Politics FB

XTREME Civics & Comparative Gov’t: Voting Systems Pt. 2: STV & US Primaries

Left out of the voting systems conversation last week were two rather unique ways of voting—one that affects only a handful of people around the world, but is still quite important to talk about, and one that, directly or indirectly, probably affects every person in the world, but is a mess to talk about. Let’s start with the former.

One of the central tensions voting systems have is how to balance the need for proportional representation with the need for local representation. A proportional system tends to favor party-selected politicians, while systems that allow voters to select their political representatives tend to become disproportionate. 

Single Transferable Voting is a way to achieve roughly proportional results using multi-member districts but without letting parties choose those candidates from lists. Voters always decide which candidate will take the seat. It also allows voters to not waste a single vote. That is to say, no voter will ever throw away their vote on a lost cause (unless they choose to) or overvote for a candidate who has won a seat. As roughly as possible, every vote in an STV election matters and has the same power as a vote in a different district.

Because of this, it is a favorite voting system of political scientists everywhere, which is why we’re talking about. It is also not a favorite of governments everywhere: it has been adopted by just two countries, both nearly 100 years ago: Ireland in 1921 and Malta in 1922. That’s it. That’s all the countries that use it. Two European islands that combine for just over 5 million people.

In its simplest form, STV is a ranked-choice voting system with multiple members, and the lowest performer is eliminated each count until you end up with all the seats in a constituency filled. But where STV really shines is with the implementation of something called the “Droop quota.” 

The Droop quota is the number of minimum votes required to fill every seat. So, looking at our states in last week’s example, we have 20 seats per state and 1,000 voters in each state. It would seem like you’d need 50 votes per seat to take a seat. But that would be wrong. You actually need 48 votes to fill 20 seats. 

Why? Because the minimum number of votes required to win a seat is actually the number of votes divided by one more seat than exists, plus one vote. It’s both legally and mathematically impossible to seat 21 people in the state using this quota: you’ll always get 20 seats (those extra +1 votes would require that there be an additional 21 voters in the state that don’t exist). 

Another way to think about it is this: If 20 people got 48 votes each, we’d have 960 total votes. If a 21st candidate got all the remaining votes, they wouldn’t be able to get into parliament, because they didn’t meet the threshold (they got 40, not 48, votes). Conversely, if 20 people got 47 votes, then we’d have 940 votes total, meaning there’d be 60 votes left, which would fill a seat if a 21st candidate were awarded it. So 48 is the lowest possible number you can get and win a seat.

Why is this important? Because if a candidate receives more than 48 votes, every vote past 48 is wasted. Voters who voted for the candidate beyond the Droop quota threw their vote away, not on a lost cause, but rather on a sure thing.  

Single Transferable Voting systems with a Droop quota seek to undo this. Once a candidate has received more than the minimum required 1st preference votes to take a seat, on the second count, a formula is used to randomly determine who the “extra” voters who voted for that candidate were and redistribute their votes to other candidates in the race using those voters preference rankings. 

There’s no point in showing you a made-up example, it’d be too complicated to come up with imaginary preference rankings. Instead, I’m going to show you a real-life example from earlier this year that comes from one of the two countries that uses STV: Ireland. We’re going to look at the Dublin West constituency, for two reasons: first, it only took six counts to fill the four seats, which is a minor miracle, given that these can run into the double digits, and second because it’s the constituency that the then-current Taoiseach Leo Varadkar hails from and it shows that STV doesn’t necessarily guarantee notoriety a seat. Here’s where we’re pulling this data from


At the top of the page, you’ll see that the number of valid votes in the constituency was 43,628. The constituency seats four. So if we divide that by 5 and then add 1, we arrive at the Droop quota: 8,726.

In the first count, only a single candidate has achieved this number: Sinn Fein’s Paul Donnelly, with 12,456 votes. Donnelly secures the first of the four seats.

For the second counting, no candidate is eliminated, not even Independent candidate Sean O’Leary with his 24 first preference votes. That’s because there are 3,730 extra votes for Donnelly which will be redistributed. In theory, any candidate, even O’Leary, could receive enough of those votes to leapfrog them up the ranking.

At the end of the second count, however, all three Independents are excluded, despite redistributing Donnelly’s excess votes. That’s because none of them can conceivably reach Aengus Ó MaolÁin’s 1,114 votes, even if you sum them all together. No other candidate has crossed the quota threshold either, though Ruth Coppinger, the sitting TD for Solidarity-People Before Profit has gained the lion’s share of the extra SF vote.

