Creation of the Datafile

All polls in the data file used a telephone interview methodology and national random digit dial (RDD) samples. The minimum sample size for all polls was 1,000 adults, ages eighteen years and older. A spreadsheet was created that included the party ID distribution as well as the following information for each poll:

•  Sample size and description

•  Dates of interviewing

•  Year and quarter of interviewing

•  Survey topic and title

Yearly and quarterly averages were calculated for party ID using the same methods employed for the authors' 2002 AAPOR paper, "Understanding Short-Term Variation in Party Identification."

The party ID distributions are based on the results of a question with the following basic wording: "In politics TODAY, do you consider yourself a Republican, Democrat, or Independent?" [Newsweek polls conducted between October 2002 and December 2004 preceded the basic party ID question with the following additional wording—"Regardless of how you might vote in November" (October 2002)/"Regardless of how you might have voted in previous elections (November 2002-December 2004).] Volunteered responses were recorded for "other party" and "no party."

In most polls, the party ID question was positioned at the beginning of the demographics section at the end of the interview. For practical reasons, the question was asked near the beginning of the interview during the primary campaign period, to identify registered Democrats. Nearly all Newsweek polls in the datafile were conducted during a two-day interview period, Thursday through Friday. Fieldwork was done by Braun Research, Inc., of Princeton, New Jersey.

Identical weighting procedures were used for all polls. Results were weighted to current U.S. Census parameters for gender, age, education, race/ethnicity, and geographic region. No weighting by party ID or any other political variable was done.