GOTV Design – If We’re Serious.

by: Grebner

Sun Feb 17, 2008 at 01:47:45 AM EST

Our Get-Out-The-Vote drives are driven by folklore, group-think, and doing what comes naturally, rather than being the product of careful design.  Especially in a high-turnout environment, where almost everybody who falls within the obvious categories is already voting, our efforts and money are targeted so inefficiently they make virtually no difference.  

An observer dropping in from another world would deduce that our goal was to provide our volunteers with tasks to keep them busy, while finding ways to spend all the available money.  It would never occur to them that we were trying to affect the outcome of a political election.

Some of the conventional tactics might be effective in a low turnout election.  But in a very high turnout environment, why do we spend election day “reminding” people to vote?   Our targets are all people who have land-line telephones, whom we have previously determined are likely to vote Democratic.  Is it conceivable that a significant number of such people will have literally forgotten Election Day?

Beyond reminding, we urge people to vote, telling them how important their votes are.  Again – is it likely the people we previously IDed as Democrats are unaware of that argument?  Or, maybe they’ve remained unpersuaded by it – until our contact tips the balance?  Baloney.

Our contacts are limited largely to people who have landline phones, or who have voluntarily provided contact information.  Again – how many such people would fail to vote in a high-turnout election, even if we don’t contact them?  I describe our general approach as one of recruiting people to staff phone banks which then attempt to contact the people staffing the phone bank.  We don’t exclude anybody from our lists because they’re too likely to vote, so we activists can find plenty of targets on whom to spend our efforts and money: ourselves.  Unfortunately, the genuinely marginal voters aren’t on the list.

Our efforts have virtually no real effect on turnout, but we are spared from hearing that because nobody ever tracks down the information specifying exactly who finally voted.  Except for a few academics, nobody sets up “control groups” to allow GOTV drive impact to be quantified.  Instead, we wait two years for the next election, at which time all the “old-timers” re-appear to tell their stories about how successful such-and-such a tactic was, and how they suggest it could be improved.  All decisions are based on stories, the wisdom of committees of volunteers, and the inscrutable directives of the badly managed state Coordinated Campaign.

If we’re trying to increase the number of Democratic votes, in order to win elections, we need to target people who are less stable, less organized, less knowledgeable about politics, and generally harder to identify and locate.  These are generally people who do not think about politics very much, whose opinions may seem unreasonable, who have personal problems that monopolize their attention, and who may be only one step ahead of creditors or police.  In general, it isn’t easy to find these genuinely marginal voters on our lists.  Contrary to the conventional wisdom, they are not particularly likely to show up on our lists as “one-election-of-three” voters or whatever.  Instead, if we ever pursue them, it will be almost one-at-a-time, through painstaking research, contact, and re-contact.

The first thing we need to do, if we’re going to re-engineer our GOTV efforts, is to lower our expectations.  Our past “accomplishments” have been exaggerated to absurdity by bragging and the passage of time.  “We could see we weren’t getting enough votes out of the Third Ward, so we sent extra canvassing teams to do a second sweep – and we ended up winning by a thousand votes!.”  And other such fantasy.

Realistically, if we could raise overall Democratic turnout by two percentage points statewide, we would be doing very well.  With unlimited funds and volunteers, and incredibly efficient management, the ceiling might be 5 percentage points.  I would guess our current efforts – if rigorously compared to preselected control groups – are responsible for no more than 0.5 percentage points – and possibly much less.

We need to identify specific classes of individuals, and then get to work finding the names and addresses of individuals who fall within them.  We need to contact them, starting long before election day, to begin the process of figuring out what obstacles exist and what needs to be done to surmount them.  Our current efforts remind me of Christmas caroling:  a mass activity mainly done for the benefit of the participants.  What we need to do is much more like home health-care:  checking up on vulnerable people, establishing relationships of trust, and taking care of specific problems they face.

So who are these people?  The following are examples, not a complete list.  But each of these will strike the experienced GOTV volunteer as “not the way we do it”.  Each category presents specific difficulties, which invite shortcuts that would allow the drive to get back into its comfort zone.  Those shortcuts must be fought, from trench to trench, unless we’d rather continue down our present ineffectual path.