On the third count, the Independents’ votes are redistributed, and now Aengus O Maolain is eliminated, narrowly unable to rise above far right Aontu’s Edward Macmanus.

On the fourth count, Macmanus is eliminated, and Taoiseach Leo Varadkar needs precisely 20 votes to retain his seat. 

On the fifth count, he gets 37 more votes than he needs from the Aontu voters, pushing him over the quota. Now both the other Fine Gael candidate, Emer Currie, and Joan Burton of Labour are eliminated. Again, that’s because even if in the unlikely event all of Currie’s voters went for Burton next, it would still not be enough votes to top former Green Party chair Roderic O’Gorman’s 6,270 votes.

On the sixth and final count, Jack Chambers of Fianna Fail secures his seat, and we stop counting, even though neither Roderic O’Gorman or Ruth Coppinger has achieved the quota. That’s because if we were to count for a seventh time, it would be to redistribute Coppinger’s voters. But merely by eliminating her, we’re left with four candidates for four seats. And so O’Gorman wins the final seat.

Run over an entire election, and you get roughly a proportional number of seats. You can do the math—most misses in party representation compared to first preference votes are off by <1%. The largest miss is by Sinn Fein, who failed not because STV failed them, but rather because they simply didn’t run enough candidates to maximize their gains (their polling bump came as a fair surprise).

And now for something completely different: The US Presidential Preference Primary System. Unlike most countries, the United States never effectively legislated or create a constitutional status for political parties, which, over time, has led to two large big tent or broad church parties. This often occurs in many other countries, whether by natural attrition in plurality systems or through coalitions or simply the size of the middle of the political spectrum’s vote in the electorate to support large, broad, parties. Where it’s relatively unique in the US, however, is how voters are allowed to participate in what is normally an internal party decision.

As we’ll cover (no doubt) in a future installment, many parties in countries across the world allow rank-and-file party adherents to participate in internal nominating processes. In the US, however, many parties (or are required to by law) to allow any voter, regardless of political affiliation, to participate in candidate selection. Sometimes they can restrict that, but often enough not. Many of those rules are further loosened for the four-year process of choosing a party’s nominee to become President of the United States. This system of public participation is the widespread adoption of what was once an outlandish reform by groups in the early 20th century to break the traditional power structures of party machines–by allowing any voter to participate in nominating contests, early Progressives hoped to dilute the strength of influential ward bosses, mayors, and other politicians who bent the process to their will and favored candidates.

As with most things related to electing the American President, voters do not, in truth, actually vote for the person they wish to win the election. While they will certainly mark their ballot as such, what they are doing instead is electing delegates who will support that candidate for the nomination at the respective party’s national convention in the summer. Much like the general election for President is actually a choice of awarding a state’s electoral college votes to a candidate, the primaries serve to choose delegates and award those to a candidate.

There are two ways states handle this. One way is called a “caucus.” This system is run entirely by the party without government oversight. It involves voters going to a designated place and then standing in a designated area in that place to show their support for the candidate. Any supporters of candidates who do not cross a “viability threshold” (we’ll talk about this in a moment) may change their support to another candidate, potentially lifting that candidate above the threshold, much like STV’s quota mechanism.

The other way states handle it is to run a primary election. In this case, it becomes a formal election managed by the Secretary of State for that state, with voters casting votes via secret ballot for a particular candidate (in some states, they also elect delegates for that candidates directly as in an open-list proportional system; in others, the candidates choose their delegates, as in a closed-list proportional system).

In both cases, candidates with votes above the threshold earn delegates for the national convention. Any candidate who secures an absolute majority of “pledged” delegates becomes the nominee.

Where the system gets rather bizarre is what it’s attempting to do. In theory, it is attempting to use a proportional system to elect a single winner. Proportional systems cannot achieve this function; by design, they’re intended to elect multiple winners because you cannot have, for instance, 40% of one person (at least, not without making a mess).  

Where it’s also bizarre is how delegates are awarded. Candidates can earn delegates both at the statewide level and at a lower district level (usually congressional districts). Some Republican primaries don’t allow lower district level awards, but we’re going to ignore that. In that way it’s similar to a parallel voting system, only if the local elections were also proportional. So you have two different proportional elections going on at the same time in each state on primary day. And because US congressional districts are notoriously drawn in such a way that they do not guarantee fair representation (and even if they were, they would not be drawn that way for entirely partisan contests such as presidential primaries), that means there are often unproportional results.

Both parties have ways of attempting to solve the problems caused by using this parallel proportional system to do something it was never intended to do. Each party uses two strategies to correct for this. One they share in common, the other is unique to each party.

