Precision vs. Accuracy

Electoral-Vote.com has an instructive post today about the difference between "margin of error" and accuracy in a poll, and about the importance of which questions are asked in what order. To a certain extent I think sites like this that predict the outcomes of elections are misleading, since they present what looks like a full prediction, like "48 Democrats, 52 Republicans," when in reality several of those races are too close to call statistically.