Canvasser Management

Disclaimer

Please do not email me about this project without reading the disclaimer above.

This project is for someone interested in political campaigns. It has a series of computer projects, from very easy to quite sophisticated to manage canvassers. Canvassers are typically volunteers, but they are still a scarce resource and need to be used efficiently. I have done some doorknocking myself and these ideas are based on that experience.

Canvasser Checklist

Any major political party will likely already have a voter/canvasser database. You might want to integrate this feature into it, or export data from it to a PC (Personal Computer) program that either prints the sheets by generating them with Swing or by generating HTML (Hypertext Markup Language) tables. Canvassers get a orientation when the rules of who gets what, but forget once they get out into the field. A low-tech solution is to compose such a form in a word processor with blanks to be hand filled later. A little sheet like this will make sure the official policy is more likely followed. It will also help the volunteer check his kit is complete before he walks out the door.

I have used a -2 .. +2 rating scheme since is obvious that +2 means the voter is enthusiastic about our part rather than the NDP (New Democratic Party) ’s traditional 1 .. 4 which where 1 means most enthusiastic. Imagine the havoc if a canvasser interpreted it backwards!

The Facts Of Life

• People tend to vote identically to their neighbours. Political party choice is highly correlated with income. Further, party affiliations of these willing to divulge them, are quite stable. Campaigning is not so much about changing anyone’s opinions, but identifying where your support is and herding them like cattle to the polls.
• By far the majority of voters are not home, either to phone calls or door knocking.
• Canvassers spend a minute fraction of their time talking to voters. Mostly they walk, scan paperwork, wait for people to come to the door who are not home and try to track down where their fellow team members disappeared to. In my own case, I wasted an embarrassing amount of time because papers and pencils refused to stay put on the clipboard and kept blowing away. A canvasser needs a pouch for the miscellaneous papers to keep them dry, clean and easy to access.
• By far the majority of voters are not willing to give you more than a few seconds of your time, even if they like your party. You have to make the most of those precious seconds by carefully prioritising what you ask. The script must reflect this. Wherever alternate sources of information exist, they should be used before canvassing so canvassers don’t waste any of the time voters give.
• If you call back later in a few days at the same time of day, it will be a waste of time.
• If somebody else comes back later in a few days at the same time, it will still be a waste of time.
• Volunteers are unpaid, but they still require costly management and grooming and coaxing. It is important to use their labour more efficiently than the other parties do.
• Using canvassers twice as efficiently is equivalent to doubling the money you raise to fund the canvassing effort.
• Most voters do not like receiving telephone or doorstep calls. If you call at the wrong time, you could well lose a vote. One guy I canvassed said, Do you realise this is Saturday afternoon? Geesh. Have you no consideration?.
• People are always moving, so your records of who is in the house and which party they support can easily be out of date. Don’t write people off who might be a new set of people in the house. There needs to be ways of keeping track of who has moved, through some other means than revisiting the house and asking.
• Even when there are too many canvass volunteers to manage, there is still not enough time to canvass everyone. You must prioritise. The success rate of a canvassing campaign depends mainly on how cleverly you prioritise.
• Phone canvassing is getting less efficient every year because people are replacing land lines with cellphones and VOIP (Voice Over Internet Protocol). Automatically dialled calls start with a silence, just like all other annoying telemarketer calls. People hang up immediately. If a voter users caller id to figure out the your party has been repeatedly calling and hanging up without leaving a message on the answering machine, that could lose you a vote.
• Door step canvassing is getting less inefficient every year because people are spending less time at home and because they are afraid to answer the door to strangers, and because they feel beleaguered by people demanding a few minutes of their time.

When-Home Tracking

By now I think you can see what I am driving at. It is necessary to track not only successful calls but unsuccessful ones to gradually, perhaps over years, glean when the optimum time to call is, both days of the week and times of day. We need canvassers to record the time of each call, completed or not. Automated phone call placing software for voice canvassers should surely know from they many probes when people tend to be home. Doorstep canvassers should know what that computer knows so they don’t waste time. With a database of such information, (similar to ones burglars keep) you could safely send out a team picking off the night owls only, or shift workers etc. They key point is this, if we knew with 95% certainty a certain voter would not be home on this round, that is not low hanging fruit. The canvasser should get on with someone else. That would result is a much higher percentage of completed calls. Just before the canvasser walked the door of the campaign office, the computer could produce him a super-fresh list of just those people should be contacted right now.

Route Lists

Canvassers are typically not familiar with the territory. They have to ask themselves, Where am I? What building am I in? On what page do find the information about this door in front of me? Am I supposed to canvas that building? I am at the end of a block, where am I supposed to go next? I spent far more time scanning sheets of paper and looking at the map than I did talking to voters on my round today. Mailmen don’t have this problem for two reasons:

1. The mail is pre-sorted in walking route order
2. The mail man is familiar with the territory.

Simply sorting the record sheets in walking route order would solve most of that problem. But how do you go about doing that? This is the second student project.

