Should Ballot Order be Randomized in Vancouver?

Written by Jack Lucas and Charles Tessier

Earlier this year, Vancouver city councillor Andrea Reimer introduced a motion to randomize the order of candidate names on the city’s election ballot. The purpose of the proposal, Reimer explained, was to eliminate the unfair advantage that some candidates get by virtue of their position at the top of the ballot. When the ballot is organized alphabetically, candidates whose last names come early in the alphabet enjoy prime real estate at the top of Vancouver’s very long council ballot – a ballot that often includes more than forty candidate names. Vancouver voters, faced with a complicated ballot and ten city council votes to cast, might simply scan the list of names from top to bottom, choosing the first names they recognize as they move down the list. If even a small number of voters actually do this, the result could be a real advantage for the candidates who are blessed with early-in-the-alphabet names. Call it the Aardvark advantage.

Reimer’s proposal, which passed in city council earlier this month, briefly reignited a longstanding debate about the very existence of the Aardvark advantage in Vancouver. Some claimed that the research showed clear evidence for name-order advantages; others argued that name-order advantages are a myth. Reimer, for her part, argued that there are “academic studies which feel conclusive and others that don’t,” a conclusion that her fellow councillor, George Affleck, shared. “For every study that says there’s bias,” Affleck said, “there is another that says there isn’t.”

So is the Aardvark advantage an “urban myth that refuses to die,” as one reporter dubbed it a few years ago, or is it a source of genuine disadvantage for unlucky candidates with names at the end of the alphabet? We’re currently working on a full analysis of this question, using a new database of election results from cities across western Canada, but in the meantime we thought we’d share some information about Vancouver’s part of the story. As we’ll see, the answer is…well, it’s a little complicated. So let’s get into it.

***

Western Canadians have been worried about the Aardvark advantage for a long time. Nearly a century ago, politicians and administrators in the city of Calgary became aware of the idea that a candidate’s position on the ballot might influence the number of votes that he or she received. After getting approval from the provincial government, the city switched in 1919 to a rotating ballot order. In this system, Mr. Aardvark would be in first position on one ballot, second position on the next ballot, and so on down through the stack of ballots until he returned to the top of the list. Other cities, such as Winnipeg, also adopted the rotating ballot in this early period.

These days, Canadian cities use a wide variety of ballot order rules for their elections. Most cities, including every municipality in Ontario and Quebec, use the alphabetical ballot. Calgary continues to use the rotating ballot that it introduced to Canadian elections nearly a century ago. Winnipeg recently switched to a randomized ballot, in which the order of the candidate names is determined by a random draw prior to each election.

Do these various ballot order rules actually matter? In 2018, political scientists Charles Tessier and Alexandre Blanchet used data from provincial and municipal elections in Quebec to provide Canada’s first systematic analysis of this question. They found that top-of-the-ballot candidates in Quebec municipal elections do have an advantage, receiving a boost of about four or five percentage points over their counterparts lower on the ballot. In provincial elections, however, the ballot order effect vanished. Tessier and Blanchet concluded that the difference had to do with political parties. “The fact that ballot order seems to be present only in elections without party cues,” they concluded, “indicates that they are indeed a crucial piece of information to voters.”

Tessier and Blanchet’s findings mirror those of researchers who have analyzed ballot order advantages in the United States and other countries. These researchers have consistently discovered that the ballot order effect is real – but it’s much more pronounced in elections that involve non-partisanship, low-profile races, or low amounts of information for voters. While the exact estimates of the advantage vary, many researchers have found that candidates at the top of the ballot in non-partisan elections receive a vote share boost of 2-4 percentage points over their competitors.

***

This brings us to Vancouver. Here’s a simple way to test for the Aardvark advantage in Vancouver’s council elections, where each voter can cast up to ten votes for council candidates. First, we divide the ballot into four equal groups, called quartiles, based on the order of the names on the ballot. For instance, if the ballot has forty names, the first ten names will be in the first quartile, the second ten names in the second quartile, and so on through to the fourth quartile. If there’s no name order advantage in Vancouver, candidates should be just as likely to be elected from each of the four quartiles: we should see that roughly twenty-five percent of elected councillors come from each quartile. If there is a name-order advantage, on the other hand, we’ll see a disproportionate number of elected candidates from the first quartile.

When we run this quartile test, we discover that the Aardvark advantage is real. You can see the numbers for yourself in the graph below, which summarizes the proportion of elected councillors from each quartile in Vancouver from 1936 to 2014. Nearly thirty-six percent of elected councillors have come from the first quartile on the ballot, compared with less than twenty-five percent in each of the three other quartiles. The chances that we’d see a result like this if there is actually no name order advantage in Vancouver are vanishingly small.quartile1But before you stop reading, there’s an important catch. The numbers in the graph above come from every city council election in Vancouver from 1936 to 2014. But a lot has changed in Vancouver elections since 1936. The most important change is probably the re-emergence of a competitive party system in Vancouver in the late 1960s and early 1970s, as well as a related policy change in 1974: adding party labels to candidate names right on the ballot itself.

