Polls are always weighted VK. The adjustment would be found in the margin of error--in this case the higher sample in BC. The normal factor is Ontario--38.3%, Quebec---23.8%, BC 12.9% Alberta 11.1% and so on. Ideally you want the number of respondents to reflect this proportion--but as an alternative one can take the higher sample--in this case BC and use it as a sort of base line for the other percentages derived. Even in this ideal----the specific margin of error per province in isolation would be somewhat high---400 respondents in Ontario is still around 5% margin of error. The calculation method is generic. So--even in this poll the overall numbers--I certainly couldn't use BC---@36% for example with Alberta @60%---Quebec etc. (for Conservatives). Obviously I could not extrapolite BC over Canada for Conservatives or Liberals or NDP---imagine what the outcome would have been if I were doing as you suggested! If you had held your breath for a moment VK--you would have gone past stage 1 to realize this.
It never ceases to amaze me what pomposity prevails upon so many 'anonymous' bloggers.