How do you decide the best way to walk a territory? Basically you canvas one side of the street until you can go no further, then you flip to the other side and keep going. You only canvass the streets inside a given boundary which runs down the center of the bounding streets. There exist digital street maps of each poll region. You could handle this mathematically as a sort of traveling salesman problem to visit all the houses to be canvassed on that pass. It does not matter if the solution is the absolutely perfectly shortest, but it has to be good enough that people perceive the route as reasonable. Probably each side of a block should be considered a single leg. The database may consider some parts of the route as higher priority, so it might put those legs on the early part of the route. Canvassers could pass there experience back to fine tune the routes. You might also define the routes by having someone wave a mouse or digitiser over the map, walking the virtual route. You then have to take that geographical information and use it to label the various streets in the region separately odd/even, so that the computer can assign a walk route sort number to each street address. For proof of concept, you could even assign the order manually with a keyboard. You could even assign the order on a house-by-house basis to deal efficiently, for example, with rear and corner entrances. You might to try the idea out first on a sampling of live canvassers to see how they like the by-route order. A canvasser could ask for it in street-name order or by-route order.

Non Voters

The base canvasser information comes from the government’s voter registration. People who are not yet registered to vote don’t show up. In my whole day of canvassing I think I definitely won over only one vote. I had accidentally gone to a newly-built house with no registered voters. The tradition is to ignore these people since the are not voting. However, there is plenty of time to register and these people are not as likely to be set in their ways. In other words, they are still influenceable, unlike nearly everyone else. I think much more attention should be tracking what is going on in house without registered voters. Sometimes there is good reason, e.g. they are not citizens. Sometimes this reason excludes them forever. Sometimes it just that nobody asked them to vote. So another student project is a database to track these potential voters.

Getting Fancy

Think about the many ways a smart cellphone, with a custom canvassing app, especially one with GPS (Global Positioning System), could be used to make canvassing more efficient.

• It could be used to report findings back immediately. The phone automatically knows the canvasser, location, date and time. All the canvasser has to do it hit a button for his assessment and to press a button to supersize it with a lawn/window sign, ride to the polls request, key issue discussion call etc. The phone would know exactly where a candidate was. This would eliminate errors in recording information for the wrong address.
• This would help coordinate teams so they don’t accidentally call the same location within minutes.
• It would help teams of canvassers keep track of where each other are.
• The phone would know exactly where it was, to it could tell the canvasser in advance the necessary information, up-to-date to the second.
• The information could be more detailed than now, since it would not need to be printed out on reams of paper, just displayed on the screen.
• There would be no need to thumb through papers looking for the information, or running out to front to recheck the addresses.
• The computer could cleverly prioritise which locations to visit on this round, depending on time of day, how tight time is, physical abilities of canvasser to climb steep driveways etc.
• The computer could guide the canvasser using something like a car GPS to the next location.
• To get the absolute best walking route, the computer just has to watch the routes that various pioneering canvassers use and keep the shortest contender. This strategy can even optimise routes inside buildings.
• If canvassers go away for a coffee break and come back, the computer can create a whole new walking route on the fly, starting from where they are now and only going through the houses left to do on this pass, taking into account how more time the canvasser is willing to devote, the priorities of remaining houses and the time to walk to them. Right now canvassers are expected to make these complicated tradeoffs themselves with very little information to make them.
• This thing cannot be too much of a Nazi. Keep in mind canvassers are volunteers. If they don’t want to go to some house, the computer should not push them to or chastise them for refusing. (Big Brother knows if they did). The canvasser should be able to press a pass button without having to come up with an excuse.
• To deal with voters who do not speak English, you must track what language they do speak (guessable from the neigbourhood) and arrange for a canvasser who speaks that language to be assigned that house. This is pretty easy for a computer.
• You might want to track who is soon to become a voter, e.g. 16 years olds, people working toward citizenship and court them specially even before they vote.
• The computer could attach voice notes to a site intended primarily for the next canvasser e.g. Don’t be afraid of the Great Dane next door. He is perfectly friendly.
• It would mean the database would have extremely accurate location data. That lets us map/graph results and project from samplings how the campaign is going and where the volatile polls and addresses are.
• The phone could also be used to get help, clarification, send messages, check on how canvasser were feeling, tell your wife when you will be home…
• Right now teams are pretty much limited to two people. If you add more they just trip over each other and it gets more and more complicated to figure which houses are already done. With computers managing things, you could use huge teams. The spirits would be higher. The impression on anyone viewing these giant teams in action was your party had overwhelming support. They would attract attention of passers by. You would never have to turn away a volunteer just because you had your quota of two per polling region. If canvas were getting behind you could easily send out more people without the canvassers having to coordinate in the new comers. You can move canvassers from poll to poll at any time without disrupting the other canvassers. The large team implies legitimacy and may open more doors.
• The workers managing the canvassers have up to the minute knowledge how the people in the field are doing, how productive, how flagged… This system gives the managers much more flexibility to adapt, e.g. rain in one part of the city, to redirect canvassers where they are most needed, to poll some houses more often than others, or to give new instructions to the field to speed up or slow down the canvas.

