[Election-Methods] Closed or open primary elections? (was: Why monotonicity?)

[Juho]

This is at least in principle an interesting area of study. In  
addition to having strategic selfish voters in this type of elections  
we may have also malicious voters (whose intention is not to elect a  
candidate that would be good for themselves but possibly a candidate  
that is so terrible that nobody wants him, or a candidate that would  
lead to problems like not being elected in the actual election after  
the primaries).

This could in principle happen also in other elections than primaries.

In a two-party system with simple linear left-right opinion structure  
it would e.g. make sense for the malicious voters to vote for the  
most leftist candidates in the left primary since they are probably  
less preferred by the right wing party, and the right wing of the  
left party too.

It makes sense to invite right party voters in the left party primary  
since that may lead to electing a candidate that is quite close to or  
otherwise acceptable to the right party. But of course if the right  
party voters are strategic, then they might vote for the extreme left  
candidates. It thus depends on the expected level of strategic voting  
if one wants to limit participation in primaries to a closed circle  
or not.

(I sort of assume that the most efficient way to win the final  
election is to have a candidate that is seen to be a centrist and  
even slightly on the side of the other party. The own party voters of  
that candidate will vote him anyway and will tolerate some political  
rhetoric around the values of the other party, while it is crucial to  
appeal to some voters of the other party to steal the last required  
votes from them to change the balance in favour of one's own party.)

I wonder how many of the primary voters today vote for the candidate  
that they consider to be the best, and how many vote for the  
candidate that they think has the best chances to win against the  
candidates of the other party.

In Finland in the presidential elections of 1993 the Social  
Democratic Party used a primary that was (exceptionally) open also to  
others than party members. Mr Martti Ahtisaari who had not been that  
active within the party won the primary and finally also the actual  
election. Mr Ahtisaari was elected in the primary although there were  
also other more distinguished party internal candidates (i.e. a long  
time party leader that many considered to be the automatic first  
choice of the party). I'm sure the party external voters must have  
contributed (more or less) towards electing Ahtisaari, and making a  
Social Democratic Party candidate win the final election. The idea of  
arranging a primary for "all the people" may also have led all the  
voters (also party internal) to evaluate the candidates using more  
general than just party internal criteria. I'm not aware of any  
strategic voting. Most voters, especially Ahtisaari supporters  
outside of the party (and non-registered party supporters too) maybe  
saw this just as an open (somewhat populistic) opportunity to  
influence positively.

I have only very limited understanding of the US primary system but I  
have understood that it is quite easy to become registered as a voter  
of "the other party" and participate in their primaries. Negative  
campaigning seems to be quite popular nowadays, and widely spread  
registration and voting in the primaries of the other party could  
actually make sense ;-). It may be easier to elect a disliked person  
in the primary of the other party than to try to figure out which one  
of the candidates of one's own party would be acceptable also to the  
other party members in the final election. Of course the end result  
might be an election between two highly incompetent and unwanted  
candidates (low utility). One way to fight against this would be to  
arrange one's own primaries in a way that is not too vulnerable to  
this type of strategic attacks.

In general methods that are designed to elect the candidate that the  
voters want probably work well also in the reverse direction (=bad  
candidates will be elected if the voters so want). It is impossible  
to separate malicious votes from the regular ones. Maybe the idea to  
limit participation works best (if widespread strategic voting is  
expected).

Good methods like Condorcet may be easy also for the strategists in  
the sense that they don't need any strategic plan on how to change  
the outcome. It is sufficient to rank the candidates in the order of  
unlikeliness of being elected in the final election. In plurality the  
strategic votes could be wasted e.g. by voting for some bad candidate  
that is so bad that he will not be elected anyway (or by picking a  
candidate that is not the worst one that could have been picked). Of  
course there is no need for primaries that would elect one candidate  
only if the method is good enough to handle multiple candidates per  
party. Influencing primaries that elect multiple candidates for the  
final election seems to be less efficient as influencing election of  
one single candidate for the final election.

One more simple approach is to collect poll data about the popularity  
of the candidates of one's own party and make the results well known  
to the voters of the closed primary. This would encourage electing  
candidates that get wide support also from other parties. Also any  
registrations that force the voters to somehow "show colour" publicly  
could limit the number of voters that do not want to show this  
particular colour.

One general observation. In theory primaries can make the outcome of  
the election worse than it would be without them since the election  
criteria in the primaries are different from the actual election.  
Primaries may well eliminate candidate that would have won the final  
election. This is maybe one additional reason for supporting methods  
that can handle numerous candidates. (Having too many candidates may  
however also make the election worse since then it is more confusing  
and based on less good analysis of the candidates by the voters.  
Allowing e.g. two candidates from the major parties and one from some  
smaller ones (and maybe few independents too) could be one way to go.)

Juho


On Jan 4, 2008, at 6:33 , Dave Ketchum wrote:

> On Wed, 02 Jan 2008 04:42:30 -0800 Steve Eppley wrote privately.   
> If you
> reply,
>> please address it to the maillist.]
>>
>> You wrote:
>>
>>> As to "closed" primary elections:
>>>      They make sense to me for Plurality general elections.
>>
>>
>> Why do you think so?
>
> In a closed primary election party members properly get to choose who
> shall be the party's candidate in the general election:
>        For a major party it matters little, for it is hard to round up
> enough outsiders to outvote party members.
>        For a minor party there are lots of outsiders, who could easily
> outvote party members.
>        Assuming there are enough interested outsiders, they should  
> be able
> to nominate the candidate they wish to elect, without interfering  
> with a
> party's nomination task.
>
> I did specify Plurality.  With some other methods, such as  
> Condorcet, more
> candidates can be allowed in the general election without destructive
> interference.
>>
>> Regards,
>> Steve Eppley
> -- 
>   davek at clarityconnect.com    people.clarityconnect.com/webpages3/ 
> davek
>   Dave Ketchum   108 Halstead Ave, Owego, NY  13827-1708    
> 607-687-5026
>             Do to no one what you would not want done to you.
>                   If you want peace, work for justice.
>
>
>
> ----
> Election-Methods mailing list - see http://electorama.com/em for  
> list info


		
___________________________________________________________ 
Copy addresses and emails from any email account to Yahoo! Mail - quick, easy and free. http://uk.docs.yahoo.com/trueswitch2.html