[EM] proportional constraints - help needed

[Peter Zbornik]

Kristofer,

to be more exact:
I need not just proportionality in the ordered list as a whole (i.e.
meaning proportional ranking), but also that seats/candidates are
quoted in proportionally within each gender too.

Proportionality within each gender is not needed, if the constraints are met.

I.e., the guiding principle when constraints are not met and when
deciding upon which seat to apply quotas on should be
i. to quote-in the seat proportionally within the gender, but
ii. without causing an unnecessary "disproportionality" within the
ordered list.

This means for instance, that if we have to decide if we should apply
the constraint (quote-in) at seat 3, 4 or 5
and the proportionality within the gender would be identical in each case,
then the candidate should be quoted-in at seat 5, since seat 5 is
"less important" than seats 3 and 4 and since there is no gain in
proportionality within the gender by quoting in at seat 3 or 4
compared to quoting-in at seat 5.

It is necessary to quantify what "less important" above exactly means,
but I am not sure of how to do it.

P.

2013/2/6 Peter Zbornik <pzbornik at gmail.com>:
> yes, that's it.
>
> P.
>
> 2013/2/6 Kristofer Munsterhjelm <km_elmet at lavabit.com>:
>> On 02/05/2013 09:37 PM, Peter Zbornik wrote:
>>>
>>> Hi Kristofer,
>>>
>>> I am afraid your approach might in some cases not lead to
>>> proportionally distributed quoted-in candidates.
>>>
>>> For instance, say we have three coalitions: A, B, C.
>>> Coalition A and B get their first place candidate
>>> Coalition C get their second place candidate quoted-in (i.e. they
>>> would prefer Agda, but they get Adam due to the quota rules).
>>> Coalition A and B get the third and fourth place candidates respectively.
>>> Coalition C, again, get their fifth place candidate quoted in (i.e.
>>> they would prefer Erica, but they get Eric due to the quota rules).
>>>
>>> This approach leads to an unproportional distribution of quoted-in
>>> seats (candidates) as Coalition C get both of the quoted-in candidates
>>> and Coalition A and B get none.
>>
>>
>> So you need not just proportionality in the group as a whole, but
>> proportionality within each gender too?
>>