Tuesday, August 4, 2009

The right to an equal vote

www.politik.md has published an interesting little table:

On average, the votes per seat is 15,003, however there is quite a wide range between the votes per seat for AMN and those for the PCRM and PLDM. The PL and PD are in the middle.

www.politik.md points out that the constitution of Moldova enshrines a principle that each citizen's vote should be equal. Obviously this cannot be perfectly achieved in practice, however the question arises as to whether the result shown above is the 'most equal'. The aim should be to minimise the standard deviation of the observed votes/seat variable.

The standard deviation of the above results is 807. If I now take a seat away from the party with the 'cheapest' seats (i.e. the PLDM) and give it to the party with the most 'expensive' (i.e. AMN), I get the following result:

Now the range is much narrower and the standard deviation drops to 428, so this is a 'more equal' and hence 'more constitutional' result.

But what if I take the seat away from the PCRM instead of the PLDM? Here's the result:

The standard deviation for this arrangement is 420, making this an even more 'equal' and 'constitutional' result than the previous one.

Oh, and by the way, it would give the liberal bloc a golden vote again ;)


  1. Just used excel's solver to work out the optimum solution. This gives a standard deviation of 416 by taking the third solution above and transferring a seat from the PLDM to the PL. The final score, on the fairest possible basis would be PCRM 47, AMN 8, PLDM 17, PL 16 & PD 13.

  2. good job, nice to see that some are going beyond the routine news and actually provide some insight.

  3. It's interesting to note that the electoral code doesn't actually make reference to the D'Hondt formula per se. Article 87(1) makes vague noises about 'successive division', which seems to imply D'Hondt. The article is so badly written, however, that it's anyone's guess what was really intended.