fight it.  Once I recognize the signs of candidate-psychosis, I know that facts don't matter, advice may be asked but is always ignored, and the only relief will come a few hours after the close of the polls.

For the few people who care about this subject, I have assembled some additional tables to ignore, based on the dates ballots were requested and returned in the August primary.  The data is derived from the Secretary of State's AV ballot tracking system, which is generally correct, but has its share of inconsistencies and glitches.

Faux Tea Party - Rejected by Repubs on Board of Canvassers

by: Grebner

Mon Aug 23, 2010 at 17:33:08 PM EDT

It's only the next scene of the continuing farce, but the two Republican members voted against certifying the Tea Party petitions as having sufficient signatures.

Previous cases have held the Board of Canvassers is NOT supposed to consider the circumstances under which petitions were circulated, nor the merits of the proposal under consideration, but solely determine whether a sufficient number of voters properly signed.  (Recall Doyle OConner, Ward Connerly, and the so-called Michigan Civil Rights Initiative.)

I don't know how this will play out, but I have a general hunch Faux-Tea won't finally make the fall ballot, because eventually the Supreme Court will find a way to thwart Mark Brewer.  (Who isn't involved, of course.) 

Understanding the August Primary

by: Grebner

Mon Aug 16, 2010 at 03:21:28 AM EDT

I've been trying to figure out how to understand the governor primary results, and this is the best I can do.  

I dialed 2000 likely Democratic primary voters (I had to guess who had voted, since data showing the actual voters won't be available until next month) and asked them which primary they had voted in, whom they had supported, which party's primary they usually vote in, and whom they intend to support this fall.  I got about 130 usable responses.

I combined that data - it may have been scanty in quantity, but I've always said scanty data beats hypothetical data every time - with turnout figures, and my linear models of voter behavior.  The result is a sort of impressionistic understanding of how each candidate got the votes they did.  I should emphasize that the small sample size, combined with the likelihood of misreporting, make my conclusions somewhat shaky.

All I claim is this sketch is plausible and at least as well documented as anybody else's. 

Absentee Ballot Return Rates

by: Grebner

Tue Aug 03, 2010 at 16:43:30 PM EDT

I'm calling it: Bernero

by: Grebner

Thu Jul 29, 2010 at 01:25:12 AM EDT

Unlike Bill Ballenger, who had the courage three days ago to say Virg was going to win, I waited until I had a few additional numbers.

For the last week, it's been clear Virg was picking up outstate; we were seeing him pull votes from "undecided" in a number of client polls where the governor's race was just incidental to the real business.  But having a lead outstate wasn't (for me) enough; I had to be sure Dillon wouldn't run up an offsetting margin in Oakland/Wayne/Macomb.

So I ran me a little robo-poll of likely Dem primary voters, limited to the tri-county area, and this is what I found:  61 (Virg) to 36 (Andy) with 29 undecided.  (As always, those are raw responses, not percents.)  Small numbers, but statistically distinguishable from a tie.

Faux Tea Party - how will it work?

by: Grebner

Thu Jul 15, 2010 at 00:31:16 AM EDT

Possibility of spurious polling on Kos

by: Grebner

Wed Jun 30, 2010 at 01:13:04 AM EDT

As more and more polls appear, the question arises:  exactly who is checking their legitimacy?  The answer, unfortunately, is that nobody is. Last year, a firm called Strategic Vision, LLC was discovered to have essentially fabricated a huge number of supposed samples in dozens of critical races around the country.

Now, Research 2000,  until recently the house polling firm for DailyKos, has been called into question.  The paper speaks for itself:

 (http://www.dailykos.com/storyonly/2010/6/29/169/32552)

In brief, we looked at some 75 published polls, and examined how much statistical variation they contained compared to the minimum expected by statistical theory given their sample sizes.  The Research 2000 polls showed too little fluctuation, were too consistent to one another, and generally showed signs that they were not properly calculated or were not based on actual interview data.

