In both polls, if you pool the two questions saying they intend to get one, and ignore the don't knows you get:
PS3: 22+10=32 saying they intend to buy one, 14 saying they will never buy one, or 70% +/- 15%.
For XBox 360, 4+2=6 saying intend to buy one, against 14, or 30% +/- 20%.
The error margins are very large because of the small number of people who answered the poll.
Using a different formula, which calculates if the % of people wanting PS3's is higher than the % wanting XBox 360's, then I get 99.7% confidences. I can be quite sure of this, even though the number of people is small because the difference between the responses is so large.
I can be very sure statistically speaking that more people want PS3's then XBox 360's, assuming that the questions were well understood, and that a random sample of people took the poll.