There are certain places where people don't vote on the issues, they don't vote on the candidates, they aren't influenced by the campaign - they vote for a particular party because they've always voted for that party, and their parents voted for that party and their grandparents voted for that party. It's most commonly true in areas where education levels are lower because those areas seem to correlate with areas where voter participation levels are also lower and if a voter isn't interested in participating, it's easier for them to blindly follow one particular party or another. This statement is as equally true of some solidly Republican voting areas as it is of Democrat voting areas, and I don't know whether the Bronx fits into this classification but if it does and if there are enough "party voters" in the area, it is possible that there is NO Republican who could win it, no matter what their opinion was. They could be the most left-wing Republican in the history of the US against the most right-wing Democrat and if enough people will automatically vote for the Democrat the Republican would still lose.
Most people are like this. Any presidential election, GOP and Dems are completely guaranteed 40% of the vote each. That leaves only 20% of the electorate that is in any way swayable. And at least half of these folks are only swayable in an unusually large (1964, 1972, 1984...) landslide.
Precisely - and there are some areas which have well over 80% partisan voters (along with some areas which have less than that). Of course, some areas can be won by the party with the minority, sometimes based on a particular candidate with a high personal vote but more frequently it's because the dominant side didn't turn out to vote for the candidate. I haven't looked at the numbers, but I'm guessing turnout was quite low in that Idaho district that turned in this election (as opposed to a large number of swinging voters).
On a slight side note, in Australia I've seen booth results where voters have swung 15-20% between two consecutive elections, and I've seen booths in which the federal and state results can vary by 20% or more.
Compare:
http://www.vec.vic.gov.au/state1999TCPbyVCKnoxDistrict.html and
http://www.vec.vic.gov.au/state2002TCPbyVCFerntreeGullyDistrict.htmlThis is the same electorate, but a redistribution changed some of the boundaries and the name of the electorate. Observe the Heany Park booth. The population growth at that booth was not due to the redistribution, it was due to population growth of a new part of the suburb. In 1999, the election result was about 64.37% two party preferred to the Liberals. In 2002, that swung to just 45.22% two party preferred, or a swing of 19.15%. Federally, the same booth in 2004 returned a two party preferred result of 63.73% - so the change between 1999 and 2002 was a genuine swing as opposed to changing demographics due to population growth. Here are the 2004 federal election results:
http://results.aec.gov.au/12246/results/HousePollingPlaceFirstPrefs-12246-31743.htmI found a few other examples, but that one should suffice for now.