Another winner-take-all system?

Yes.  Of course.  But what electoral system isn’t?

This objection usually arises from people who want proportional representation (PR), as if proportional representation itself wasn’t winner take all.


With single-member representation, where we use single-member FPTP, aka Single Member Plurality (SMP);  Instant Runoff Voting (IRV) (aka Alternative Vote (AV));  or… whatever including single-member Ranked-Pairs — we have one seat to fill, and the candidate who ‘wins’ gets that seat.  All the other candidates get nothing:  Winner-take-all.

Multiple- and Proportional-Representation

With multiple-member representation we use Multiple-Member FPTP, aka Multi-Member Plurality (MMP);  Single-Member Transferable, STV, et al, or whatever… including Multi-member Condorcet/Ranked-Pairs, and including as well proportional representation systems — we now have n seats to fill and the n candidates who win get those seats, while the other candidates get nothing:  Again — winner(s)-take-all.

When we go from a single-member representation system to a multi-member system, unless we want to increase the size of the given elected body, we must combine a number of single-member-representation areas, constituencies, into a single new constituency.  To elect n representatives for a given single area we’d combine n areas.  So at the end of the day, instead of electing one candidate (winner-take-all) in each of n areas, we’d elect n candidates (winner(s) take all) from the one combined area.  It’s the same number of winners in each case, and all equally “winner-take-all.”

Choosing the Winners

The difference is in how we choose those n winners.

With STV (which is multi-member but not PR) we choose the n ‘top’ candidates:  once a candidate receives enough votes to win (which number of votes depends on how many seats we’re filling), any votes exceeding this required minimum to win are considered “wasted” so these so-called excess votes are reallocated to next-preference candidates using a weighting algorithm so that “the whole ballot” is used.

In doing this more candidates from more-popular parties will tend to be elected, which will have a tendency to look like proportional representation, but since party isn’t actually taken into account in the vote allocation it’s not really PR, though it is often called that.

With PR we’d choose the winners based on the relative percentages of party-based support:  we vote primarily for the party (and in an open-list system, secondarily for candidates within the party to adjust their order in the party-list), and then elect a number of candidates from party lists according each party’s relative proportion of the over-all vote and the number of seats to be filled.

This delivers a result that is intended to approximate the proportional support of each party in the over all results (to the limit of the granularity permitted by the number of seats involved), and that’s significant for those who believe that the outcome should reflect the overall party support, but it is nonetheless still winner-take-all.

The Best of Both

Mixed-member PR (MMPR) is a bridge between these approaches.

With MMPR Some number of representatives are elected to single-member constituencies, and these numbers are topped-off by electing the balance of the legislature from party-lists in order to achieve, taking account the single-member constituency members as well, party-proportionality for the legislative body as a whole (to the limit of the granularity permitted by the number of seats involved).

The size of the legislature will typically remain the same, or nearly so, when converting from a wholly single-member constituency system as the number of single member seats are reduced, leaving the balance to be elected on a party-list basis over a given region, or for the body as a whole.

So in the end we get party proportionality for those who care about that, as well as single-member accountability for constituencies for those who care about that as well.  But all, nevertheless, winner-take-all.

Condorcet/Ranked-Pairs MMPR, augments such a system so that the elections for the individual constituencies are done by Condorcet/Ranked-Pairs, rather than FPTP — given that whatever the virtues of proportionality, they’re not enough to absolve the distortion of the FPTP elections themselves.

Condorcet/Ranked-Pairs instead of FPTP, means that those single-member representatives are chosen based on the true preferences of the electors, and this, topped-off by party-preference party-list elections to give over-all proportionality, arguably delivers the best of both.


PR has its virtues (and faults), as does any other system, but being somehow “not winner-take-all” is not one of these virtues.  Support PR by all means, if you believe party-proportionality is a better way to elect your representatives, but, please, not because of a flaky claim that somehow it’s not winner-take-all!

Bookmark the permalink.