>>I heard an interesting Canadian Broadcast article last night - an interview with Prof. of Pyschology Robert St. Clair (St. Clare? Sinclair?) described a double-blind research experiment he performed twice (once w/students, again with random public) contrasting a linear ballot and a butterfly ballot. The latter proved problematic for in both experiments. Specifically, people had problems with the butterfly ballot when indicating a selection other than the first option listed. (I think this was "As It Happens" but I couldn't fine a reference at their web-site.)
>>
>>I was not an issue of how responsible these people were. And when considering machinery, I suspect most here can can readily recognize (when not politisizing) that in a whole system - the instructions, the sample ballots published in the media, the previous method that people may have become accustomed to, the UI, the input device(s), the tabulation hardware - all parts are factors. It is disingenious to charachterize the accuracy issue as 'people vs. machine'.
>
>
>Steven, it would be very intesting if Prof. St. Clair were to rerun his experiment with a large instruction board in front of the subjects saying (in large print) "4. Vote every page."
>
>I wonder how many of the subjects would follow the instructions rather than figuring out that in the case of the butterfly ballot they had to ignore that instruction.
>
>There was also a place in the standard instruction book which said to make the punch to the RIGHT of the name.
>
>This is a case where "RTFM" does not apply.


That's OK. I think there's only one documented case where a user *ever* RTFM. <s>