Electoral reform and the (un)importance of simplicity

By JD Mussel

Perhaps among the most frivolous arguments one is likely to hear against any electoral reform is that a proposed system is ‘too complex’. Let me illustrate:

The 1953 Dutch State Commission on the electoral system had this (among other things) to say in consideration of a reform along the lines of the Danish system:

After further consideration of the pros and cons, the commission will not conceal that there are also reservations to be expressed against this system. We fear that a large part of the electorate, accustomed as it is in principle to a simple system, as is established by our present electoral law, will not be attracted by it, probably at the expense of the so desired political interest.

Under the ‘Danish system’ as reviewed by the commission, voters can vote either for a candidate on a party list or for the list as a whole. Seats are first divided proportionally among parties in multi-seat districts, while an ample number of ‘levelling seats’ are then distributed in a compensatory fashion, resulting in a very proportional result overall.

To be fair, the system used in the Netherlands is, on the face of it, quite a bit simpler: seats are simply distributed to parties in proportion to their national vote in one nationwide district, being allocated first to candidates having received more than a certain threshold (in 1953: half of the quota) and then according to the party-list ordering.

However, on a closer look, the Dutch system also includes nomination-districts, within which parties can, but do not have to, propose different lists (a practice which was still relatively common, although limited, in 1953, but which has all but disappeared by 2006). This does not affect the allocation of seats to parties, but can rather complicate the allocation to candidates.

But does any of that really matter? I somewhat doubt that there are too many voters in the Netherlands, or in Denmark, for that matter*, know the intricacies of the D’Hondt formula, the workings of nomination districts or even the number of preference votes candidates need to be elected in their own right.

Does this undermine the electoral process or people’s interest in politics? Of course not. As long as the system is generally responsive and the act of voting is not a complicated or confusing hassle, the complexity of an electoral system shouldn’t pose any problems. Few systems have ever really made voting difficult after voters got used to them, even in countries with illiteracy problems. Australian Senate STV before the introduction of above-the-line ticket voting was one of them, though even then the vast majority of voters still managed to cast a formal ballot despite the increasingly onerous nature of the task. In the case of the Netherlands, adopting the Danish system could actually make voting simpler, considering that the current bed-sheet sized ballot would become smaller (due to the lower district magnitude) with otherwise no changes to it; apart from the candidates, most voters would hardly notice any difference.

When it comes to electoral reform, ‘it’s too complex’ is usually little more than a red herring.

* Although the commission would seem to imply that Danish voters are better at this than the Dutch…

25 thoughts on “Electoral reform and the (un)importance of simplicity

  1. Thanks, JD. Very interesting.

    On the Netherlands practice, I thought parties actually had to present lists that were at least somewhat different in different nominating districts. That is, that if they repeated the same list in each nominating districts, the larger parties might not have sufficient candidates for their nationwide seat entitlement. Is that incorrect?

    • Virtually all parties currently submit the very same list in each nomination district. The current limit on the number of candidates is high enough to allow this, though perhaps this was not always the case – I’m not sure. I’ll see if I can find some more information about that.

      • Thanks, JD. I recall the point from the chapter on the Netherlands by Rudy Andeweg in the Gallagher and Mitchell (2005) volume. It could have been a historical point, not relevant to the then-current system, though I surely recalled that they meant now (i.e. 2005).

        Speaking of Gallagher and Mitchell, the title of the chapter on Denmark by Jorgen Elklit is classic, “Simplicity embedded in complexity (or is it the other way round?)”.

  2. Although I agree that often the issue of complexity is unnecessarily used as a counterargument against electoral reform, I think you should not dismiss it right away. Some systems are more complex than others. Assuming that it is important that voters feel their vote is counted in a fair way, their understanding of the system is quite relevant I think. Even if the electorate does not understand every detail, it often does understand the main lines.

    While saying that all electoral systems but the Dutch one are incomprehensible to the Dutch voter clearly is nonsense, the relative ease of comprehension of a system is relevant, because people need time to adjust to the new system. Path dependency plays a role here. A new system should only be adopted when the benefits of the system outweigh the adjustment costs(that is, the time it takes the electorate to familiarize itself with the new system, and potential temporary lower faith in the democratic system due to people not understanding the new system, and thus potentially thinking it ‘unfair’). Taking the case of the Netherlands, because the nomination districts are not actually used by parties anymore to have different eligible candidates, it is not crucial to understand as a voter. The Danish system thus probably is more complex than the Dutch system, thus adopting this system is more costly than adopting another simpler system, which in turn is more costly than keeping the current system.

