Tunisia preliminary results

See the Tunisian Election Live Results Table for regularly updated district-level results.

There are still many districts not reported, and some missing data in those districts that are reported. However, the pattern is clear. Out of a total of 217 seats, 145 have been allocated as of my check. The Ennahda has 60 of these, which is 41%. The next largest parties, Congress for the Republic (CPR) and Aridha Chaabia (Popular Petition) have 21 and 20, respectively. Then Ettakatol with 13 and PDP with 6. No other party currently has more than 4. Thirteen parties are currently on one seat apiece.

The picture is thus one of a single dominant, but likely not majority, party, and a fragmented rest of the field.

At the district level, the expected pattern for the simple (Hare) quota and largest remainders (SQLR) system can be observed. This system is well known among electoral-systems specialists for its tendency to benefit small parties. Seats are much “cheaper” in terms of votes required to win, via remainders than via quotas.

The way SQLR works is that a quota is defined as 1/M of the votes, where M is the magnitude of the district (the number of seats available). In the first state of seat allocation, a party wins seats according to how many of these full quotas it has. Whatever number of seats are not yet allocated after seats by quota are determined are then allocated via largest remainders. In this stage of allocation, the parties’ remainders are ordered from largest to smallest, and a seat–at most one per party–is assigned to each in descending order until all are filled. ((It is this incorrect to call systems like this a “largest remainder” system, as is often said, because that term describes only the stage of allocation that takes place after quota seats are assigned. There are several different quotas (for instance, 1/(M+1) instead of 1/M), and the definition of quota that is used will have a significant effect on the remainders that parties have left over once quotas are assigned.))

A party’s seats in each district are thus the sum of its quotas earned, plus a remainder seat, if any. Under SQLR a large party will use up most of its votes on quotas, and then be eligible for at most one remainder seat. Smaller parties, on the other hand, often will have their full vote totals in the district as their remainder, and hence the last seats allocated in a district often require a much smaller number of votes to win than did the first (quota) seats that were assigned.

The process, and its impact on the cost of a seat, can be demonstrated by reference to a couple of the declared district results.

In the district of Gabes, M=7, Ennahda won 4 seats, and three other parties have one each. All three seats allocated to parties other than Ennahda were remainder seats. Total votes were 138,375. Ennahda had 73,416, or 53.5%. So it is only slightly over-represented, with 57.5% of the district’s seats. Its cost per seat is 18,354 (73,416/4). Each of the three other parties that won a seat has fewer votes than this, and the smallest of them paid only 7,351 votes (5.3% of the total) for its one seat, or 40% of what Ennahda paid for each of its four seats, three of which it won via quotas, and one by remainder.

A second, very striking, case shows how SQLR can sometimes under-represent the largest party. In Jendouba, M=8, Ennahda won 2 seats, and six other parties won a single seat apiece. Ennahda’s votes were 33,136 out of 118,376, or 28%. Yet it won only 25% of the district’s seats. The quota is 118.376/8=14,797. Thus both of Ennahda’s seats were won via quota, and it paid 16,568 votes for each of these. The other parties that won a seat each had vote totals ranging from 12,433 to 3,599. So the smallest party won a seat at a cost that was a mere 21.7% of Ennahda’s per-seat cost. Ennahda missed winning a third seat by 57 votes, as its remainder was 33,136-(2*14,797)=3,542. One party winning a seat on 3% of the votes cast, while another wins 2 seats on 28% is an odd result, but one that is inherent to the formula used, SQLR.

Tunisia Live does not seem to have a running total of national-level votes, so it not possible to tell the vote percentage upon which the preliminary 41% of seats for Ennahda is based. As the two district results discussed here show, SQLR can either over-represent or under-represent the largest party, depending on whether the precise vote totals allow the largest party to win one of the remainder seats or not, and of course on the district magnitude.

It is possible that Ennahda has won a bit more than 41% of the votes. It is also possible that it has less than 40%. Even the latter would imply a lesser degree of over-representation than most “proportional representation” systems would provide the largest party.

With one party so dominating the rest of the field, and smaller parties earning seats much more cheaply than the dominant party, we might anticipate that SQLR will not be the electoral system that this Constituent Assembly adopts for the next election.

