This morning, two new approval polls, one from Pew and one from the AP, show President Obama's approval rating at 63% and 64% respectively. These are huge numbers for President Obama, and stand in fairly sharp relief from the 55% approval number he received this morning from Rasmussen Reports.
Why are different polls shows such a substantial variance in President Obama's approval rating? The first place to start looking for an answer is Pollster.com, which has cool charts and tools that allow you to measure President Obama's approval rating according to a variety of filters. According to these charts, there are two big differences between the vast majority of polls showing President Obama's approval rating in the low-60s, and the other polls showing President Obama's approval rating in the mid-50's. Specifically, President Obama does better in polls conducted with live telephone interviews, and in polls that survey "all adults," rather than "likely" or "registered" voters:
Likely Voters (Democracy Corps, NPR, Rasmussen, Zogby Internet)
Approve 54.7%--44.6% Disapprove
Registered Voters (Cook/RT, Diageo/Hotline, Fox, Marist, Q-Poll, Winston Group)
Approve 58.3%--31.6% disapprove
All Adults (everyone else)
Approve 61.9%--28.8% disapprove
Approve 51.5%--48.0% disapprove
Automated phone interview (Rasmussen, PPP)
Approve 54.5%--44.% disapprove
Live phone interview (everyone else)
Approve 61.5%--28.7% disapprove
Given the close correspondence between a drop in Obama's approval ratings among, first, likely and registered voters, and, second, among automated phone and internet interviews, it is difficult to tell whether the type of interview or voting habits is the main cause of the difference. This is especially given that Rasmussen and Zogby both use "likely" voter models, but neither use the live phone interview poll method. One solution is to look only at live phone interview polls that sample all adults versus live interview polls that sample registered voters:
Live telephone interview, all adults
Approve 62.1%--28.9% Disapprove
Liver telephone interviews, registered voters
Approve 58.3%--31.6% Disapprove
This shows a clear difference between presidential approval for "all adults" and for "registered voters" using the same interview type. President Obama has a net job rating of +33.2% among all adults, but only +26.7% among "registered voters." While there is not data from a sufficient number of polling organizations to say so definitively, it is highly probable that live phone interview polls among "likely voters" would show a further drop in President Obama's ratings from the "registered voter" level. Thus, it is reasonable to assume that the type of interview is not the cause for the difference in the polls (and thus, there is not "Bradley effect," where people are more likely to tell a machine than a person that they disapprove of President Obama) but rather the perceived voting likelihood of the survey sample. As such, non-voters appear to be inflating President Obama's ratings a bit (unfortunately, non-voters always seem to slightly prefer Democrats), although his approval ratings are still strong compared to recent Presidents.
Taking a look at the 23 approval polls of President Bush Jr. conducted entirely between March 1st, 2001 and April 23rd, 2001, his mean approval at the time was 56.7%--27.5%. That number shrank slightly to 56.4%--28.0% when looking only at the 19 "all adults" polls (Fox News produced three of the four "registered voter" polls during that time). This means the following President Obama's net approval rating among "all adults" is roughly 5% higher than President Bush's at the same point during his Presidency. Among all adults and registered voter polls combined (there were no early "likely voter" approval polls of President Bush), Obama is about 3.7% higher (61.5%--28.6% vs. 56.7%--27.5%, or +32.9% vs. +29.2%) their net approval ratings are virtually identical.
It is also worth noting that there are significantly fewer undecideds in early Obama approval polls (9.9%) than in early Bush approval polls (15.8%), which could indicate an increasing level of partisanship nationwide.
For further comparison, President Clinton's approval was polled on 19 occassions between March 1st, 1993 and April 24th, 1993. The mean of these 19 polls was 55.2%--29.5%, which was a bit lower than Bush Jr.'s.
Earlier President's were only polled by Gallup, and nowhere near as frequently as Obama, Bush and Clinton, making comparisons difficult. Overall, Obama appears to be doing a bit better than his predecessors, which is impressive given the rotten mood in the country. His improved performance is probably closely connected to his higher score in the popular vote, earning 52.87% versus 47.87% for Bush and 43.01% for Clinton. So, on balance, President Obama is starting from a slightly stronger position than both Clinton and Bush Two, both of whom were re-elected to second terms.
Update: New Gallup poll shows Obama at 65% approval. According to Pollster.com, his live-interview poll trend is now slightly up.