Originally posted by Not a Number If a 2% difference is not statistically significant or within the margin of error then yes 82% vs 84% can be considered a tie.
The size of the sample can't be six thousand voters, it'll be one or more of the DPR staff, right?
So it's like a panel of judges in a singing contest, or the jury in a court case - there might be variation in opinion polls before the event, but there shouldn't be a discrepancy in the actual tally on the day!
This decision to redefine what a 'tie' is diminishes the winner's achievement.
Why not more conventionally say, "The Pentax is the best camera, with an honourable mention to the Sony"?