Read an Excerpt

Timing and Turnout

How Off-Cycle Elections Favor Organized Groups

THE UNIVERSITY OF CHICAGO PRESS

CHAPTER 1

Timing and Turnout

The Basics

When it comes to voter turnout rates in industrialized democracies, the United States is at the bottom of the list. About 55 to 60% of eligible American voters participate in US presidential elections, compared to turnout rates of 91% in Belgium, 90% in Italy, 86% in Sweden, and 81% in Australia. Some of the gap between the United States and other democracies can be explained by unusual voter registration requirements in the United States and compulsory voting laws in countries like Belgium and Australia, but nonetheless, the differences are striking. In fact, the sole industrialized democracy whose turnout rates in national elections regularly fall below that of the United States is Switzerland, and with its average turnout rate of 54%, it is not far behind.

Moreover, turnout rates of 55 to 60% are by no means typical of elections in the United States; those figures describe American voter participation at its highest. Turnout in midterm congressional elections runs closer to 35%, and participation rates in national primary elections are lower still. Of course, American government is much more than just its national government. According to the preliminary counts of the 2012 Census of Governments, there are 89,055 individual governments in the United States, most of which are local governments like counties, cities, townships, school districts, and the like. And in local government elections, voter turnout rates of 20%, 10%, or even lower are common. Thus in comparison to the turnout levels of most elections conducted in the United States, the 55 to 60% turnout rates of presidential elections are actually quite high.

However, citing average turnout statistics by the type of government in the United States masks an enormous amount of variation across governments. Consider, for example, two cities in the San Francisco Bay Area: Palo Alto and Berkeley. Palo Alto, the municipality adjacent to Stanford University, has a considerably smaller, older, and wealthier population than Berkeley, and so based on what scholars have established about the correlates of political participation, many would guess that Palo Alto municipal elections would have higher turnout rates than Berkeley municipal elections. But in fact the opposite is true. In 2008, a full 65% of Berkeley registered voters cast a vote in their city election, whereas the following year, only 38% of Palo Alto registered voters participated in their city races. Those elections were not anomalous. In 2002 and 2006, Berkeley city races had participation rates of 57% and 59%, respectively, while Palo Alto turnout was a mere 38% in 2007.

Why the large turnout gap between the two cities—and in the unexpected direction? The explanation is actually quite simple: Berkeley's city elections are held in November of even-numbered years, concurrently with presidential elections, gubernatorial elections, and congressional elections, whereas Palo Alto's city elections have historically been held in November of the odd-numbered years, concurrent only with local school board elections. Anyone skeptical of this explanation for the turnout gap need only look at turnout rates in some of the ballot measure elections that have been held in Palo Alto in November of even-numbered years: In November 2008, a full 82% of Palo Alto registered voters cast a vote for or against local Measure N—a measure dealing with library improvements. In 2002, 56% voted in Palo Alto ballot measure races. Thus within Palo Alto, turnout was 18 percentage points higher in city elections concurrent with gubernatorial and congressional elections than in the odd-numbered years, and it was a whopping 44 percentage points higher when city elections were concurrent with a presidential election.

The case of Palo Alto versus Berkeley illustrates a pattern that is already familiar to some political scientists: that is, the timing of elections affects voter turnout. In a study of California cities, for example, Zoltan Hajnal, Paul Lewis, and Hugh Louch found that turnout in off-cycle city elections averaged 36 percentage points lower than in city elections held during presidential elections. Similarly, using a sample of fifty-seven cities across the country, Curtis Wood estimated that off-cycle election timing dampened city election turnout by an average of 29 percentage points. Analyzing a sample of elections in thirty-eight large cities over twenty-five years, Neal Caren also found that turnout was significantly lower in cities that did not hold elections at the same time as presidential elections, by about 27 percentage points. Not only are these effects large, but they also dwarf the effects of other institutional variables these scholars examine. As Hajnal concluded, "election timing is the most important factor in explaining local voter turnout." Wood summarized his findings similarly, saying, "the single largest predictor of voter turnout is holding city elections concurrently with national and state elections." Clearly, turnout in municipal elections hinges greatly on whether they are held on the same day as national elections.

