Say NO to STV

November 14, 2004

Disadvantages Revealed in Assembly's Recommendation

The Citizens' Assembly named its recommended voting system "BC-STV".

The Citizens' Assembly website has added a column that encourages visitors to check out an animated look at how the single transferable vote (STV) electoral system works, according to the Electoral Office of South Australia. The site, http://www.seo.sa.gov.au/flash.htm, which requires the Macromedia Flash plug-in, is worth viewing. Everyone might not make it through the maze of mouse clicks to see the list of disadvantages of STV, so here are a few of them from proportional systems in general and STV in particular:

Election results take longer to work because they are more difficult to calculate,
Some voters have their preferences counted more times than others, so they appear to be more valuable,
Results may fail to show the support for small parties,
It is still possible for a party with an overall minority vote to be elected to government (because one party may win by a small amount in over 50% of the electorates, but another party may win by a large majority in just under 50% of the electorates),
Two major parties may be compromised by minority parties holding the balance of power in parliament,
Ballot papers are more difficult to mark correctly.

The last disadvantage in the list may surprise some supporters of STV. The Citizens' Assembly's website kicked off its advertisement of the Assembly's recommendation with the claim that "It's as easy as 1, 2, 3" to mark a ballot. In 2001 New Zealand adopted legislation which required health boards to use STV (with Meek's method) for its 2004 health board elections, and it gave municipalities the option of using STV. In the first two health board elections to finish their count, the number of spoiled ballots increased 15 fold and to as high as 13% in one of the districts. It turns out that many voters simply marked an "X" next to their first choice thereby spoiling their ballot. Marking the ballot may be simple in Ireland where they've done it for 82 years, but for many who are used to marking an "X" when they vote, it is difficult to make the change.

Some of the other disadvantages listed on the Australian site are being glossed over by proponents of the system in BC, who claim that STV will give greater representation to small parties. It is highly unlikely that a party with less than 20% popular support could win a seat in an area with 2 or 3 MLAs to be elected; such a party might win a seat in an area represented by 5 or more MLAS. That means BC would give different treatment to different regions. Whether the tradeoff of greater regional representation for less proportionality is fair, or even constitutional, might have to be determined by the courts.

Supporters of STV claim that it eliminates wasted votes. Of course, a "wasted vote" is a strange concept for many who feel that voting for a candidate who doesn't win is not a waste. Imagine how the system would work in a 5 MLA riding in Vancouver, where there would probably be 40 or more candidates running. Those who vote for the leading candidates would probably have their second, and perhaps third, preferences transferred and counted again. Those who vote for the least favoured candidate would also likely have at least their second preference counted, but those who voted for candidates in the middle of the pack would likely have nothing but their first preference counted - just as it is now. That is what the Australian site means when it says "Some voters have their preferences counted more times than others, so they appear to be more valuable." What's fair, or simple, about that?

Whether having a minor party hold the balance of power is a disadvantage or not depends on how one votes. New Democrats have always been proud of what they accomplished during periods of federal minority governments. If the provincial balance of power were held by the Unity Party, some might agree that it is a disadvantage to change to a system that is less likely to give governments a majority.

November 1, 2004

Correction: The example given below fails to properly transfer votes with the diminished "transfer value". Find the error and demonstrate that you fully understand STV.

Say NO to STV

The Citizens' Assembly's website cautions that all of the details have not been finalized for its proposed Single Transferable Vote (STV). Nevertheless, the Assembly has provided enough specifics to make it possible to work out some examples. STV could give rough proportionality between the BC Liberals and the NDP, but contrary to claims about the system, parties with less than 20% support and independents would not necessarily win a seat. With 10% support, the Green Party would be out in the cold, and the way the counting works, those who chose it as a first preference might not have their second preferences counted.

The Assembly provides a link that illustrates how STV worked in the last few Irish elections. It is interesting to note that in those examples the major parties did not run full slates. That may not be the case in British Columbia where the BC Liberals, the NDP, and the Marijuana Party ran full slates in 2001; the Greens almost did. Candidates run in hopeless situations for many reasons: to make a statement, to gain experience and to prepare for a future election. That's a tradition that is likely to continue in BC.

In BC's Election Act the counting rule is primarily for determining what constitutes a valid ballot. The count itself is implied because it is so simple; the candidate with the most ballots wins. Part XIX of Ireland's Electoral Act (1992) provides the rules for counting ballots. This takes six printed pages of mental gymnastics, not on what constitutes valid ballots, but on how to count them. Consider the following portion of those rules:

"(8) The returning officer need not necessarily transfer the surplus of a candidate deemed to be elected whenever that surplus, together with any other surplus not transferred, is less than both the difference between the quota and the number of votes credited to the highest continuing candidate and the difference between the numbers of the votes credited to the two lowest continuing candidates and either-
( a ) the number of votes credited to the lowest candidate is greater than one quarter of the quota, or
( b ) the sum of the number of votes credited to the lowest candidate together with that surplus and any other surplus not transferred is not greater than one quarter of the quota."