    I think the issue boils down to the empirical question of how strong the path dependency is and which electoral systems present higher adjustment costs than others due to their complexity. While not saying that the ‘argument of complexity’ should override any proposal to reform, dismissing adjusment costs out of hand is uncalled for I think. When the benefits of reforms outweigh their costs, reforms should take place, in case they do not, reforms should not be undertaken. The complexity argument thus is relevant to the degree that the complexity determines these adjusment costs.

    • I don’t wish to completely dismiss complexity; I simply believe that serious problems arising from it are rare, even among situations where it is an important issue. There are two key issues arising from complexity: 1) do people understand how to vote? and 2) do people broadly understand how votes translate into seats? I definitely think that these are relevant, but in the vast majority of cases they are not an obstacle, and therefore not really an argument against reform. For both, a public awareness campaign of some sort will go a long way, and the more complex the system, the more attention should be given to it by the authorities, although the media will undoubtedly also play a major role in any case.

      While I don’t deny that complexity *can* be an issue, I think it’s usually very much overstated, even when noted as a serious argument rather than an excuse. Dutch voters, even both when parties *did* nominate separate lists, had no problem with the complexity of the system and the impact on the categories I offer above (1 & 2) was very much negligible. The same would be true if the Danish system were adopted.

      My comments in the post are ultimately mostly about the sort of reforms that have been proposed in real life, which is why I offer such examples; they are not about the full range of possibilities, which is of course virtually endless. More complex systems could certainly be proposed, and complexity could certainly become an important issue. I am simply saying that thus far, the problem has generally been exaggerated and/or misused.

  3. Clearly, and it’s not surprising that politicians have constructed and enacted “complicated” electoral rules when it served their interests. Here I have in mind early American uses of various numbered-post systems, as well as later American and Canadian variants of ranked, single-winner systems.

    • “A new system should only be adopted when the benefits of the system outweigh the adjustment costs” (Jose) seems like a good principle to me, and pretty close to what Rein Taagepera and I argued in a piece in Electoral Studies in 1990 or thereabouts.

      What is a numbered post system, Jack. (And I am not sure why your comment was held in the moderation queue.)

      • I thought I had left a reply on Thursday, when I was doing it from a smartphone in a moving vehicle (not driving), but I guess not.

        By numbered post, I meant plurality where the voter can V≤M, but each of the M winners must come from one geographically defined district. GA and MD, I think, used these rules in the federal elections of 1789, and they made a brief resurgence post-1842 Apportionment Act. I have a chart here.

        It would be nice to know whether V<M invalidated a ballot. It would also be nice to know whether each of a voter's V choices had to come from different geographical districts. Both would take some sleuthing.

      • Thanks, Jack. I have never heard that term before. However, I recall that at least up to the 1980s, Santa Ana, California, had the usual (for municipalities of its size or smaller) system of at-large plurality (V≤M), but the winners somehow had to be from sub-divisions of the city. It was the only such case I ever knew of in Orange County, where most were straight winner-take all, at-large, with 3 or 2 elected in alternate elections.

      • I think I saw the term “numbered post” many years ago in a FairVote-LWV primer on municipal electoral system choice. Functionally analogous, more or less, I guess, to MNTV with nomination districts.

        (Not sure why I can’t reply directly to your reply below, but no biggie.)

      • Sounds similar to the Lebanese system, except the winners have to come from different religious/ethnic groups.

  4. I have a vague recollection of an APSA presentation by Randy Stevenson of survey work he and Ray Duch did (most likely for their book) that showed that voters are as sophisticated as they need to be, given the complexity of the electoral and (relatedly) party systems they must deal with. That is, voters learn as much or as little as the system demands. I don’t have a precise cite, but it stuck with me – made me think at the time that maybe civic education is a good argument in favor of complexity!

    On the other hand, the mere presence of complexity can become a point of grievance. Witness the Electoral College / popular vote “reversal” in the 2000 US Presidential Election (and a few prior elections).

    • “voters are as sophisticated as they need to be, given the complexity of the electoral and (relatedly) party systems they must deal with.” That’s precisely what I’d expect – but the important thing here is that the necessary sophistication consists of understanding how the ballot works and little more. Understanding how the system then converts votes into seats is of little importance, and when it is important voters won’t need much to realise what it means for their vote, for example which parties are likely to exceed the threshold.

      Arguably popular vote reversals such as the electoral college in 2000 were not a problem of complexity per se but of unresponsiveness, and then you would argue against the system on the latter basis. But there are probably cases where the unresponsiveness or other grievance is inherently connected to the complexity. The ACT’s D’Hondt-STV hybrid perhaps? (Although I’m apt to say that had more to do with voting, as people were used to voting preferentially in a way that wasn’t allowed under the system. Can someone shed more light on this case?

