TRENDS IN NON-RESPONSE RATES

Within survey research it is widely accepted that response rates have been and are continuing to decline both in the United States (Bradburn 1992, Brehm 1993, Brehm 1994, Fan 1994, Groves 1989, Groves and Lyberg Ig88, Remington 1992, Schillmoeller 1988, Schorr 1992, Singer and Martin 1994, Steeb 1981) and elsewhere (Bairn 1991, Davis eta/. 1994, Groves 1989, Meier 1991, Yamada and Synodinos 1994). The drop in the initial mail return rate for the U.S. Census from 75 percent in 1980 to 63-65 percent in 1990 is seen as emblematic of this problem (Fay eta/. 1991 and Kulka et a/. 1991). A corollary holds that increases in refusal rates have been the main cause of the falling response rates and that only decreases in the non-contact and other non-response rates have prevented an even greater decline in response rates. Moreover, most researchers think that response rates will continue to deteriorate in the future. In a survey of American survey research organizations in 1993/94 by NORC, refusals and the related problem of call screening and answering machines were considered the most important problems over the next 10 years (both mentioned as a major challenge by 43 percent of firms). These response problems were cited much more frequently than the nine other concerns (high costs 24 percent, unethical studies 20 percent, biased s~udies 16.5 percent, lack of trained staff 10 percent, data privacy 7 percent, incorporating new technology 6 percent, lack of public interest in issues 6 percent, and consistency of new technology 4·5 percent-Smith 1994). Similarly,