Which Polls Are Most Accurate?
Many of our thoughts on the upcoming races are based on polling numbers. With this in mind I went back and looked at the last several contests to compare polling numbers to the final results. I used reliable polling information from contests conducted "post" Super Tuesday. This is because I assumed polling information would be more accurate once certain demographic and voter turnout patterns were established.
The polling numbers are from popular pollsters such as Rasmussen, Survey USA, etc. In each case the numbers represent the last polls conducted before voting. The polling numbers are from pollister.com. The actual result percentages are from the New York Times.
Wisconsin/Clinton/Obama
ARG 42/52......................................... Obama +10
PPP 40/53......................................... Obama +13
Rasmussen 43/47.............................. Obama +4
Result 41/58.............................Obama +17
---------------------------------------------------------------------------
Virginia/Clinton/Obama
Survey USA 38/60............................ Obama + 22
ARG 38/56............................. Obama + 18
Rasmuseen 37/55............................. Obama + 18
Insider Adv. 37/52............................. Obama + 15
Result 35/64.................................... Obama +29
---------------------------------------------------------------
Vermont/Clinton/Obama
Rasmussen 33/57........................... Obama + 24
ARG 34/60...................................... Obama + 26
Result 39/59................................... Obama + 20
---------------------------------------------------------------------------
Texas/Clinton/Obama
ARG 50/47.................................... Clinton + 3
Zogby 47/44................................... Clinton +3
Rasmussen 47/48............................... Obama +1
Insider Adv. 49/44............................ Clinton +5
PPP 50/44.......................................Clinton +6
Survey USA 48/49................................Obama +1
Result 51/47...........................Clinton +4
---------------------------------------------------------------------------
Rhode Island/Clinton/Obama
Rasmussen 53/38......................... Clinton +15
ARG 52/40................................... Clinton +12
Results 58/40................................Clinton +18
---------------------------------------------------------------------------
Ohio/Clinton/Obama
ARG 56/42.................................. Clinton +14
Zogby 44/44................................. Tie
Rasmussen 50/44.......................... Clinton +6
Survey USA 54/44..........................Clinton +10
PPP 51/42....................................Clinton + 9
Quinnipiac 49/44.............................Clinton + 5
Results 54/44..................................Clinton +10
---------------------------------------------------------------------------
Mississippi/Clinton/Obama
ARG 38/54.................................... Obama +16
Insider Adv. 37/54........................... Obama +17
Rasmussen 39/53.............................. Obama + 14
Results 37/61.................................. Obama +24
---------------------------------------------------------------------------
Maryland/Clinton/Obama
Survey USA 32/55................................ Obama+23
ARG 37/55....................................... Obama+ 18
Rasmussen 31/57................................. Obama+26
Results 37/60........................................ Obama +23
----------------------------------------------------------------------
Quick Observations....
SurveyUSA was exactly right with thier polls in Maryland and Ohio. However they were wrong in Texas where they had Obama winning. (albeit by 1)
When Public Policy Polling (PPP) came out with their poll showing Obama leading in Pennsylvania, it was widely dismissed. However looking at their recent track record, PPP was the most accurate pollster for the Wisconsin primary. And they were the best pollster overall for "Super Tuesday II." Their polling is 100% legitimate. This means Obama is very close in Pennsylvania.
Finally, there is now more than enough evidence to put to rest "the Bradley effect" when it comes to Obama and public polls. This became a popular issue following Clinton's win in New Hampshire.
However, if you look at the numbers objectively, Obama has actually outperformed the polls in several contests, including Wisconsin which is overwhelmingly white.
Supporters of the "Bradley effect" also frequently leave out polls which had Clinton winning in Missouri and Alabama even though she lost both. (She was actually "blown out" in Alabama, and the "always reliable" Survey USA had her leading 54/43 in Missouri when she lost that primary.)
One more note, Rasmussen seems to poll more frequently than an other major pollster.
This is my first entry here, so I apoligize if the topic or format is "off-base."