Given what we’ve learned from Tessier and Blanchet, we’d expect that this partisan context might substantially reduce the name order advantage in Calgary. And that’s exactly what we find. In the graph below, we’ve split city council elections in Vancouver into two groups: elections before the mid-1970s without party labels on the ballot, and elections after the mid-1970s with party labels on the ballot. In the graph on the left, the name order advantage is huge. After the 1970s, it vanishes. Give voters clear party labels and robust party competition, and it seems that the Aardvark advantage is a much less serious concern.quartile2

***

So far, we’ve suggested that the name order advantage isn’t as serious today as it once was in Vancouver. But for critics of Vancouver’s alphabetical ballot, the effects of the system on actual election outcomes are only one part of the problem. There’s also the concern that the alphabetical ballot will discourage some candidates from even running for office because of a belief that ballot order advantages are real. This was a point that Andrea Reimer emphasized in her argument for the randomized ballot in Vancouver. Reimer recounted “dozens and dozens” of occasions in which, when trying to recruit possible candidates for office, people responded to her by asking, “with the ABCD (alphabetical listing of names), do I really have a shot?”

Reimer’s remarks suggest a concern with what we might call the Aardvark arms race: if people think there’s a name order advantage, they’ll be less inclined to run for office if their last name happens to be Wilson or Zorn. Over time, the proportion of candidates whose last names start with the early letters of the alphabet will increase.

We can use our Vancouver election results to investigate this possibility, and our results suggest that the Aardvark arms race is actually a real phenomenon in Vancouver. In the graph below, each circle plots the proportion of candidates in an election whose last names begin with the first four letters of the alphabet (ABCD). The dashed line captures the basic trend through time. As the line suggests, the proportion of ABCD names in Vancouver elections has been increasing through time. But it’s important to notice that the overall increase is quite modest: an increase of about 1.5 percent in the proportion of candidates with ABCD names each decade. You might think that this increase has to do with partisan strategy, as parties nominate more and more ABCD candidates for election, but it’s actually driven almost entirely by non-partisan candidates, who are much more likely to have ABCD last names today than they were decades ago.top4

***

Should Vancouver randomize its municipal ballot? Was council’s decision earlier this month a mistake? Many have argued that a randomized ballot is a harmless reform, so it should be adopted even if there’s little evidence for the Aardvark advantage. But it’s worth remembering that the randomized ballot does come with some costs. The most important is voter confusion: on a list of thirty or forty candidates, it’s pretty easy to find the people you’d like to vote for when the names are organized alphabetically. When the names are randomized, on the other hand, locating your ten preferred candidates on the ballot can be a pain. In fact, when Vancouver experimented with a randomized ballot in 1993, this was one of the reasons that the city switched back to the alphabetical ballot.

If we’re worried about the basic unfairness of the ballot order advantage, we should also remember that a randomized ballot is just as unfair as an alphabetical ballot: rather than winning the ballot order lottery because your last name happens to be Aardvark, you win the lottery because your name was the first to be pulled from the hat. If ballot order advantage is a problem, a randomized ballot will produce a different set of advantaged names, rather than addressing the underlying unfairness. And given the challenges involved in casting ten votes on a randomly ordered list of thirty or forty or even fifty names, there’s good reason to believe that ballot order advantages would probably increase under a randomized ballot. To avoid this problem, it’s vital that Vancouver adopt a rotating system alongside its recent decision to randomize the ballot – this practice, used in some American states, rotates the order of the randomized names in different polling stations across the city. Unfortunately, however, Vancouver’s charter does not permit this measure – so unless there’s a quick change to the legislation, the city is stuck with either an alphabetical ballot or a randomized ballot with a standardized order across the entire city.

In our view, what’s most important to remember are the basic findings that we described above: since Vancouver’s competitive party system emerged, the ballot order advantage has not been a serious problem. Given what Tessier and Blanchet found in Quebec — ballot order advantages only in non-partisan races — this is just as we’d expect. When voters have the information they need to make their decision, they’re much less inclined to skim through the ballot and arbitrarily pick a name or two at the top.

The real solution to the Aardvark advantage, in other words, is information: and when it comes to elections, there’s no information that’s quite as useful as a candidate’s party affiliation. Vancouver can use an alphabetical ballot, a randomized ballot, a rotating ballot, or any other ballot it likes. What really matters is that Vancouver voters have a chance to learn about the competing parties, to understand what each party stands for, and to easily determine which candidates belong to each party. To do this will require hard work from the political parties, from election administrators, from Vancouver media, and from voters themselves. With clear political parties and high-quality information, we think the Aardvark advantage will remain a lingering memory from Vancouver’s distant past.

Jack Lucas is an Assistant Professor in the Department of Political Science and Director of the Urban Policy Program at the School of Public Policy at the University of Calgary. 

Charles Tessier is a Postdoctoral Fellow in the Department of Political Science at McGill University. 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s