The 2004 CCFR survey of the public is based on an internet-derived sample, rather than the dual mode (telephone/in-person) survey employed in 2002 or the personal face-to-face interviews used between 1974 and 1998. Tests in 2002 using the various foreign policy goals questions found that their "very important" responses on
separate internet and combined telephone/in-person samples
differed by an average of five percentage points. Thus, assertions of recent opinion change in this analysis are generally confined to changes of ten points or more in order to account for potential sample "mode effect."
