• Sample size and composition: 57,967 randomly selected US adults
• Interviewing dates: Daily, December 17, 2007 – November 3, 2008

Respondents were sampled for the National Rolling Cross-Section using the following random-digit dialing and respondent selection procedures:

Random telephone number generation: A list of 10-digit US telephone numbers was randomly generated, inclusive of the 48 continental states and Washington, DC:

• The first eight digits of phone numbers were randomly generated proportional to estimates of the percentage of household phones beginning with those digits.
• The last two digits were generated entirely at random.
• Known cell phone numbers were excluded.

Household identification: Interviewers called the random phone list to identify numbers that were households. When interviewers determined that a number was not a household or was out of service, the number was removed from the study.

Selection of respondents within households: Upon reaching a household, interviewers asked to speak with an adult age 18 or older, and then asked this person how many adults lived in the household. One adult per household was selected as a survey respondent:

• If there was one adult in the household, he or she was selected as the survey respondent.
• If there were two adults, interviewers asked at random for either the younger or older adult.
• If there were three adults, interviewers asked at random for the youngest, middle, or oldest adult.
• If there were four adults, interviewers asked randomly for either the adult who most recently celebrated a birthday or the adult with the next birthday.
• If there were no adults age 18 or older, the household was ineligible and removed from the study.

Interviewers called sampled phone numbers multiple times to identify households, select respondents, and complete interviews. The following phone procedures were used:

Calling schedule: Randomly generated phone numbers were added to the study in daily batches. Each phone number was called up to a maximum of 18 total calls over 14 days from the date of addition to the study:

• If a number resulted in no answer, a busy signal, or the call being forwarded to an answering machine, the number was scheduled to be called again no later than the next day.
• If households asked to be called back, interviewers accommodated requests to call at specific times, or called back within the next few days if no time was specified.
• Most calls were made on weekday evenings and weekends. When possible, at least one attempt was made to each phone number during a weekday morning or afternoon.
• After a number was called 18 times over 14 days, no additional calls were made to that number.

Refusals and breakoffs: If a household refused to participate in the survey, or started and broke off an interview, interviewers called back to try to elicit participation:

• If a household refused during the first week of calling to the number, generally the callback was made a few days later.
• If a household refused during the second week of calling to the number, generally the callback was made the next day.
• If a respondent started and broke off an interview, he or she was called back to restart and complete the interview. Generally, callbacks to complete partial interviews were initiated within one day of the breakoff.

Interview language: Interviews were conducted in English or Spanish at the respondent's preference. Variable WHa01 indicates the interview language.

The number of National Rolling Cross-Section interviews completed per day varied from 40 to 313, depending on the campaign period. The highest interviewing volumes were allocated for the primary and general election periods:

Holidays / days without interviews: National RCS interviews were conducted daily, with the following exceptions:

• December 24 and 25, 2007, Christmas holiday
• December 31, 2007, and January 1, 2008, New Year's holiday
• March 23, 2008, Easter
• July 4, 2008, Independence Day
• October 7, 2008 — due to a field house technical problem, no interviews were completed this day

Call outcomes: The following table shows the total telephone numbers called for the National Rolling Cross-Section, and the final outcome of calling each:

Response rate calculation: Response rate is the ratio of completed interviews to total households called:

• A conservative approach is to assume that the total number of households includes both those of known eligibility (A, table above) and those of unknown status (D). Using this method, the response rate for the National Rolling Cross-Section is 19% [A1 / (A + D)].
• A less conservative approach is to include only households of known eligibility. Using this method, the response rate is 23% (A1 / A).

• Sample size and composition: 3,737 US adults initially interviewed for the National Rolling Cross-Section during the 13 weeks before the general election
• Interviewing dates: November 5 – 10, 2008

The Post-Election Reinterview consists of second interviews with respondents initially interviewed for the National Rolling Cross-Section during the period August 8 through November 3, 2008.

A total of 21,698 respondents were interviewed for the National RCS during this period, of which 78% consented at the end of the interview to being recontacted for a second interview.

Of these 78% of National RCS respondents who consented to a second interview, 6,990 were randomly selected for the Post-Election Reinterview.

Calling schedule: The goal was to complete as many reinterviews as possible in the days immediately following the election. Interviewers called sampled respondents daily to re-establish contact and complete reinterviews.

Respondent matching: At the end of the National RCS interview, respondents indicated their first names. At the start of the reinterview, interviewers confirmed they were speaking to the same respondent by asking for him or her by name, and confirming that gender and age (within one year) matched with the data collected for the National RCS interview.

Interview language: Interviews were conducted in English or Spanish at the respondent's preference. Variable WHa01 indicates the interview language.

Reinterviews were completed with 3,737 of 6,990 respondents called for the Post-Election Reinterview, a 53% recontact rate. The overall response rate is the product of the recontact rate and the percentage of National RCS respondents who consented to being recontacted (78%), which is 41%.