# [EM] Reverse NO columns - Required Number Votes

Juho Laatu juho4880 at yahoo.co.uk
Thu May 22 02:13:23 PDT 2014

```On 22 May 2014, at 06:04, DNOW1 at aol.com wrote:

> Reverse NO columns
>
> IF Number Votes are to be deemed YES in a table, then there should be a reverse columns NO table -- esp. for checking math purposes.
>
> Example
> YES [left to right]
> 26 AB
> 25 BA
> 49 Z
> 100
>
> NO [right to left]
> 26 Z
> 25 Z
> 49 A+B
> 100
>
> Z should obviously lose.
>
> Should A then beat B ???

Yes.

> ---- 26 to 25 (assuming none of the Z voters vote for the lesser evils of A or B -- noting the 51 A/B majority group.
> ------------
> Which brings up --
> Should voters be required to put a Number Vote on ALL choices - both YES and NO ???

Allowing truncation of the vote may be useful if there are e.g. 100 candidates, and it would be too tedious to rank them all.

Incomplete rankings can be accepted also to avoid the problem of rejecting too many ballots because of some minor errors thatvoters are likely to make in the rather complex ranked ballots. In this case those small errors could lead e.g. to just ignoring markings that are not clear enough.

But if the election has only few candidates, no major risk of errors, and no risk of losing too many voters due to too complex voting, then requiring full ranking could be possible.

>
> Thus YES votes left of *, NO votes right of *
> 26 AB*Z
> 25 BA*Z
> 30 Z*BA
> 19 Z*AB
> 100
>
> Thus - does 26 A YES (+19 NO) beat 25 B YES (+30 NO) ???

You can use the approvals in many ways. In typical competitive political elections my preference would be to put more weight on the rankings and less weight on the approvals. Approvals could be used e.g. to resolve ties or to set some acceptability limits to the winner. (Putting more weight on approvals may easily lead to problems with strategic voting.) Many elections work also with rankings only, without measuring approvals at all. If one looks at the rankings only in the ballot set above, B seems to be a good winner since it would beat both A and Z in pairwise comparisons (i.e. B is a Condorcet winner, and electing the Condorcet winner is a criterion that most people on this list want to respect).

It seems that both A and B have 49 no votes, so that would not influence the "rankings only" results that I discussed above.

>
> Noting again that some of the Z minority voters may vote YES on A or B.
> ----
> Which brings up -- whether a NO minority should be able to determine the lesser of its perceived evils in the YES majority ???

It makes sense to allow the "opposite side" to rank the candidates of the "AB side" since typically we want to elect compromise candidates that are acceptable to all voters. In the example above it is possible that A is an extremist that very far from Z voters, and B is much closer to Z voters. In this situation it makes sense to allow the Z voters to influence on whether A or B wins.

>
> i.e. Should B 55 (YES+NO) beat A 45 (YES+NO) ???----

This means that B beats A 55-45 in pairwise comparison. If the method puts lots of weight on rankings, the B should win.

In summary, thee is no general rule on how rankings and approvals should be combined. Often rankings can do the job alone, but sometimes approvals can add something useful to the equation. For most needs I see rankings as primary information and approvals as secondary information.

Juho

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20140522/c7638824/attachment-0002.htm>
```