I misread paragraph second from bottom.
One thing I’ve noticed on RCP is all the national polls with the smallest sample sizes have Obama up...
Noticed this too. While getting a larger sample is important, it is not going to bias a poll one way or another on average...it just gets you a smaller margin of error...but the tricky thing about polls and why they tell different stories outside of their margin of error (like this one compared to Gallup) as often as they do...is that they have to choose among several imperfect ways to attempt to get a random sample of the population of people actually going to vote with an unbiased chance of each real voter being represented in that sample. This is a very difficult problem...and the only real way to check the success of this selection is to do sanity checking of your results...such as comparing the break down of the party id to larger to voter registration patterns and such. This poll is dubious to me not because of the small sample...but because, like I said, it has lots of big talk about conclusions but does not seem to provide me with any internal data which I could use to check its veracity.