This is not purely a municipal election phenomenon. Using a national sample of school districts, Frederick Hess estimated an 18-percentage-point turnout gap between school board elections held during state and national elections and school board elections held on other days. Nor is it limited to local government. Gubernatorial elections attract the most voters when they are held concurrently with presidential elections and the fewest voters when they are held in the odd-numbered years. Midterm congressional election turnout averages 13 percentage points lower than turnout in presidential-year congressional elections. In general, voter turnout in the United States varies predictably with whether elections of one type are bundled with—or separated from—elections that attract greater participation.

Given that election timing is so consequential for voter turnout, one would think the electoral calendar would be a major focus of research among political scientists. After all, it is a political institution, and the study of political institutions is at the very heart of the discipline. And yet the topic of election timing in the United States barely shows up in the literature on institutions. The most basic questions remain pretty much unexplored: Who decides when elections will be held, and how are those decisions made? Does the timing of elections matter for more than just voter turnout? Does election timing affect election outcomes, or public policy? These are fundamental questions about the operation and substance of American democracy, and yet the extant literature offers little in the way of answers.

This book is an attempt to make some headway. It focuses on the causes and consequences of election timing, by which I mean whether or not elections of one type (for example, city elections) are concurrent with elections of another type (for example, presidential elections). My main argument is that the timing of elections has great potential to affect not only how many people vote but also who votes, which candidates win, and to whom elected officials respond in designing public policy. Specifically, I argue that shifting from on-cycle to off-cycle election timing has the effect of increasing the electoral presence of the organized, for two main reasons: First, those who have a large stake in an election outcome turn out to vote at high rates regardless of when that election is held, and many of the individuals with the greatest stake in election outcomes are members of organized groups. Second, off-cycle election timing enhances the effectiveness of organized groups' mobilization efforts, since each additional mobilized supporter goes further toward tipping the election outcome when turnout is low. In general, the members and mobilized supporters of organized groups should make up a greater proportion of the active electorate in off-cycle elections than in on-cycle elections.

This change in the composition of the electorate that results from a shift in election timing has great potential to shape election outcomes and policy making, and in some political contexts, it's relatively easy to see how. In particular, when organized interest groups seek policies with concentrated benefits and distributed costs, and when they face relatively little organized competition, they should be more successful in electing their preferred candidates and securing favorable policies when elections are held off-cycle rather than on-cycle. However, even when organized groups compete over policy, and even when voters on both sides of an issue are equally motivated to turn out, election timing can still tip the balance of power in favor of one group or its rival, with potential to change the outcome. The choice of election timing is therefore a consequential one, and not just for voter turnout but also for election results and the substance of public policy.

A Snapshot of Election Timing in the United States

Before I can move forward with an evaluation of the causes and consequences of election timing, a very basic problem must be resolved. That is, aside from national elections and a few studies that have examined samples of local governments, we do not even have basic information on when elections are held in most American governments. There is no single source that provides such information, and the laws governing election timing vary tremendously by state. Thus creating a sketch of when different types of elections are held across the United States requires researching each state one by one. Considering that the rules governing election timing in many states are quite complex, collecting that kind of information is a nontrivial task.

Take the state of New Jersey as an example. The schedule of elections in New Jersey is so complex that it makes one wonder whether it should be called a "schedule" at all. Its state-level elections are held in November of odd-numbered years, separate from US presidential and congressional elections. Elections for the state's twenty-one counties and any of its 566 municipal governments that hold partisan elections are held annually in November, but most of its municipalities that have nonpartisan elections hold their regular elections on the second Tuesday in May. School district elections, on the other hand, have historically been held on the third Tuesday in April. Fire district commissioners are elected on an entirely separate day: the third Saturday in February. The dates of most other special district elections are so elusive that groups of citizens have actually banded together to try to figure out when governments like sanitation districts elect their representatives. So, in addition to national Election Day, which refers to the Tuesday after the first Monday of November of even-numbered years, New Jersey voters are asked to go to the polls in the odd-numbered years, in the winter, on a couple of dates in the spring, and in many cases during the summer too. This does not even account for primary elections or special elections, which add to the list of voters' responsibilities.

