i think kellyanne conway is right. would say i disagree and never sayg weighting, but i would with limited parameters. at me clarifyest: the entirewith mountain -- margin of error. if you do a nationwide poll of 1000 voters, the margin of error 3%.lus or minus it means that 19 out of 20 cases, had you asked the same poll, you would get the same 3.1 points in either direction in 19 out of 20 cases. there is that one out of 20 cases where it is not going to be correct, number one. number two, if you have that engineercentage martin -- margin of error, we will wait three in one or the other. wait, butnever avoiding over-waiting. i think some of the firms that are way too busy, have way too many clients at one time are anded to mass-produce polls andhington journal the -- weight them. when you do polling, it is a way to look at the electorate in the case of consumers or corporations. people often ask, how can you tell what a population's opinion is if you only talk to 1000 300 million, and the answer -- -- 1000 out of 300