[EM] STV ballot question

[Ed Still]

Let's say we are trying to use existing optical scanner voting machines for 
STV (Preference Voting).  This machine will not recognize characters like 
"1", "A", etc., but senses marks inside small circles or bubbles on the 
ballot.  

What we plan to do is list the candidates down the left side of the ballot 
and have columns for marking candidate-choices:

          --------choices------------
candidate 1st 2nd 3rd 4th 5th 6th 7th
Anderson   ()  ()  ()  ()  ()  ()  ()
Bush       ()  ()  ()  ()  ()  ()  ()
Clinton    ()  ()  ()  ()  ()  ()  () 
Dukakis    ()  ()  ()  ()  ()  ()  ()
Eisenhower ()  ()  ()  ()  ()  ()  () 
Forbes     ()  ()  ()  ()  ()  ()  () 

Ideally, we would like to have the same number of columns as we have 
candidates, but there is a finite width and length to the ballot, so we may 
have to truncate the number of columns.

When I met George Hallett in 1974, he told me that he had once voted in the 
New York City council election and something like his 21st choice helped 
elect a candidate.  His first choice never made the quota, but he remained an 
active candidate until Hallett's choices 2-20 had either been elected or 
eliminated.  So I realize there are some practical consequences to limiting 
the number of columns.

Since (as someone said recently on the elections-reform list) we need to be 
the guys in the white lab coats, I need to be able to tell folks the real 
world consequences of the limitation of the choices to less than the number 
of candidates.

Are there any empirical studies or theoretical ones that I can cite?

Ed Still