Let's say if Ralph Nader has 2% of the votes of those polled, with a 3% margin of error [assuming 1000 people polled, so 1/(1000^1/2) = 3%].

With a 95% confidence level and a 3% margin of error, should we then conclude that on 95 out of 100 polls, Ralph Nader who has 2% of the votes will receive between 0 to 5% of the votes? (of course, ignoring the negative percentage points since that's not possible.)

Am I on the right path, or is there something more to it?

I have a feeling I'm doing something wrong. Any ideas?