How the AP-GfK Poll on the midterm elections was conducted

The Associated Press-GfK poll on the midterm elections was conducted by GfK Public Affairs and Corporate Communications September 25-29. It is based on online interviews of 1,845 adults who are members of GfK's nationally representative KnowledgePanel.

The original sample was drawn from a panel of respondents recruited via phone or mail survey methods. GfK provides Internet access to panel recruits who don't already have it. With a probability basis and coverage of people who otherwise couldn't access the Internet, online surveys using KnowledgePanel are nationally representative.

The survey included interviews with 958 likely voters. A respondent was determined to be a likely voter based on responses to a series of questions on voting intention, voting habits and interest in the upcoming election.

Interviews were conducted in both English and Spanish.

As is done routinely in surveys, results were weighted, or adjusted, to ensure that responses accurately reflect the population's makeup by factors such as age, sex, race, education and phone usage.

No more than 1 time in 20 should chance variations in the sample cause the results to vary by more than plus or minus 2.5 percentage points from the answers that would be obtained if all adults in the U.S. were polled. For results among likely voters, it is 3.5 percentage points.

There are other sources of potential error in polls, including the wording and order of questions.

The poll had a cooperation rate of 47 percent.

The questions and results are available at http://www.ap-gfkpoll.com