If we have a rational expectation of tracking polls (and other polls that are repeated over time with the same methodology) they are useful and all pretty consistent at what they measure, which is
change.
Just because polls are consistent doesn't make them right. Ten pollsters could ask people if they cheat on their taxes and get consistent results, but all could be wrong if people are intrinsically less than candid when confessing crimes to random callers on the phone.
But, with that caveat, the apples-to-apples consistency of various Presidential Approval polling sets suggests that a real-world opinion phenomenon is being measured.
The most recent results from some presidential approval series that PollingReport.com collects: (Tracking polls and other polls that are repeated often with the same questions and internal methodology)
Gallup 48
FOX/OD RV 46
CBS 53
Quinnipiac U. RV 48
CNN/ORC 55
ABC/Washington Post 56
AP-GfK 54
Pew 51
Ipsos/McClatchy 53
The first thing we realize is that CNN is in league with ORCs. Hmmm...
You can look at the range from 46% to 56% and conclude (erroneously) that polls are meaningless... if you think the primary purpose of a tracking poll or polling series is to generate a top-line number. (The media, being a sorry lot, focus on top-line numbers but that doesn't mean we have to.)
The fact that tracking polls and polling series are only good for what they do well doesn't make them useless, no more than a rake is useless because it is hard to comb our hair with it. (Or that a comb is useless because it will take forever to collect up leaves with it.)
A couple of things things about the varied approval polls:
We have an expectation from history that this country is politically split. We expect presidential approval to be in the 40-60 range. But if these polls were measuring something really unknown, like whether people prefer the planet Venus or the planet Jupiter, a range of 46%-56% would tell us a lot. Going into it Venus might have been at 10% or 90% so a host of polls saying the distribution of planet favorability is somewhere near 50% is useful information.
Comparing apples and oranges is a problem. Though the different polls show different results compared head to head
they all show the same result when compared apples to apples. All polling series had Obama doing well in June. All polling series have him down substantially since then.
Gallup from the 60% range to the 50% range.
FOX from 62% to 46%
Quinippiac from 57% to 48%
CBS 63% to 53%
CNN 61% to 55%
Is Obama at 56% or 46%? Nobody knows for sure. Somewhere around 51% seems likely.
Is that "all adults" or registered voters? Depends on the poll. What is the party distribution assumed? Depends on the poll.
But what we do know is that whatever the *right* number is, that number is down about 10% from whatever the *right* number was in June.
All tracking polls have biases and assumptions, yet measure apples-to-apples change pretty well. Rasmussen is a republican outfit and he sets up his poll to depress the top-line Obama number. But even he tracks the same
relative change. (Considerably higher in June than November.) FOX has the lowest number of the network polls at 46%, but in June they had a very solid number of 62%. They poll only RV... is that more volatile than "all adults"? Hmmm... would make sense, I guess.
Democrats always do better in CBS polls, for whatever reason. In this case that means a drop from 63% to 53%. Quinippiac is usually more in-the-middle, showing a drop from 57% to 48%.
And so on...
Numbers used in OP:
http://www.pollingreport.com/obama_job.htm