I challenge anyone to re-state the above provision in plain language.

The "Droop Quota" is the number of votes it takes to be elected in a constituency; it is equal to one plus the number of valid votes divided by the number of seats plus one. (Quota = 1 +[valid votes/(vacancies+1)]) It is different in each constituency. Votes received above the quota are surplus and are redistributed, but since the second choices of voters may differ, all of the winning candidate's votes are redistributed but reduced in proportion to the surplus divided by the candidate's total vote. The examples shown below are constructed under assumptions that avoid that extra level of complexity. Votes from candidates who finish last are not redistributed until all surpluses have been eliminated through redistribution.

Supporters of STV argue that complexity doesn't matter; "after all", they say, "you use your TV or computer without understanding how it works." Maybe supporters of STV are counting on people not understanding how it works. That allows them to make dubious claims that go unchallenged. One such claim is that STV makes it easier for small parties and independents to get elected. In constituencies that are 2 to 7 times larger than now, it will cost 2 to 7 times as much to provide each household with an election leaflet; that's not easier. It is hard, time consuming work for candidates to knock on doors and personally speak to voters. With 2 to 7 times more voters, candidates will have more difficulty reaching all of them, so most will go without a personal visit. That makes campaigning harder. Most importantly, the counting system does not help small parties in a polarized system where people vote party lines.

Consider two examples, first, a constituency with 5 MLAs and, second, a constituency with 2. Assume that the Irish counting rules apply, that parties run full slates, and that voters plump (meaning that they only vote for their party). Also assume that parties overcome competition between their own candidates by convincing voters to mark 1, 2, 3, 4, 5 next to their candidates in the order they appear on the ballot paper under their party's name. In the absence of that assumption, the fifth ranked candidate for a party could help defeat a preferred candidate to the benefit of another party. In practice, that is probably why Irish parties don't run full slates, but for the purpose of this example, the ranking assumption makes it easier to show how the count proceeds. The ranking assumption also sweeps away the fact that with STV, elections are competitions between all of the candidates, not just between parties. Finally, assume that the votes split 51% for Party A, 39% for Party B and 10% for Party C.

With 10% of BC's 79 seats, under the system used in New Zealand or Germany (Mixed Member Proportional Representation), Party C would get 7 or 8 seats, but in our example of STV, Party C wins nothing!

Call the first example "South Island". It is the amalgamation of the current constituencies of Esquimalt Metchosin, Oak Bay-Gordon Head, Saanich-South, Victoria-Beacon Hill and Victoria-Hillside. In 2001, 116,562 valid votes were cast in those five ridings.

We can call the 15 candidates, five for each party, A1, A2, A3, A4, A5 and so on for B and C. Under our assumptions only candidates A1, B1 and C1 receive number one, first preferences. The counting then proceeds as shown below. In order to understand the examples, it is probably necessary to click on the example of the Irish system and to read the rules they use to count votes. This is not easy!

Valid votes: 116,562
Vacancies: 5
Quota: 19,428

first round

59447

45459

11656

surplus

40019

A1 elected

second round

40019

45459

11656

surplus

26031

B1 elected

third round

40019

26031

11656

surplus

20591

A2 elected

fourth round

20591

26031

11656

surplus

6603

B2 elected

fifth round

20591

6603

11656

A3 elected

Call our second example "Peace River". It is larger than most countries although its population is small. It is made up of the former ridings of Peace River North and Peace River South where a combined total of 19,065 valid ballots were cast in the 2001 election. The counting then proceeds as follows:

Valid votes: 19,065
Vacancies: 2
Quota: 6,356

	A1	A2	B1	B2	C1	C2
first round	9533		7626		1906
surplus	3177
A1 elected

second round		3177	7626		1906
B1 elected

The examples provided here illustrate why Adriane Carr is very upset that the Assembly rejected the system used in New Zealand and Germany and recommended the Irish model. The examples also provide another reason why the Assembly's recommendation should be rejected.

In the absence of the simplifying assumptions used here, the counting becomes even more confusing. It is not good enough to say "the ballot is simple, don't worry about the count". The reason people vote is for the outcome of the count; it's not something that can be dismissed or trusted to others. Third parties may win a few seats with STV, but that is a matter of luck, not something guaranteed as it is with the New Zealand and German systems.

Why would British Columbians vote to throw out our easy to understand method of determining who gets elected and replace it with a very confusing system that cannot guarantee any seats for a party that wins 10% of the vote?

Click here for two more articles on why you should vote NO to BC-STV.