Polls for congressional races combine traditional polling techniques with generic balloting to predict overall partisan distribution of nationwide congressional races. The typical generic ballot question asks voters whether they would vote for the Democrat or Republican candidate if the congressional election was held that day.
The generic question asking voters whether they would vote for a Democrat or Republican has limited value in predicting individual election results but is a reliable forecaster of overall national outcomes. Polling organizations believe this method accurately depicts the national political environment and find it useful for classifying political preferences according to demographic factors to develop voter profiles and for identifying emerging trends.
Polling organizations obtain surprisingly accurate results by combining generic ballot questions with traditional polling methods such as identifying likely voters, obtaining representative samples and adjusting data to reflect actual voting populations, particularly in mid-term election years. Since 1950, predicted mid-term election results deviate from actual outcomes by an average of 1.1 percent using this method, according to Gallup and Pew Researchers. Rasmussen Reports claims a plus-or-minus 2 percent margin of error and a 95 percent confidence rating in its data using generic balloting.
Despite the accuracy associated with generic balloting during mid-term election years, deviation rates tend to increase to 2 to 5 percent in presidential election years. Researchers are not sure why this occurs, but some attribute this to the tendency of popular sentiment toward presidential candidates leading voters to inadequately consider congressional races.