Combined Multi-Member & Top-Up

The federal Electoral Reform Committee recommended exploring a combined approach of multi-member ridings in more urban areas and single-member ridings in more rural areas to better address Canada’s unique geography.

Why We Might Want to Combine Approaches

While both Multi-Member and Top-Up approaches offer significant improvements over our current voting system (check out our Scorecard), there are always things that could be improved.

For example, one concern raised about the Single Transferable Voting system recommended by BC’s Citizens’ Assembly on Electoral Reform (a Multi-Member model) was that there was a possibility that all MLAs elected in a multi-member district in a more rural part of the province would come from the population centre in that combined riding and not from the less densely populated areas (note that the Local PR Multi-Member model can ‘protect’ existing ridings so that each existing riding could be guaranteed to continue electing an MLA).

A second concern about STV was that if we try to keep the number of seats low (e.g., 2 or 3 or 4), then there can be larger mismatches between the vote share attracted by candidates from a particular party and the number of seats won by candidates from that party, and it can be more difficult for candidates from smaller parties (e.g., the Greens) to get elected.

A concern raised about the Mixed Member Proportional voting system is that the single member ridings need to be enlarged by two-thirds in order to free up an appropriate number of top-up seats (assuming a 60-40 constituency/regional seat balance), and that this likewise might be a concern in the more rural parts of the province.

For reasons such as these, former Chief Electoral Officer Jean-Pierre Kingsley suggested that we might consider using multi-member ridings in the more urban areas and keeping some single member ridings in the more rural areas.

The problem with this idea is that voters in the more rural areas who don’t support the winning candidate wouldn’t have their votes count, so the overall result would be less representative than it could be.

How a Combined Approach Might Work

During the federal Electoral Reform process, Fair Voting BC and Fair Vote Canada proposed a combined system to address this concern with Kingsley’s suggestion, which we called Rural Urban Proportional Representation (RU-PR).

As with Kingsley’s proposal, we suggested using moderate-sized multi-member ridings (in the range of 2-5 seats) wherever possible, and single-member ridings in the most rural locales. These would be supplemented by a few top-up seats, but many fewer than would be necessary with MMP, because the multi-member ridings would already produce a considerable degree of proportionality. We estimate that only about 10-15% of the seats would need to be set aside under the RU-PR approach (approximately one in every 7 or 8 seats), as compared with 40% using MMP. This would mean that we could have a few single member seats in the most rural parts of the country that would be only modestly larger than the current single-member ridings (and, if we integrated the top-up seats into the largest multi-member ridings, it would even be possible to avoid changing the existing single member rural riding boundaries at all).

Combined Example: Vancouver Island: The following images illustrate how this combined model might work on Vancouver Island. The first map shows where the MLAs elected in 2017 came from. Despite the NDP winning just over 40% of the votes on the Island, they won 10 of the 14 seats (just over 70%), and the Liberal Party, despite winning 29% of the votes (1% more than Green Party candidates) only took one seat.

Results of the 2017 BC election for the 14 Vancouver Island ridings under our current First Past the Post voting system. NDP MLAs are shown in orange, Liberals in blue and Greens in green. The inset pie charts show the Island-wide shares of the popular vote and seats.

In using a combined approach, we might set aside 2 of the 12 seats as top-up seats and allocate 12 of the 14 in single and multi-member ridings (shown as yellow and green ovals, respectively). Here, we suggest creating a six-seat riding in the Greater Victoria area, a three-seater in the Parksville-Nanaimo-Cowichan Valley region, a two-seat Mid-Island riding, and a single seat North Island riding. With this configuration, we estimate that the NDP would take six seats, and the Liberals and the Greens three seats each. Note that two or more parties are represented in each multi-member riding – this already brings the seat shares Island-wide much closer in line with the vote shares by party.

Estimated results had the 2017 provincial election been run under an RUPR model with 12 seats in the first tier and 2 top-up seats. The yellow oval identifies a single-member riding; green ovals indicate multi-member ridings.

The two top-up seats are determined based on the shares of the popular vote secured by local candidates from each party. Here, we estimate that the two top-up seats would likely have gone to the Liberal and Green Parties. The vote share and seat share charts are now very closely aligned, so virtually all voters on the Island are represented by a candidate of their choosing.

Estimated results following selection of the top-up MLAs using the RUPR model.

Ballot Design

In single-member districts, the ballot would look essentially identical to what we use now. In multi-member districts, the ballot would be the same as described in the section on Multi-Member voting. Below is a sample ballot we might have for the three-seat Parksville-Nanaimo-Cowichan Valley riding.

VI RUPR Ballot Nanaimo

Here we’re suggesting that voters would be invited to rank candidates in numerical order, but, as described in the section on Multi-Member voting, we might also choose to ask voters to simply mark their preferred candidate with a single ‘X’.

Counting Ballots: The ballots for the single- and multi-member ridings would be counted as described in the section on Multi-Member voting. Essentially, the top local candidates would win seats, and these would tend to reflect the range of political views in each locale. If the Local PR riding protection rules are applied, we might call this variant LPR+ (see our Scorecard).

Determining Top-Up MLAs: There are a couple of possibilities here. Perhaps the most straightforward would simply be to determine, on the basis of first preference votes, how many seats each party would win if the seat distribution matched the popular vote. These numbers would be compared with the seats won in the single- and multi-member ridings, and the most under-represented party/parties would win the top-up seats, which would be awarded to the strongest remaining candidates from each party.

‘Integrated’ Top-Up Variant: The RU-PR system described above has the top-up seats explicitly assigned to a top-up layer that covers the entire region. This means that the electoral boundaries need to be slightly adjusted as there would now only be twelve first-tier seats in a region that formerly held fourteen. However, it would also be possible to simply include the top-up seats directly into the larger multi-member ridings (in this case, the Greater Victoria and the Parksville-Nanaimo-Cowichan Valley ridings). That is, the Greater Victoria riding would elect six seats using the multi-member ballot counting procedure, but leave the 7th seat vacant. Similarly, the PNCV riding would elect three seats, leaving the 4th vacant. The two top-up MLAs would be elected from their riding’s ballot into these last seats. In this way, we wouldn’t have to change any electoral boundaries. For the purposes of our Scorecard, we call this integrated top-up variant Multi-Member Proportional Plus (MMP+).


With either the explicit or integrated top-up approaches, the final group of candidates elected would broadly represent the diversity within each electoral district and would be the individual candidates most supported by the voters.