Hand Held Canvasser App

In projects like this:

1. First focus on collecting accurate data.
2. Work out a simple UI (User Interface) that you are sure you grandmother could master. Then build a prototype and make sure. Projects of this type often fail from featuritis and the over enthusiasm of young nerds. What the canvassers see has to be dead simple. You can hide all the cleverness you want behind the green curtain.
3. Then later you find ever more sophisticated ways to use your data to help you make ever more optimal decisions. This is the payoff. It is hard to predict the myriad ways you will be able look and at and mine that data. However, anything you do of this form requires collecting verified, clean, accurate, unambiguous data. Skimping here is the royal road to disaster.

Your task as student programmer is to write a call phone Java app to implement some of the above ideas. You also have to write the JSP (Java Server Pages) server-side code to talk to all the canvassers’ cellphones in the field. You might eventually write an HTML version of the interface so that people with lap tops or other non- GPS devices could get some of the benefits. You probably should start with Android and then once it is working thing about branching out. It should run on an a variety of cellphones and laptops, such as iPhone, iPad, BlackBerry, Windows 8 and possibly even the Kindle Fire. You need to find out what your volunteers tend to own or have access to. After you have proved proof of concept with a miniature database for one polling region, you need to work on integrating your work with what has already been done. This will end up being the more difficult part, especially selling your party on the advantages of doing this and investing in this cutting edge technology. In the process you will pretty well have to simulate to some degree everything that has been done before. A finished system has to integrate phone and foot canvassers for example and has to provide all kinds of reporting. All through your design process you must provide hooks to make it possible to integrate your work with what has already been done, with data exchange in real time. If you are interested in selling this as a generic too for use by any party in any sort of election, you need to think about ways to may it customisable while retaining a solid common core. Study up on the use of delegates, (the computer programming kind, not the political kind.)

One the system were shaken down and proven, a party might buy some smart cellphones to lend to volunteers. This in itself would be a perk to get people to volunteer. It is a fun toy. It would be like Star Trek. It would also help establish the party’s prestige as competent in technology.

Mockup

canvaser app mockup

#401 - 1111 Douglas St.
	Newman, Alfred
	(250) 555-1212
	Newman, Gertrude
pass next review
map other

Legend

• increase the rating one notch.
• decrease the rating one notch.
• pass I don’t want to do this house just now, show me another.
• next I have finished marking for this house. Next address please.
• review review past work. You can see every house, or just houses where you made changes.
• map Show me a map of where I am, where others on the team are, where my next house is, and guide me to it.
• other Other functions such as messaging, adding text or voice notes.
• Various ways of getting help, including the button legend, texting, voice calls, 911.

Concerns

• To do this properly requires a fairly large screen. Most people’s cellphone screens are tiny.
• I did not do any analysis of the cost of the phones, insurance, air time… which I just assumed would be borne primarily by the canvassers.
• You must sort out who is liable for lost, stolen or damaged phones, whether they are owned by the canvasser or the party. You must sort out who pays for air time.
• You must ensure canvassers cannot rack up long distance bills, porn calls etc and stick the party with them.
• Some people are technophobic and no matter how simple the system, will claim they cannot master it. If there are a sufficient percentage you will need to accommodate them with paper, however, more intelligently generated paper than before.
• You can’t hide the fact you are using this technology. It is only a matter of time until your rivals catch up. So your edge requires constant innovation.
• Costs of hardware are constantly dropping, so you must plan for hardware and air time costs as they will be for full deployment, not as they are now.

• Le mieux est l’ennemi du bien
  The perfect is the enemy of the good.
  ~ Voltaire (1694-11-21 1778-05-30 age:83) [François Marie d’Arouet Voltaire]

  Beware of trying to plan and implement everything all of a piece. It is best to develop some small part of it and get it into the field. That experience will make you ever so much smarter than you are now to design and build the next step. Things that sound so useful later drop away and other things you never thought of become the top priority. Besides, the goal is to stay ahead of the competition. If you get too fancy too fast, you won’t have any edge at all over the competition for quite a while. See Kent Beck’s Extreme Programming. Arthur C. Clarke wrote a short science fiction story, whose name escapes me for the moment, about the phenomenon. It is part of engineering school curriculums.
• You need a backup plan in case the opposition sabotages your computer system or in case it fails. You want to do sanity checks on when the canvassers tell you so that infiltrators cannot sabotage you with false data.

Go For It

I wrote this student project outline based on my experience canvassing for the NDP but the ideas probably apply in other countries for other parties and causes — anyone who does doorstep canvassing. It would even work for charities or organisations trying to get petitions signed, or for third party canvassers who work for various clients. Though I hope this work advances pro-environmental causes, pro-peace, anti-poverty causes, the ideas are in the public domain and anyone is free to implement them.

standard footer

	This page is posted on the web at:	http://mindprod.com/project/canvasser.html
	Optional Replicator mirror of mindprod.com on local hard disk J:	J:\mindprod\project\canvasser.html
	Please read the feedback from other visitors, or send your own feedback about the site. Contact Roedy. Please feel free to link to this page without explicit permission.
	Canadian Mind Products IP:[65.110.21.43] Your face IP:[3.17.128.129]
Feedback	You are visitor number