Mark Grebner is a political consultant from East Lansing. He made the following comments in regard to the effort to recall Governor Rick Snyder, but some observations are pertinent to any petition drive. The comments were posted August 20, 2011 under the title "Arithmetic of Recall Petition Signature Gathering" on the site michiganliberal.com.

I've been quoted as saying it's essentially impossible for volunteers to gather enough signatures to force a recall election for the governor. I'm sure it sounds as if I'm disparaging the volunteer spirit, or the importance of commitment in bringing political change. But I was really just trying to make a technical point: collecting signatures for a recall is different from other kinds of petition drives. I'll try to explain.

The basic problem is the fraction of signatures needed: 25% of the gubernatorial vote. (This post refers entirely to Michigan law; it applies to a greater or lesser extent in other states which permit recall elections.) Understand, it's not the number of signatures needed, but the fraction. That is, an all-volunteer effort might be able to collect 160,000 signatures statewide to place an issue on the ballot - 5% of the gubernatorial vote - if they were sufficiently motivated by the underlying issue. But if that same organization tried to collect the 22,000 signatures needed to recall a state senator, they would fail.

That doesn't mean recall campaigns are impossible, only that they can't be all-volunteer. In order to succeed, a substantial amount of money would be needed, much of which will be needed to pay some of the circulators.

I'll try to explain why.

Let's define some ideas. First, for every potential recall campaign, some percentage of the registered voters will simply refuse to sign, when given the opportunity. In the case of an extremely unpopular official (say, caught in a financial scandal) the percentage willing to sign might be as high as 90%. (Some people are hostile to recalls on principle, while others are afraid that people like me will gather their names and commit nefarious acts upon them. I've never figured out what that would be.)

For somebody who is run-of-the-mill unpopular, as Rick Snyder is today, the percentage willing to sign when presented a chance is around 60%. For somebody who is quite popular and represents a district of congenial partisan balance - such as Gretchen Whitmer - the percentage willing to sign might be as low as 30%. (Some people will sign almost anything presented to them.)

Every petition drive starts with a committed core of people, who sign the petition themselves, and then circulate it among their immediate circle of friends, neighbors, co-workers, and fellow members of like-minded groups. Collecting those signatures is easy, and helps create a feeling that the whole process will be manageable. But the reality is that only a small fraction of the total electorate falls within these circles; the larger the group of volunteers, the more they find themselves overlapping socially and geographically.

After taking the low-hanging fruit, volunteers turn their attention to public events, rather than door-to-door. They overwhelmingly prefer to collect their signatures by sitting at tables or circulating with clipboards at public places. There are a number of reasons for this. First, door-to-door work is strenuous. Second, door-to-door doesn't initially produce as many signatures per hour as a public event. Third, if any significant number of volunteers actually started canvassing their own neighborhoods, they would quickly trip over one another, since people sufficiently enthusiastic to serve as volunteers tend to cluster geographically. Unless there's a strong hand coordinating the areas canvassed - and volunteers don't react well to strong guidance - each person will tend to choose a small nearby area without regard to what has already been done. Fourth, it's essentially impossible for volunteers to keep detailed accurate records, since doing so requires training, close supervision, and effective feedback. Without detailed records, it's impossible to direct the effort away from areas which have already been worked and toward areas which are untouched.

As a result, we can assume that after collecting signatures from their immediate friends and neighbors, an all-volunteer effort will confine its activity almost exclusively to collecting signatures in public places. If they do, they soon discover that the pool of voters who appear at such places - and are willing to stop to sign petitions - is limited to perhaps 50% of all voters. If half the eligible voters don't come to such events, and many of those who do aren't willing to sign, the total potential number of signers at such events is roughly 25% of the registered voter pool, or about 1.5 times the number of signatures needed for a recall petition.

At first blush, that number seems auspicious - just convince EVERYBODY at EVERY event to sign your petition, and you'll have enough. But there are practical problems with working public events.

First, because recall petitions must be divided by township and city, a substantial number of signatures will be invalid because they will be written onto the wrong petition, especially if the petition sheets are overseen by untrained volunteers.

Second, it's not that 50% of the voters go to every event, but to some event, which means you need to cover many events over an extended period of time - which can't be done consistently with unpaid volunteers.

Third, as the drive continues into a second and third month, a larger and larger fraction of the signatures gathered will turn out to be duplicates - people who have already signed at some other location. This is particularly pernicious because under Michigan law, turning in the same name twice results in NEITHER signature being counted, so if you reach a point where 10% of your signatures are duplicates, for every 100 signatures you gather, you're only adding 80 net to your valid total.

Fourth, as you continue to work public events, your volunteers will begin to perceive that "everybody has already signed", because a larger and larger fraction of people will tell you so, or tell you that they've already refused. The efficiency of collection drops by 30% and then 50% as you pursue the few remaining fish in your pool.

If 10% of all signatures gathered turn out not to be registered (which is typical) and 10% are collected on the wrong sheet (because the voter doesn't know their township or city) and 20% are rejected as duplicates (meaning 10% sign twice), you'd have to gather 1.24 million raw signatures to have 800,000 valid ones - with no margin for error. Collecting 800,000 raw signatures would take an immense effort, but going beyond that number would become progressively harder and harder.