"Tea Party" petitions - I'd bet against them.

by: Grebner

Mon May 17, 2010 at 22:39:50 PM EDT

"Somebody" is paying to circulate petitions to place the "Tea Party" on Michigan's November ballot.  Chet Zarko was apparently the first to notice them, and now the story has also been picked up by the Freep, DetNews, and MIRS.

Who's behind it? I'm betting on "the usual suspects". 

More about nicknames

by: Grebner

Thu May 13, 2010 at 21:46:15 PM EDT

What do these people have in common?

Mike Bouchard (R)
Mike Cox (R)
Tom George (R)
Pete Hoekstra (R)
Rick Snyder (R)
Virg Bernero (D)
Andy Dillon (D)

Some good news on the ethnic conflict front.

by: Grebner

Mon May 10, 2010 at 00:22:23 AM EDT

For many years, while picking through election results, I've noticed that Carl Levin's weakest performance relative to the rest of the Democratic ticket has been in certain precincts in eastern Dearborn, and Dearborn Hts.  It's never amounted to very many votes, but it was consistently true that he didn't get quite as many votes as I would expect wherever there were pockets of Arab-Americans.

To my surprise, I can't find any trace of that phenomenon in the 2008 precinct results.  The comparison is fairly tricky, since the six years between Levin's appearances on the ballot gives time for changes to precinct boundaries, not to mention ordinary demographic changes. I don't know whether my evidence will convince anybody else, but I don't have any doubt what I'm seeing.

More obscure discussion of modeling voter preferences

by: Grebner

Sat Jan 23, 2010 at 01:24:46 AM EST

A couple weeks ago, I reported a little bit of research that showed that knowing the partisan composition of a voter's former neighborhood provides useful information beyond what we get by considering their present neighborhood.

Today, I present evidence that knowing about former household members is similarly useful.

Modeling Party Orientation at the Level of Individual Voters

by: Grebner

Sat Jan 02, 2010 at 19:24:03 PM EST

I'm reporting an interesting finding in an obscure and narrow field: predicting an individual's partisan orientation from publicly available information.  I'm afraid most ML readers may find it of little interest, but I'm taking advantage of this forum to make the information available to the handful who may wish to link to it.

To state the result briefly, I've just completed some research which shows that voters' partisan attitudes are predicted not only by the partisan composition of the precinct in which they currently live, but also by precincts in which they previously lived.

The ACORN scandal - explained.

by: Grebner

Sat Dec 26, 2009 at 04:43:50 AM EST

I guess ACORN is never going to become old news, so maybe it's worth explaining what really went on.  My insight comes from actual contact with ACORN and also from running a very similar voter drive for the last 30 years, which once had very similar problems.

Basically, there are two ways to run a voter drive that is based on paying low-wage canvassers.  Let's assume we have $10,000 to spend in a mid-size city, say Jackson or Battle Creek.

Plan A:  Hire a bunch of unemployed teenagers.  Tell them how the forms are supposed to be filled out.  Pass out clipboards, pens and forms and assign them to various "territories" - door-to-door, downtown street corners, bus station, et cetera.  Supervise the process either with volunteers, or with people whose real jobs are only tangentially related.  If you pay $1 per completed form, since you have virtually no overhead, at the end of the drive you can announce that you've registered 10,000 people - increasing the total number of names on the City's voter rolls by an amazing 15%

Plan B:  Spend 30% of the available cash on supervision.  Review each form that comes back in the presence of the canvasser, and only pay for the forms that meet standards.  Emphasize that they'll only get paid for NEW registrations - anybody already registered at the right address doesn't count.  Require the canvasser to go back out (or phone) to correct forms with missing or incorrect information.  Immediately fire anybody who is caught cheating.  Paying minimum wage, plus $1 per valid form, you'll struggle to collect 2000 valid forms.  You'll have a LOT of turnover, as canvassers chafe at being told what to do and they discover finding and registering actually unregistered people is hard work.  At the end of the drive, you'll read in the paper that the "other" drive in town was a lot more successful than you were.

The "right" method depends on your ultimate goals.