      • There were numerous problems with the ‘modified D’Hondt’ system employed in the ACT. Voters didn’t really want the Legislative Assembly anyway, so any voting system would be unpopular. The threshold of one Hare quota was perceived to be fairly arbitrary. Tony Fleming and the Fair Elections Coalition achieved 99% of this quota, but were shut out. Then, there were the preference transfers after the exclusions. Group voting tickets previously lodged by parties were used for this (under certain circumstances) and in one case, the preferences of voters for the frivolous Sun-Ripened Warm Tomato Party went to the Residents Rally (a self described ‘community based urban green party’). The flow of preferences to the Rally meant that they received more seats than the No Self Government Party, despite winning fewer votes. The Labor Party (the largest in the assembly) recieved just 22% of the vote, and were forced into minority government. Eventually, the Labor government fell, and the Liberals (with four seats off 17% of the vote) went into coalition with No Self Government (3 seats) and Residents Rally (4 at election, but with one defection). Eventually, this coalition collapsed, and Labor came back just in time for the 1992 election. At this election (held alongside a referendum that chose Hare-Clark STV for future elections) D’Hondt nearly elected a majority Labor government off 40% of the vote. Voters had no trouble with the actual mechanics of voting (all that was needed was a valid first preference for a party or candidate), but the results, for an area used to majority governments, were utterly perplexing. For some info on the system and results, see the Elections ACT website.

  5. I suppose there is a difference between everyday (or every election) complexity, and that which arises as consequential only every generation or so.

  6. Regarding pre-ticket voting STV in the Australian Senate, the main problem was not the complexity of the system (random transfer last parcel STV) or whether the ‘vast majority’ of voters could cast a formal vote. The problem was that the system advantaged the Liberal/National party over the Labor party. Labor voters, at that time, were less well educated, and were thus less likely to fill out a formal ballot paper with full preferences. In 1974, a group of activists opposed to the Whitlam government stacked the Senate ballot paper in New South Wales in an attempt to get Labor voters to make mistakes. The informal vote in the Senate went up from 10.1% to 11.1%, and while the House of Representatives vote in NSW was 53%, the Senate vote was 50%.

    • Well, clearly the problem was indeed that it was too complicated to cast a formal vote. I don’t think it a situation where it would have affected both parties equally would have been possible, but if it were, and the rate of informal votes had included equal parts Liberal and Labor supporters, it would still have to be seen as a problem.

      • Yes, but the point I am trying to make (and was cut off in making by an iPad), is that a ‘complicated’ electoral system can be kept by politicians who have their needs served by it. STV for the Senate was introduced by the Labor government in preperation for the 1949 election, which they thought that they would lose, in order to ensure that the incoming Menzies Liberal government would not have a majority in the Senate. I’m not quite sure why they opted for full preferencing (perhaps consistency). The system was kept by the Menzies government when it was made clear to them that STV would both cause Labor voters to vote informal and leave them with a majority in the Senate. The Whitlam Labor government would have probably liked to change the system, but did not have the numbers in the Senate to do so. The Fraser Liberal government didn’t really mind (for reasons mentioned above), and it took the Hawke Labor government, which had a vested interest in decreasing informal voting rates, to introduce the (possibly even more complicated) group ticket STV system.

  7. I agree with Joes de Natris above. The problem is path dependency, and “complexity” when it is used to criticize a proposed change in the electoral system usually really means “too different”.

    If you change the rules on voters, they assume that you are pulling a fast one, and its up to the changers to show that its still basically the system they are used to with a few improvements, or adopting a totally new system is really, really necessary. When the governments and the old system lose their legitimacy, as happened in New Zealand during a bout of neoliberalism in the late 1980s and early 1990s, you can make a big change. If the system is still seemed to work well enough, such as the UK, its hard to make even fairly minor changes. There are situations where big changes have a higher probability of being adopted than small changes.

    The problem with STV isn’t the complexity, its that its not been used that often.

    Its not helpful for academics to insist that the proposed system really isn’t that complex, which is the usual response. First, complexity isn’t the issue, and second the system still has to be sold.

