Sainte-Lague/Schepers
The German physician Hans Schepers, at the time Head of the Data Processing Group of the German
Bundestag, in 1980 suggested that the distribution of seats according to d’Hondt be modified to
avoid putting smaller parties at a disadvantage. By a different calculation method, the procedure
proposed by Schepers arrives at the same results as the method developed by the French
mathematician André Sainte-Laguë in 1912.
The procedure according to Sainte-Laguë/Schepers has been used since 1980 for the distribution
of seats in the committees and bodies of the German Bundestag, and since the Election to the 17th
German Bundestag in 2009 for the distribution of seats in Bundestag elections. In 2009, the
procedure was for the first time applied also in a European election. The seats in the Länder
parliaments of Hamburg and Bremen have already been allocated according to that procedure. In
Nordrhein-Westfalen, too, the seats have been allocated according to Sainte-Laguë/Schepers since
the 2010 Election to the Land Parliament. Baden-Württemberg and Rheinland-Pfalz have
introduced the procedure according to Sainte-Laguë/Schepers for the distribution of seats in the
forthcoming elections to their Länder parliaments.
In this procedure, which is also called
divisor method with standard rounding, the respective numbers of second votes cast for the
individual parties are divided by a joint divisor. The resulting quotients are rounded according to
standard practice to obtain numbers of seats, i.e. the figure is rounded up or down when the
remaining fraction is larger or smaller than 0.5, where the residual equals 0.5 exactly, a lot will
be drawn. The divisor is determined in a way which ensures that the total of the numbers of seats
equals the total of the seats to be distributed. Three different methods may be used for the
calculation which produce the same result and thus are regarded as equal from the legal
perspective:
-
Highest average method: this method follows the train of thought on which the procedure
according to d’Hondt is based, with the respective number of votes being divided by 0.5, 1.5, 2.5
etc. and the seats, in turn, being successively allocated by descending maximum numbers. The
calculation according to d’Hondt is based on the full entitlement to a seat and therefore uses
whole numbers for division, with smaller parties obtaining their first and further seats
disproportionately late. In comparison, the requirements for a seat to be assigned have been
lowered with this procedure. Once there is an entitlement to more than half a seat, it is already
allocated.
-
Rank order statistic procedure: here, the inverse values are considered instead of the
maximum numbers and the seats are successively assigned according to these ascending rank order
statistics.
-
Iterative procedure: with this method, an approximate allocation is calculated in a first
step. The total number of votes to be considered is divided by the total number of seats to be
distributed, thus determining a provisional divisor for allocation. Any remaining discrepancies are
reduced in the following steps by increasing or reducing the divisor until the final allocation has
been found where the distribution of seats corresponds with the number of seats to be
distributed.
For the distribution of seats in the elections to the German Bundestag, the legislator
selected the last-mentioned
iterative procedure when the procedure according to Sainte-Laguë/Schepers was introduced.
For the above example, the distribution of seats would be calculated as follows:
|
Procedure in accordance to Section 6 (2) of the Federal Elections Act with divisor for
allocation
|
|
Formula:
|
|
Party’s number of second votes
________________________
Divisor for allocation |
= Party’s number of seats
|
(after standard rounding)
|
|
Determining the divisor for allocation
(Criterion: allocation of as many seats to Land lists as there are seats to be
distributed) |
|
Total number of second votes to be considered
_____________________________________________
Total number of seats to be distributed
1 |
=
provisional divisor for allocation |
|
If necessary, increasing or reducing the divisor for allocation until the total calculated
corresponds to the total of the seats to be distributed. |
|
1 Total number of seats minus the seats obtained by successful individual candidates
(constituency nomination pursuant to Section 20 (3) of the Federal Elections Act) or successful
party candidates, where the party has obtained less than five percent of the valid second votes and
fewer than three direct seats or has not been admitted with a Land list in the respective Land
(Section 6 (2) sentence 6 in conjunction with Section 6 (1) sentence 3 of the Federal Elections
Act).
|
Allocation of 8 seats
|
1st step:
|
17 500
__________
8
|
=
2187.5
|
= provisional divisor for allocation
|
|
Party
|
Calculation
|
Result
|
Result seats
to be
after standard = distributed
rounding accordingly |
|
Party A
|
10 000
___________
2187.5
|
= 4.57
|
5
|
|
Party B
|
6 000
____________
2187.5
|
= 2.74
|
3
|
|
Party C
|
1 500
____________
2187.5
|
= 0.69
|
1
|
As a total of 9 seats is attributable to the parties when using the divisor 2187.5 for
allocation purposes while there are only 8 seats to be distributed, the divisor has to be increased
until the calculation of the allocation of seats sums up to the number of seats to be distributed.
To this end, the calculation is repeated with the higher divisor of 2300:
2nd step:
|
Party
|
Calculation
|
Result
|
Result seats
to be
after standard = distributed
rounding accordingly |
|
Party A
|
10 000
____________
2 300
|
= 4.35
|
4
|
|
Party B
|
6 000
____________
2 300
|
= 2.61
|
3
|
|
Party C
|
1 500
____________
2 300
|
= 0.65
|
1
|
The procedure according to Sainte-Laguë/Schepers eliminates paradoxes that may occur when
seats are distributed according to the Hare/Niemeyer method.
Last update: November 2010
See also:
©2013 The Federal Returning Officer