I believe new numbers will be out at the end of the month for partisan ID. Hopefully will trend more GOP or stay even. If trend more Dem, will be a bit concerned but we’ll see. Other data out looks very good.
1) according to the chart kindly cited by Stonewall Jackson, Rasmussen is talking about, "Summary of Party Affiliation" and this chart tells us that as at the last month, August, Rasmussen has party affiliation for Republicans as at August 31, 2012 at 4.3%.
Interestingly, Rasmussen has party affiliation as at August 2010 at 1.2% and as at November, 2010, at 1.3%.
[Obviously if party affiliation is now running at three times the advantage we had in 2010, we would expect to perform even better than in 2010, all things being equal. But of course, all things are never equal for example this is a presidential election year and that was a midterm cycle. Nevertheless, are we not entitled to say that turnout for Republicans will be robust because if people are polling as "affiliated" with Republicans they are more likely to come out and vote for Republicans?]
If we attempt to compare election cycle apples to election cycle apples we see that in August 2008 Rasmussen has party affiliation polling at 5.7% for Republicans and 7.6% in November on its way to 8.8% in December.
[So we do have a flip of over 10% in Republicans favor since 2008; Are we not entitled to simply strip away from any polls those points which do not reflect Republicans up 7.6%?]
2) Dick Morris in a column which appeared on these threads yesterday said that Rasmussen was the most accurate because he takes a rolling average of party affiliation in weighting Republicans vs. Democrats vs. independents among those whom he questions. Frustratingly, Morris does not tell us what these data are or have been. Nor does he explicitly state that Rasmussen is actually polling according to these numbers. Implicitly, Morris seems to except Rasmussen's numbers which show Romney either tied or slightly behind in the critical swing states.
3) I am not sure when we attempt to adjust these polls from various polling organizations if we are mixing party "affiliation" with party "identification" (whatever that difference might be) and party "registration" (which presumably means actual official data concerning people who have publicly identified themselves for the government).
4) finally we have Rasmussen providing polling data on congressional preference which the last I saw showed Republicans up 1% but they had recently lost a rather favorable lead.
The upshot of all of this is that very difficult to know what we are dealing with.