11 thoughts on “Tunisia preliminary results

  1. It will be interesting to see whether women end up significantly under-represented, despite enforcement of “zipper lists”. Most lists seem to be headed by men, and with so many parties and small district magnitude, there will surely be many incidences of parties winning odd numbers of seats off a list.

  2. Yes. In the previous thread, I posted that “only 6 percent of the more than a 1,000 lists are actually headed by women.” And, of course, that is out of all lists, not the most viable ones.

    We now can see that, as I anticipated, there are a lot of lists that elected only their top-ranked candidate. That means the share of seats held by women will be well below parity, despite the gender quota of 50%, alternating.

  3. By my count of what is currently reported, 73 of 159 allocated seats (45.9%) are the sole seat won by their list in their district.

    Sidi Bouzid is the only district so far declared in which Ennahda is not the largest party. (No seats from Tunis are reported yet.) Ennahda still won 2 of the 8 seats in Sidi Bouzid.

    Only 5 of the 20 districts reported thus far have a party other than Ennahda with more than one seat.

    There is no district at present in which Ennahda does not have more than one seat.

    The upshot of all this is that almost the only women elected will be those on the Ennahda slate, and Ennahda will have at least one woman from every district! (Again, based on the 20 districts already declared.)

    (All these counts based on domestic Tunisia seats only.)

  4. I wonder if the electoral law has a provision that a woman who steps aside must be replaced by the next available woman, or the next available candidate on the list.

  5. The next candidate on the list is the replacement, irrespective of gender (a male list-header will be replaced by a woman)

    Décret-loi n° 2011-35 du 10 mai 2011, relatif à l’élection d’une assemblée nationale constituante:

    Art. 23: “En cas de vacance d’un siège au sein de l’assemblée nationale constituante, le membre en question sera remplacé par le candidat suivant dans le classement de la même liste.”

    (more recent decrees can be relevant too, but they are still not yet translated in the Journal Officiel de la République Tunisienne)

  6. With all results in (unfortunately, only for lists with seats), it is clear the SQLR method was wisely chosen if they wanted to constrain the largest party: other methods would have given Ennahda (37% of votes) far more seats:

    SQLR : 91 seats (42%)
    Sainte-Laguë : 122 seats (56%)
    D’Hondt : 155 seats (71%)

    Note the remarkable difference between SQLR and Sainte-Laguë.

    All other parties would suffer, except for ‘The Initiative’.

    PS to MSS: The figures you use in the Jendouba example, seem to have changed: the vote total was only 112974, with a quota of 14122 and Ennahda winning a third seat with a remainder of 4893.

  7. You would not believe how difficult it is to get across to people – even (or perhaps especially) those who know a little about how PR systems work – the admittedly counter-intuitive point that “If you reduce the quota [ie, substituting the D’Hondt or the Droop for the Hare quota, and assuming no extra seats are vacant], you actually make it harder for smaller parties to win a seat.”

  8. Tom, I hadn’t realized that myself! I knew that small quotas favoured large parties overall, but I suppose I assumed that gaining a first seat might be easier. It took a bit of math to convince myself otherwise, I’m glad I took the time to work that out. Thanks!

    Now I have to try to explain it to someone else and see whether it’s as hard as you claim…🙂

  9. Vasi, I first heard of this when the Australian Senate was increased from 5 vacancies to 6 per State per half-Senate [periodic] election in 1983-84. The (now-defunct) Bulletin magazine pointed out that adding an extra seat could actually make it harder for the (now-defunct) Aust Democrats to elect a Senator with 8-10% of the first-preference votes. Labor and Coalition would each take 3 quotas on first preferences (42+%), leaving none up for grabs after eliminations. If the votes fall a particular way, that can in fact happen.

  10. From Josep M. Colomer, Handbook of Electoral System Choice, 2004

    “More generally, the choice of electoral system seems to follow what could be called the ‘Micro-mega rule’, by which the large prefer the small and the small prefer the large: a few large parties tend to prefer small assemblies, small district magnitudes and rules based on small quotas of votes for allocating seats, while multiple small parties tend to prefer large assemblies, large district magnitudes, and large quotas.

    In a nutshell, large parties prefer small institutions in order to exclude others from competition, while small parties prefer large institutions able to include them within.”

  11. John Carey offers a detailed analysis of how SQLR worked in the Tunisian election, and how it might have been different under D’Hondt.

    In fact, Ennahda might have won more than 2/3 of the seats with D’Hondt!

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 )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s