This report contains data from several different sources. The data on overall Twitter usage and demographics comes from the Pew Internet Project’s November 2010 tracking survey, while the data on frequency of use and types of material posted by Twitter users comes from two Omnibus Surveys conducted in October 2010.
The two types of surveys, tracking and omnibus, collect data from nationally representative dual-frame (landline and cell phone) samples, employ the same respondent selection process, and identify internet users using identical questions. They are conducted by the same survey research firm, Princeton Survey Research Associates International, at the same field house. However, there are differences between the two types of surveys that should be noted when comparing data across them. First, tracking surveys consist of roughly 2,250 interviews completed over the course of three to four weeks. These surveys maintain a very close 2-to-5 ratio of weekend-to-weekday interviews, to minimize the impact of day-of-the-week effects. Omnibus surveys, in contrast, consist of roughly 1,000 interviews completed over the course of four days, usually a Thursday-to-Sunday timeframe. There is no specific control in omnibus surveys for weekend-to-weekday interview ratio. To the extent that day of the week impacts technology use and online behavior, this may introduce variance in the data across the two types of surveys.
Moreover, tracking surveys follow a 7-call design in which sample that has not reached a final disposition at the end of seven days is retired, unless there is an outstanding appointment or callback for that telephone number. The omnibus surveys use a 4-call design over the course of the 4-day field period. One result of these different approaches is that tracking surveys generally achieve higher response rates than omnibus surveys. Again, this difference could introduce variance in the data across the two types of surveys.