1) Prisoners in local jails.  Contrary to public belief, even convicted felons can vote in Michigan, as long as they are not currently incarcerated serving a sentence.  Since half of all jail inmates are technically awaiting trial or sentencing, they are eligible to vote.  The problem isn’t a legal bar to obtaining absentee ballots for them, but the painstaking work needed to figure out which ones are registered to vote, and then to identify someone on the outside (likely to be a regular visitor to the prisoner) who can carry the application and get a signature on it.  Another problem is the high rate of “churn” – people are booked and released constantly, and many are only in jail for a few days or weeks.  A third problem is the typical inmate’s utter lack of political interest or knowledge.  A fourth is their typical low level of literacy and inability to follow directions.  The only “plus” is that typical inmates – if they somehow manage to vote – are at least 80% Democratic in orientation, so there isn’t much point in screening for partisan orientation.  In the two months preceding the November election, at least 20,000 people will spend at least 7 days in jail in Michigan.  If we could get 10% of them to vote – virtually none do so on their own – we’d be adding 2000 votes statewide.

2) Residents of mobile home parks.  Aside from a few middle-class properties, residents of mobile homes have extremely poor turnout, for various reasons.  One important problem is that the management of the parks often try to prevent outsiders from conducting political activities.  Another is that long-term residents are often beset by personal problems, which include legal, substance abuse, credit, employment, and family troubles.  The only successful approach is to deal with each park on its own, figuring out how to co-opt or evade the management,for example.  Carrying a voter list door-to-door will permit a canvasser to identify people who are already registered (who may not remember they are) and register those who aren’t.  (Contrary to common belief, explicitly partisan voter drives are now legal in Michigan – the law changed in 1996.)  Rather than sending different people through for each contact, each mobile home park should have a single coordinator, so residents have a chance to develop a personal relationship to an individual who knows them and their situation.  That coordinator, perhaps augmented by a handful of volunteers working under their direction, needs to commit a substantial amount of time each week, from August through election day.  A 200-unit park, with 280 adult residents, might yield 40 additional Democratic votes, requiring ten hours each week for 12 weeks.  We could reasonably hope to create 5000 additional Democratic votes statewide, if we cultivate all the easily accessible parks.

3) College students in college towns.  The problem with college students isn’t “turnout” but “registration”, which is a knotty legal problem.  We’ve completely solved the problem in East Lansing – look at our returns for the 11/04 election, if you don’t believe me – but our methods are very time-consuming and difficult to export.  Essentially, it’s necessary to build an accurate database of names, addresses, registration status, and political views, and then aggressively solicit the Democrat-leaning students to sign pre-printed voter registration cards we carry.  We’ve added 5000 Democratic votes to the rolls at MSU, which suggests we might be able to add another 10,000 at other campuses, if there were sufficient resources.  But half of these votes are actually only transferred from somewhere else – many students would have cast absentee ballots if we had not come around to make things easier for them.  5000 additional net Democratic votes is plausible statewide.  The cost would be roughly $100,000 using the tactics we employ in East Lansing.

4) College students voting by absentee ballot.  There are over 1 million identified Democrats in Michigan, but as I argue above, they’re not generally worth much GOTV attention, because for most of them, their natural turnout rates are very close to 100% in a presidential general election.  One important exception is young voters (say, ages 18-26) who are registered to vote at their parents’ address.  Rather than waste our time “reminding” middle-aged, middle-class, Democratic homeowners, we should focus on the ones who appear to have have college-aged children.  Where are the kids living?  How can we contact them?  Would mom promise to take care of getting an AV application signed if we mail it to her?  In this case, combining the AV contact with mailed campaign literature (including down-ballot races where the student may otherwise never hear a word) is valuable.  This operation will require a huge amount of data entry, list matching, individualized printing and mailing, and detailed follow-up.  Many of these students will be living out-of-state, and the absentee applications need to be individualized for the specific township or city where the parents reside.  The potential number of additional Democratic votes is hard to pin down, but is probably over 25,000 statewide. Even with substantial volunteer assistance, the cost will run into hundreds of thousands of dollars.

