Wednesday, November 21, 2012

The Waiting Is The Hardest Part

The best piece yet on, as it's titled, "The Real Reasons You Waited Hours in Line to Vote" comes from Richard H. Pildes, a law professor who worked on the Obama campaign's "vote protection" operation in Virginia.

The second section, on what poll workers face on election day, GETS it. Bears repeating in whole, with my comments and emphasis added.
Voting has become increasingly complex, both technologically and legally. Yet the poll workers who run the process are temporary volunteers paid $100 a day. (Deeth: Typical Johnson County official working a full Election Day would make roughly twice that, about $205.)

They serve episodically and cannot develop much expertise; they tend to be older and less technologically knowledgeable; they are mostly not lawyers, but must adapt, with minimal training, to constantly changing election laws.

On the legal side, poll workers now have to apply state election law for regular voters, the federal Help America Vote Act for those who are going to cast provisional ballots, and absentee-ballot laws for in-person absentee voting. (Deeth: And for Iowa, add another category of rules for election day registrations.) Each category of voting has its own distinct set of rules to be mastered. Every additional layer of complexity creates more capacity to confuse poll workers and slow down the voting process, even if the law, such as the Help America Vote Act, is well intentioned. We have to assess the costs and benefits of these laws more fully. Virginia had also (like many other states) changed its own election laws in 2011; these changes both expanded and contracted the valid forms of voter identification.

When a voter comes to check in, if his or her situation is anything other than the most routine, the process simply grinds to a halt. Poll workers are terrified of making a mistake, not sure of what the law requires, confused, and unclear about how to resolve the situation.

On the technology side, we are in the midst of moving from old-style printed poll books, where the names of registered voters are listed, to electronic poll books. (Same here in Iowa.) In the long run, this change should speed the process up, but for now, there is resistance to this change and panic when it doesn't function properly, all of which further slows voting down. Sometimes the technology doesn't work properly; the electronic poll books won't open up, for example. Poll workers have little ability to deal with these technological problems on the spot. (Deeth: You can throw all the computers and workers you want at the problem, but you need workers who can work the computers.)

Once a year volunteers have trouble making this system run smoothly, especially in the face of constantly changing laws and technology. (Deeth: With all due respect to our seniors, some of whom are very good, but...) Younger poll workers are more likely to be comfortable with new technologies, and we could improve things a bit by creating incentives to encourage college students, for example, to volunteer as poll workers. (Deeth: I've also suggested teachers and other government workers.)
Pildes makes two other main points. One of them is largely addressed in Iowa: "pressure can be taken off the system by generous early voting, mail-in voting, and no-excuse absentee voting." Been there, done that, and Johnson County saw 58% of its voting BEFORE election day. (Election Day turnout was actually DOWN by 570 voters from 2008 to 2012.)

Other counties could still expand their satellite efforts, but every auditor's office in the state is required by law to start early voting 40 days before a primary or general election, and the last two Saturdays are also required. That's a minimum of 30 calendar days of early voting.

The other point Pildes raises concerns resource allocation, and concerns me a bit more. He argues implicitly for more state and national control:
State law does impose minimal levels of machines per capita, but these requirements are so low as to be close to meaningless. In addition, registrars typically allocate the machines they do have based on the number of "active voters" in various polling locations. "Active" means voting regularly every election, not just in presidential election years. Areas that have high turnout in presidential elections, but much lower turnout in non-presidential years, therefore have far too few machines in presidential-election years.
First off, a factual error: EVERYONE seems to get the concept of "active" vs. "inactive" voter wrong. The categories were created under the federal Motor Voter law of 1993. Under that law, no one gets completely canceled simply for not voting. Instead, everything depends on the mail. An "inactive" voter is someone who has had mail from the election office returned to sender. They stay on this status through the next two general elections, unless something changes, and only then do they get canceled.

But that's a tangent. If election administrators are basing their presidential election planning on off-year or off-off-year  turnout, that's a problem. I realize that mis-allocation of resources, perhaps deliberately, has been a problem in some states. But it's important to remember Hanlon's Razor: "Never attribute to malice that which is adequately explained by stupidity."

One of my pre-election jobs is trying to predict the future and figure out how many voters to expect. It's part science, part art. But the baseline is always, always always elections of a similar type. You don't use presidential numbers (83% turnout this month) to plan for a school board election (more like 5% in a GOOD year) and vice versa.

Estimating turnout requires some sense of history. It was made more complicated this year by reprecincting. I did my best to account for it, and we saw some trends during early voting that helped us adapt, especially in North Liberty.

Turnout estimates require some local knowledge of population shifts and housing construction. It also requires, frankly, some political common sense, more for local contests than for the big ones. Is this year's city election going to be a big deal or a flop? How about a June primary, when you have in effect two elections competing against each other? Republicans have a hot governor race and Democrats have a US Senate primary with a prohibitive favorite? Democrats have a decisive courthouse contest?

Johnson County is looking at the possibility of three elections in the first half of next year: two county-wide and one in the Iowa City school district which covers more that 80% of the county's voters. All three put together are going to be way, way less workload than one presidential election.

Increases in turnout increase the workload not in a linear way, but exponentially. That's because the voters added to the process in a large election are very different than the every election voters. They're less familiar with the process and more likely to need help.

Here's a quiz about Johnson County:

A) Which local city has the biggest zoning fight relative to size?
B) Which neighborhood has a new subdivision every time you drive by?
C) Which precincts got bought out for floodplain since the last presidential election?
D) Which precinct has nothing in it but dorms and frat houses?
E)  Which precinct had a large percentage of married foreign students?

This is a very, very easy quiz for locals. A) University Heights B) North Liberty 6 C) Iowa City 4 and Coralville 1 D) Iowa City 5 E) Iowa City 7.

My point is, that's a very easy quiz for locals. Multiply that quiz by 3000 counties nation-wide.

Now, try to draft legislation.

With shifting local dynamics, and thousands of local jurisdictions, you can't set a one size fits all, same way every year standard based on census population or voter registration stats, to tell you how many ballots to order, booths to set up, or workers to hire. If changes in the law don't take local knowledge into account, you'll be way off, short some places and feather-bedded in others.

Pildes doesn't have the answer yet, but in his critique he's starting to get the question right.

No comments: