Latest Presidential Tracking Polls 2008

Bookmark Us!

General Election Matchups *NEW *

National Polls

House/Senate Polls

Gubernatorial Polls

Democratic Exit Polls

2004 Primary Polls

Political Odds

Invest in Foreign Currencies

General Election Battlegrounds

FL	PA	OH
NV	MI	NH
CO	OR	NJ
WA	IA	WI
MN	TN	OR
VA	MO	NM

Pollster Grades

American Research Group

Rasmussen Reports

Insider Advantage

Mason Dixon

Rasmussen Reports Grade - Solid Pollster But Incorrect on Obama's Magnitude of Victories

2/28/08

State	Date	Poll (HC, BO)	Primary Date	Results (HC, BO)	Offset (HC, BO)	Gross Error
Wisconsin	2/13	(43, 47)	2/19	(41, 58)	-2, +11	13
Virginia	2/6-7	(37, 55)	2/12	(35, 64)	-2, +9	11
Tennessee	1/30	(49, 35)	2/5	(54, 41)	+5, +6	1
South Carolina	1/21	(28, 43)	1/26	(27, 55)	-1, +12	13
New Hampshire	1/5-6	(28, 38)	1/8	(39, 37)	+11, -1	12
Maryland	2/6	(31, 57)	2/12	(37, 60)	+6, +3	3
Connecticut	1/27	(40, 40)	2/5	(47, 51)	+7, +11	4
New York	1/31-2/1	(52, 34)	2/5	(57, 40)	+5, +6	1
New Jersey	1/30	(49, 37)	2/5	(54, 44)	+5, +7	3
Georgia	1/22	(35, 41)	2/5	(31, 67)	-4, +26	30
Florida	1/27	(47, 25)	1/29	(50, 33, 14)	+3, +8	5
Arizona	1/31	(45, 39)	2/5	(51, 42)	+6, +3	3
Alabama	1/31	(46, 41)	2/5	(42, 56)	-4, +15	19
California	1/29	(43, 40)	2/5	(52, 43)	+9, +3	6
AVERAGE						8.9%

The problem with dissecting these results is that Rasmussen Reports tends to release its latest polls about a week before the actual primary. For example the latest Wisconsin poll was surveyed on February 13th when the primary was held on February 19th. And the Super Tuesday state polls were surveyed at the end of January. So it is tough to say how accurate of a pollster they are when their results are a week old but we can look at this data to find patterns.

First of all, despite having a gross error of 30% in Georgia, their average error is just a shade under 9% which is better than American Research Group and Mason Dixon -- who have generally had polls taken just before the primaries. So we'll give more credibility to Rasmussen Reports polls when we see them.

Here's another interesting tidbit, when Rasmussen Reports predicts Clinton ahead (even by 1%) they are generally dead on accurate but when they predict that Obama is ahead (even by 1%), they are wrong in the magnitude of Obama's victory. Meaning that when Obama is expected to win a primary according to Rasmussen Reports, he will generally win by a greater margin.

Here are the states that Rasmussen had Clinton ahead: Tennessee, New York, New Jersey, Florida, Arizona, Alabama, and California. The average error in these 7 states is 5.4%. They were correct in 6 of 7 of those states. Obama won Alabama by 14%.

The states that Rasmussen had Obama ahead were: Wisconsin, Virginia, South Carolina, New Hampshire, Maryland, and Georgia. The average error in these 6 states is 13.7% and they were correct in 5 of the 6 states -- with New Hampshire being the outlier.