New Jersey therefore has something that looks more like a steady stream of elections every year than a single Election Day once every two years. And in some other states, the rules governing election timing are even harder to follow. For example, some states give local governments discretion to hold elections whenever they want. Other states give local governments a menu of election dates from which they can choose. There are even some states that establish a uniform election day for a particular type of government but then carve out exceptions for certain places. The thought of trying to keep track and make sense of all these rules is quite daunting. And yet at the same time, it is difficult to ask questions about how election timing matters, or why elections are held when they are, if we don't actually know when elections are held. So as a necessary preliminary step, I set out to collect some basic information on the rules governing election timing in each of the fifty states.

First off, it is important to note that there are many different types of elections in the United States: regular elections, primary elections, runoff elections, special elections to fill vacancies, ballot measure elections, tax and bond elections, charter or constitutional elections, recall elections, and so on. To make the task at hand manageable, I focus on the timing of regular elections for government officials. I also exclude nonschool special districts from my search, since determining when special districts hold elections is prohibitively difficult for many states. I used a variety of resources to assemble the information, including state statutes and constitutions, information provided by the offices of secretaries of state, reports compiled by state municipal leagues and state school board associations, and many phone and e-mail communications with government officials and associations of local governments.

The final product of the data collection effort is a snapshot of state and local election timing in the United States as of January 2012, which I summarize in table 1.1. Since government is structured differently in each state, I group the information in the table into four categories: the rules governing state elections, county elections, municipal elections, and school district elections. For each category in each state, I document the month of the regular elections as well as whether those elections are held in even-numbered years, odd-numbered years, or annually. In many states, this is straightforward, since all of the elections within each of the four categories are held on the same day. As I alluded to above, however, oftentimes the election timing rules for a certain category of government cannot be summarized so simply. There might be subtypes of governments within a particular category that hold elections at different times. For example, within the municipal government category, villages might be required to hold elections on one day and cities on another. Alternatively, individual governments within a category might have discretion to choose from among a menu of established election dates. In table 1.1, I label all such cases as having "Multiple schedules." I also distinguish between cases with "Multiple schedules" and cases in which governments in a given category have broad latitude to choose whatever election date they please with very few restrictions. For the latter cases, I indicate that election timing for the category "Varies." Where necessary, the table notes document detail on individual cells of the table.

Even with all of these simplifications, the information in table 1.1 is a lot to process. In a way, this is testament to just how complex the American electoral calendar really is. But ultimately, my interest in this book is in whether elections for governments like states, cities, and school districts are held on the same day as other elections or not. Therefore, to summarize the information about election concurrence in each state, I have used a bold font if at least some elections of that type of government in that state are held at times other than November of even-numbered years. The cells left in a normal font are those for which all elections in that category and state are held concurrently with congressional and presidential elections.

It is the bold font that really highlights the patterns of election scheduling in the United States. At the level of state government, there are only five states that hold some or all of their regular elections in the odd-numbered years: Kentucky, Louisiana, Mississippi, New Jersey, and Virginia. The rest of the states hold elections concurrently with national elections. While November of even-numbered years is also the modal choice for county elections, there are several states that hold county elections at other times. In California and Tennessee, regular county elections are concurrent with statewide primaries. In Louisiana, Mississippi, Pennsylvania, and Virginia, county elections are in November of the odd-numbered years. Two states—New Jersey and New York—hold county elections annually in November (with some exceptions), and in Wisconsin, there are both nonpartisan county elections in April and partisan county elections in November. Alaska's boroughs, which are similar to counties, hold elections annually in October. Thus even at the county level, there is quite a bit of variation in when regular elections are held.

But the variation really takes off when we move to municipal elections. For municipal government, elections in November of even-numbered years are the exception rather than the norm, a schedule used by only five of the fifty states: Arkansas, Kentucky, Nebraska, Oregon, and Rhode Island. In several states, some municipal elections are held concurrently with national general elections while others are held off- cycle, either because individual governments have a choice over their election scheduling or because certain types of municipalities have on-cycle elections and others have off-cycle elections. In North Dakota, municipal elections are concurrent with the state's primary elections, and in South Dakota, it is permitted but not required for municipalities to hold elections during the statewide primary. In many states, however, all municipal elections are held on days other than national general or primary elections. Thus the rules for municipal elections in most states are quite different from the rules for state and county elections. Across the United States, municipal elections tend to be held on days other than national elections.
(Continues...)

---

Excerpted from Timing and Turnout by SARAH F. ANZIA. Copyright © 2014 The University of Chicago. Excerpted by permission of THE UNIVERSITY OF CHICAGO PRESS.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.

---