Below the orange thingy are the list of Rasmussen results from 2010 Senate races that I have deemed competitive looking only at Rasmussens numbers. Before looking at the data, I arbitrarily selected a 5% margin as what would define a close race. If I have the time, I'll go back and try this for the house and 2008. I do prefer the idea of using the 2010 race as a basis. It was considered a strong R year, one where R leaning pollsters should have outperformed.
The data is shown below. The first listing is the Rasmussen poll difference between the frontrunner and the top competitor. The second listing is the actual result. Third is whether they were correct or incorrect. Finally, the last number shows how far off they were and which side their bias benefited.
Alaska
Final Ras Miller (R) +1 Results Murkowski (un) +5 Incorrect +6 R bias
California
Final Ras Boxer (D)+3 Results Boxer (D) +10 Correct +7 R bias
Colorado
Final Ras Ken Buck (R) +4 Results Bennet (D) +1 Incorrect +5 R bias
Illonois
Final Ras Kirk (R) +4 Results Hirk (R) +2 Correct +2 R bias
Nevada
Final Ras Angle (R) +4 Results Reid (D) +5 Incorrect +9 R bias
Pennsylvania
Final Ras Toomey (R) +4 Results Toomey (R) +2 Correct +2 R bias
Washington
Final Ras Rossi (R) +1 Results Murray (D) +4 Incorrect +5 R bias
West Virgina
Final Ras Manchin (D) +4 Results Manchin (D) +10 Correct +6 R bias
Scorecard. On average Rasmussen overestimated the Republican candidates numbers by 5.125. As far as whether he picked the winner, he was 50-50. Heads or Tails?
http://www.rasmussenreports.com/...