5) “Additional household contacts.”  This category is difficult to explain generally, but it’s easy to understand specific instances.  Instead of wasting our time – and theirs – by reminding solid Democrats to vote, we should use our contact to interview them about people they know who aren’t likely to get to the polls on their own.  A mentally ill relative?  Somebody in trouble with the law?  The next door neighbor?  An almost-18-year-old high school student who hasn’t registered yet?  24-year-old non-college kid living in an apartment nearby?  The interviews need to be conducted by people who are skillful and well-trained, combining access to voter lists (including voting history) with open-ended questions intended to recruit the “index” voter (the one who is already a solid voter) to be an extension of our GOTV campaign.  In general, the index voter should be asked to help with exactly ONE person, and their willingness to assist should be formalized and followed up with additional phone and mail contacts.  It’s obvious this effort could absorb any amount of resources poured into it.  The emphasis should be on building a smaller number of high-quality/high-intensity relationships, rather than spreading the effort to a large number of people who listen to mumbled recitals over the phone of “how much we appreciate your help with [fill-in blank] thank you very much…”.  Lots of vague urging will not cause anyone to vote who wouldn’t otherwise.  A small number of specific, intrusive, pointed contacts will get a small number of additional people to the polls.  If $1 million were available for such an effort, we might hope for 10,000 additional Democratic votes, if we successfully fight off the kinds of “improvements” which would suggest themselves to any committee.

6) People who have moved recently.  We often submit voter lists to be compared with the USPS “National Change of Address” file, in order to remove people who have moved away.  But far more important is to track down known Democrats who have moved but not yet re-registered –  and work with them to re-register.  Some will take care of this on their own, as they update their drivers license, but many others will fail.  “Urging” them to re-register is a waste of time – people who are sufficiently motivated will take care of it on their own, so we’re only concerned with the un-motivated people.  We need is print and fill out the necessary forms, and hand-carry them to the household, so they understand we think their vote is important and that we’re aware of them.  For people who miss the registration deadline, the law allows the use of absentee ballots, and also allows them to return to their old precinct under certain conditions – which the voter will not know unless told.  Finally, we ought to obtain information about moves INTO Michigan by Democrats who had previously lived in other states.  This should be available from the DNC or other Democratic list sources.  The total size of this universe is approximately 50,000 individuals, or 30,000 households.  We would be doing well to obtain 5000 additional Democratic votes from a well-organized effort.

7) “Wannabes”.  One of the great untold stories of American politics is that immediately after each election, 60% of the adults surveyed claim they voted, while only 50% really did.  The other 10% – one-sixth of all apparent voters – has learned to lie routinely whenever voting is discussed.  In a sense, these people represent the triumph of our conventional GOTV methods.  We remind them to vote, and they tell us “thanks for reminding me.”  We tell them how important their vote is, and they agree.  We ask them if they’re going to vote, and they assure us they will.  They lie to their families, their neighbors, to pollsters, and even to themselves.  As a group, they turn out to be just as well-informed as actual voters, and similar in education, home ownership, age, and so on.  Unfortunately, they also tend to substitute fantasy for action; they think of themselves as athletic because they watch so much football on TV.  Additional pressure just rolls off them – they’ve learned that nobody checks their voting record, so all they have to do is give lip-service – which they’re good at.

In order to put effective pressure on these people, we need to identify them, and then let them know in a low-key, personal way that we are disappointed they didn’t vote in election X, but now we really need their vote.  The effect, beyond causing deep embarrassment, will be far more effective than continuing to play along with the game they’ve already mastered.  I’m sure we could identify 100,000 people who are likely to flake out if they don’t think they’re being watched.  Some of them will vote on their own – after all the 2008 turnout will be very high – and some won’t vote even following our contact.  10,000 additional votes is a plausible target.

There are other target groups, but these are certainly enough to illustrate the general ideas.  To summarize:

  1. Contact needs to be made long before the election, not just (or mainly) on election day.
  2. The focus needs to be on specific individuals, including the details of their specific situations.
  3. The contacts needs to be intense and continuing.
  4. The numbers involved are much smaller than traditional GOTV’s target of 2 million or whatever fantasy grips the Coordinated Campaign in a particular year.
  5. Conventional wisdom is not merely wrong, but allowing it to creep in will be destructive.  No committee of volunteers would allow any of the programs to proceed without major and fatal “improvements”.

There’s one final principle:  the names of each target must be unambiguously assigned to specific volunteers and/or staff, who are responsible for turnout of that individual.  And after the campaign, the data must be obtained from the State’s Qualified Voter File to determine who succeeded and who failed to turn out their assigned voters.  The point isn’t punishment, but the need to make sure everyone in the organization understands the seriousness of the new system, and that their role isn’t merely to play along.  This information will permit the next round to be made more powerful by adjusting tactics and shifting personnel to maximize effectiveness.  (My friends at Yale would ask us to randomly chose a few percent of the targets in each category to be held back as a control group.  I’m sure they’d like to incorporate the results in their book.)

I don’t think there’s any realistic chance these tactics will be adopted in Michigan during my lifetime, but I was asked to put my theories in writing – and now I have.