The shared strategy is a viability threshold. Thresholds are a common feature found in many proportional voting systems. Thresholds establish that only parties who gather a certain percentage of the national (or regional) vote can earn seats in the legislature. This is to exclude a vast number of small parties with strong bases of support from driving down the seat totals, leading to coalitions where a single person could bring down the government with minimal consequences for themselves. In a proportional system without thresholds, you could conceivably have a 100 seat legislature with a governing majority made up of a party with 40 seats, a party with 8 seats, and 3 independents. Each of the three independents could collapse the government at any time, over any reason, giving them undue influence over policy compared to the larger coalition partners. In proportional systems, a 5% threshold (such as in Germany) is considered rather high, while a 2% threshold is considered a relatively normal barrier if you must construct such a barrier.

Because the US Presidential Preference Primaries are attempting to elect a single candidate, their thresholds are set significantly higher. For the Republican primary, the thresholds can be anywhere from 0 to an astronomically high 20%. For the Democratic primary, the threshold is set across the board at 15%. If a candidate fails to cross the threshold, their voters are excluded from the total for calculating proportionality. For instance, if an election broke down with one candidate getting 50% of the vote, another candidate getting 30% of the vote, and two candidates each getting 10% of the vote, then 20% of the vote that went for the last two candidates would be dropped, and the candidate getting 50% of the vote would earn 62.5% of the delegates (50 divided by 80), while the candidate who gets 30% of the vote would earn 37.5% of the delegates. In that way, voters who vote for a failed candidate can spoil the candidate they might have chosen second, and it’s tremendously easy for voters to play spoiler and waste their votes because the threshold is so high.

The other strategies are different, and in both cases, they serve to correct for proportionality. The easiest to explain is the Republican process. In the Republican primary, many states (though not all) are “winner-take-all” states. In effect, those states award delegates to the person with the most votes (although, again, in many of those states it’s possible for other candidates to win some delegates at the congressional district level). Other states are “winner-take-most” states, which award the majority of delegates to the person with the most votes. This usually means that the candidate who wins the most states achieves an absolute majority of the delegates, even if they do not win a majority of the vote (for instance, Donald Trump won 58% of the delegates with just 44% of the primary vote in 2016). And then the remaining states are proportional. And some states directly elect delegates. Some states mix and match these systems, where a statewide primary and a congressional district primary are can award delegates through different means (i.e. one might be winner-take-all statewide and proportional in congressional districts).

The Democratic primary has an even less democratic system that the Republican primary’s mix-and-match awards, but it is at least consistent across all states, and in keeping with how many other nations’ parties choose party officials. Since 1984, the Democratic National Convention has given luminaries within the party “superdelegate” status. These delegates are not bound by the results of primaries or caucuses or state conventions, though they’re assigned to states and territories as though they were. Superdelegates were originally set at 14% of the convention delegates, eventually scaling up to as much as 20% in 2008, and included ex-presidents and vice presidents, elected officials across the nation, state party chairs, activists, labor officials, nonprofit leaders, and various other high-level members of the Democratic Party’s wider coalition. After 2016, the party reformed the role of superdelegates, reducing them back down to 16% and limiting them to elected officials and party leaders. In theory, superdelegates are a sea anchor on the popular primaries, allowing the party to choose a winner over the will of the voters (this is, after all, a partisan decision, not necessarily a democratic one). As of the 2016 reforms, they may only do this if voters in the election fail to give any candidate an absolute majority of pledged delegates, since they are prohibited from voting on the first convention ballot. In practice, the only time the superdelegates have had a significant influence on the outcome of the primary was 2008, when they chose Barack Obama over Hillary Clinton, clinching his nomination.

The final bizarre feature of the US Presidential Preference Primary system is that, while it is a national election covering all 50 states, most territories and citizens living abroad, not all those constituent parts of the nation vote at the same time. Rather, as a result of the haphazard way primaries and caucuses were first introduced into the parties nominating structure prior to 1968 and then the process was reformed over a century, a handful of small states go first, followed by various waves of subsequent states. There is usually a fairly early election day that has the largest number of states vote on the same day, traditionally termed “Super Tuesday” although this doesn’t always happen and there may be two Super Tuesdays in a single primary season.

All in all, the United States has one of the most convoluted and above all longest campaign seasons in the world to elect its head of government.

That’s it for this week’s installment of XTREME Civics & Comparative Government. We’re off next week (this is going to be once every couple of weeks or so anyhow). Give your thoughts on STV or the US Primary System (as far as structure goes–this is a political thread, not a Politics Thread) in the comments!