    • You may very well day that the problem with STV isn’t complexity, but that’s exactly what you’re going to hear from politicians and other who really just prefer the current system (and that’s one of the main things that were said against it in the Netherlands when it has been proposed before (and after) the adoption of PR – a debate which perhaps deserves a post to itself). And that’s my main point: ‘complexity’ is usually just an excuse. The only area where complexity is likely to matter is voting itself, and where STV has been adopted, voters have generally adapted rather quickly and successfully. In the example I gave above from 1953, the notion that the added complexity would have been a problem for voters just makes no sense; the only change – a smaller ballot – would, if anything, make voting simpler.

      I completely agree that complexity isn’t the issue; that’s why I called it a red herring. The problem is that I see too many people accepting the ‘complexity’ argument without any burden of proof for why that even matters. Complexity may sometimes pose problems (though I think that’s rarely the case, but that doesn’t necessarily mean that simpler = better. More often than not, a system’s complexity is a deliberate feature meant to introduce a certain effect. I suppose you’re right that, outside our circle, it isn’t helpful for academics to insist a system isn’t all that complex; instead, the correct response is ‘so what?’

      • I’ve heard MMP described in the Australian media as a ‘complex’ system when they report NZ election results. What they actually mean is ‘We are trying to call a horse race here and the government formation could take weeks!’

        Apparently they have never visited Florida where the simplicity of FPTP somehow produced a result that took weeks to resolve in 2000.

  8. MSS: I assume a “numbered post” system means a “slot” system, ie MNTV but where candidates may nominate for only one of several otherwise identical seats representing the same electorate. Eg, rather than “tick up to three candidates” for three seats, a voter may tick one of the candidates for Seat A, one of a different set of candidates for Seat B, and likewise for seat C. I remember reading one or another US State Constitution (a couple of decades ago, in the Ghisbert H Flanz collection in the ring-binders) that mandated this system. It’s thus midway between pure MNTV at one end (a single pool of candidates for all of several identical positions) and “vote for one candidate for Governor, one for Treasurer, one for Lieutenant Governor, etc” at the other (a separate pool of candidates of each of several distinct positions).

    Henry: Modified D’Hondt used the Droop, not Hare, quota as the threshold. 5.5555…% for 17 seats.

    I think Ed is on the right track. Australian politicians occasionally propose changing from compulsory to voluntary voting, or from optional-preferential AV to full-preferential AV (or vice versa). These changes are not complex and are, I can attest, easily grasped in principle by the bloke who cuts my hair or the grandma waiting at the bus stop, even if the longer-term effects are the subject of debate by political scientists. Nonetheless they are also regarded with great suspicion. On the other hand, STV-PR has been introduced over the past 35 years in NSW, WA, the ACT and Victoria (in the NSW and ACT cases, was adopted at one referendum and then confirmed at another) without running into any of the forehead-thumping ignorance that the British proudly displayed in the 2011 AV referendum campaign.

    As MSS has noted, the media and some politicians tend to lazily describe any proportional voting system as “complex” – even the Israeli Knesset method, which can be explained in 90 seconds to an intelligent ten-year-old.

  9. Further thoughts on the larger philosophical question…
    There’s “a simple[r] process” (few[er] moving parts) and “a simple[r] result” (a clear[er], [more] monotonic criterion for success or failure), and the two often don’t coincide. “Put a man on the moon by the end of this decade” is a simple goal even though the mechanics are, er, rocket science. “Let’s change to 366-day years with 31 days for every odd-numbered month and 30 for every other month”, on the other hand, is a simple process but the ramifications would be immensely complex.
    A typewriter is much simpler to build or repair than a word processor, but a word processor is much simpler to use. Hard to believe now but, even 20 years ago, the average Western citizen didn’t type. Most university students hand-wrote their assignments. Offices had a special typing pool with trained professionals. Most of those who could type learned it as a special subject at school. Today, using a word processor or smart phone (if you have access to one) is much simpler (grammar, spelling and syntax are a separate issue).
    Relating this to electoral systems, compare STV to SNTV. The latter is much simpler to describe and legislate. It is however much harder to voters and organised parties to operate, at least if you care about maximising the number of your highest-preferred candidates elected and/or defeating candidates you dislike. STV, on the other hand – provided candidates’ party labels are shown and the minimum number of preferences required is not excessive – is very simple for the voter.
    (It is true that academics can identify rare situations where, in hindsight, STV might have rewarded tactical voting. But this is a long way from demonstrating that STV ever offers any rational incentive to vote tactically. And when these situations are highlighted by proponents of FPTP or Borda, for example, I feel rather as if I’m being warned against the grave dangers of card-counting at the blackjack table – why, don’t you know, clever players can game the system!! – by someone who’s promoting professional wrestling as a better alternative…)

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

This site uses Akismet to reduce spam. Learn how your comment data is processed.