Are Oversampled Political Polls Designed To Sway Minds?

I am not a pollster, but if I conduct a political survey composed of 318 Democrats, 236 Republicans and 78 Independents, would I be surprised to know that the Democrat comes out on top by a big number? D 50% vs R 39%  is the published result, but the oversampling of one party over another is 14%  (The numbers above are taken from a Franklin and Marshall Poll, cited in the Pittsburgh Tribune Review, 9/26/12, by Salena Zeto.)

After months of polls where the presidential candidates are locked in an even contest, suddenly – on the same day – in the pivotal states of Ohio, Florida and Pennsylvania, polls are released by Quinnipiac University showing the democrat is ahead by 8-12 points, using samples that look like the above.

Quinnipiac's demographic samples include Republican, Democrats and Independents from Ohio, Florida and Pennsylvania and they ask the respondents 48 questions -- 48, each with five choices!  Think for a moment about the time it would take out a respondent's day to participate.  If each question consumed 30 seconds and the set-up and wind-down were considered, this is probably a time commitment of 30 minutes, listening to an intentionally dry voice drone on for 48 questions with five choices per.  The design of such a survey may be good on paper, but does not appear to consider that a human is required to participate.

Back to numbers, Quinnipiac numbers.  Ignoring the impact of self-described Independents sampled, are the survey results predetermined simply by oversampling of one party? Would it have been a dead-heat if there was no predominance of one group over another?

Florida Political Survey:
Republicans surveyed     344
Democrats surveyed       417
Result:             Democrat preferred 53% - 44%, a 9% difference
Oversample of Democrats:    9%

Pennsylvania Survey:
Republicans    344
Democrats      457
Result:             Democrat Preferred 54% - 42, a 12% difference
Oversample of Democrats:    11%

Ohio Survey:
Republicans   326
Democrats     387
Result:            Democrat Preferred  53% - 43, a 10% difference
Oversample of Democrats:   9%

Judge for yourself: http://www.quinnipiac.edu/institutes-and-centers/polling-institute/presidential-swing-states-%28fl-oh-and-pa%29/release-detail?ReleaseID=1800