The result of these factors is that the practical maximum for signature gathering at public events is somewhere around 10% to 15% of the gubernatorial vote, or roughly half what's needed for a recall petition. As you approach 15%, everything combines to slow you down: the duplicate rate rises, the volunteers become frustrated, the crowds become less sympathetic, and the daily collection rate plummets. Taking the obvious steps like adding more people, covering more events, or staying longer, doesn't solve the problem.

The only solution I know is organized, door-to-door coverage. And that isn't a solution that can be implemented by pure volunteers.

And here is more from Mark, posted August 21, 2011 under the title "How to Recall the Governor":

There's a long joke about a bunch of academics shipwrecked on a barren island. To skip the body of the story, the punchline has the economist declare that he's found a simple solution to their predicament: "First, I assume a boat...."

I know how to collect enough signatures to place a statewide recall question on the ballot: "First, assume someone gives us $10,000,000..."

If anyone is interested in the details, here they are.

I've been involved in quite a number of recall campaigns - on both offense and defense - and I've seen what works and what doesn't. As I've argued previously, because of the requirement to collect the signatures of such a large fraction of the voting population, recall campaigns are fundamentally different from other petition drives.

For a recall to be successful, it's necessary to have close supervision over the circulation of the petitions and immediate feedback about problems to the circulators. The first thing a statewide recall campaign would need would be office space where the blank petitions would be stored, handed out, returned when filled out, and checked. Imagine maps taped to the walls, computers with data entry workers searching and flagging the names of signers, a separate area for training, and waiting areas for drivers and circulators. Maybe there'd be seven locations statewide, of varying sizes. Cheap, used furniture, with some basic amenities like coffee and restrooms.

Second, because so many people would be on payroll (often for small amounts) it would be necessary to have employment forms and somebody to hand them out, check them, and file them. Payroll would probably be handled by a contract firm, in order to minimize the problems associated with cash-handling.

Third, because each petition would be checked within 24 hours of being turned in, there would be about 25 data entry stations scattered among the headquarters, each with one or two assigned workers. Data entry workers would earn varying amounts depending on their productivity, but would each cost close to $1000 per week including fringes.

Fourth, because of the substantial data processing infrastructure needed, and because of likely security threats, there would be a small MIS staff - maybe 5 total - who would handle setup, data manipulation, backup, trouble-shooting, and training the data entry staff. These folks will be fairly highly paid.

Fifth, there would be large number of people dealing full-time with the circulators: people to train newbies, to assign circulators to locations and times, to assemble groups of circulators and drivers, to record who has been assigned where and when, to answer questions that arise in the field, to check over the petitions and other materials when people return, and to record what areas have been completed. I would imagine at least 50 full time equivalents statewide, a mix of volunteers and low-paid staff.

Sixth, it would of course be necessary to print petitions; buy clipboards, pens, file folders, plastic bins and dozens of other minor office supplies; supply plastic rain ponchos and umbrellas, buy or print tens of thousands of maps, and so on.

Seventh would be the central management of the campaign, responsible for controlling the money, handling press relations, dealing with legal problems, and - most important - making daily tactical decisions in response to events.

These first seven requirements could be called "overhead", and would probably cost $1.5 to $2 million over the course of a 90-day drive. But the largest expenses would be the final three:

Eighth would be paying wages and mileage to an army of driver/supervisors. As PPC has discovered, most of the people who are available to circulate petitions simply don't have reliable transportation, and many of them don't know their way around the region. On the other hand, there are lots of people who are willing to drive (and even supervise others) who don't care to knock on doors themselves. The advent of ubiquitous cellphones has made it feasible to send out small teams of canvassers, each under the supervision of the driver who actually brings the forms back to headquarters at the end of each shift. Depending on circumstances, some of the drivers also canvass part-time.

Ninth, and largest, are the wages of the canvassers themselves. I envision starting at $1/signature (or minimum wage) for the first few weeks, and then gradually escalating the pay to $2, $3 or possibly even $5 by the end. There are several reasons for the escalating pay, but basically it will be easier to collect signatures at the beginning and it will get progressively more difficult as the drive goes on, both because so many people will have been previously approached and because canvassing will move to less and less productive territory. The actual cost per signature will be about 30% higher than the nominal cost, given various employer payroll costs.

Tenth, will be payments to people collecting signatures at random, rather than door-to-door. In general, we'll try to cover events such as high school football games or farmers markets using volunteers. But as the drive wears on, and it becomes clear where volunteers are insufficient to cover targeted events, it may be necessary to begin paying.

If everything goes well - which would be a surprise in a recall campaign - the total cost to collect 900,000 valid signatures would be about $6 million. The most important assumptions behind that estimate are that 1) 60% of people approached would be willing to sign, and 2) that unpaid volunteers would collect 250,000 valid signatures. If public resistance is higher than I expect, or volunteer enthusiasm weaker, the only solution with which I'm familiar is the one I call "Plan B", which is to bump the going rate by $1.