Here's the deal about these approval polls - they can be manipulated any way you want. A few things I'd want to see in a poll before I gave it too much credibility.
1) How were the questions asked? Injecting bias into the sample is difficult for even experienced pollsters. Was there some sort of oversight committee that viewed the questions and determined them to be free of bias? Extremely unlikely.
As an example - let's assume they asked a question about Obama and the economy and it was worded:
Given the sad state of the economy when Obama was elected, how do you think he's doing with the economy?
That's a tainted question. You've put in the respondents mind that the economy was horrible when he started.
Also, sometimes neutral answers - since they're not negative get reported as positive responses.
2) Sampling errors and modeling. How many people were polled and what were the demographics? Just because you poll a thousand people it doesn't mean anything when it comes to the validity of a sample. You have to ask in the same proportion as the demographic make-up of the country. Little mistakes in a sample of 1000 people that are supposed to represent 300 million are huge. You get into sampling at that point which is a whole 'nother tangent of potential inaccuracies.
Lastly, Democrats hisorically are more likely to respond to any kind of poll than Republicans. My own personal theory is that they are at work and don't have time for this BS.
I remember when Bush suppossedly had the worst approval rating ever. For that to have been true there would have had to been a standard poll given to every US President since Washington and the Dems and Repubs would have had to agree on the verbiage. I believe it was CBS who ran that poll. Dan Rather essentially lost his job over Bush - why was there no credibility issue with them being involved in that sample?
So relax - this is nothing but propaganda most suitably used to wipe your ass. They can't continue to ignore the growing discontent.