Social surveys continue to dominate research articles published in the social work literature and the research conducted by agency-based practitioners (Fraser, Taylor, Jackson, & O'Jack; 1991; Glisson, 1990, 1995; Marino, Green, & Young, 1998). Consequently, the choice of methods and techniques for the construction and presentation of questionnaire items on research instruments is particularly pertinent to social work knowledge building and to the evaluation of professional practice. Dillman's (1978) well-known but somewhat dated text, Telephone and Mail Surveys, provides the most complete set of practical instructions for designing surveys and survey instruments. Subsequent methodological studies have evaluated the use of alternative methods for maximizing the likelihood that recipients will respond to survey instruments and minimizing the likelihood that selected items on returned questionnaires will be left unanswered (Dillman, 1991). However, these studies have not provided empirical guidelines for one of the most frequent methodological questions asked by graduate students and agency-based practitioners seeking survey research consultation: Should the demographic questions, used primarily to describe the sample, be placed at the end or at the beginning of the survey questionnaire? (personal communication with D. Dillman, professor, Washington State University, April 8, 1999). Dillman (1978) originally recommended placing the demographics, usually perceived by respondents to be the least and socially important items, at the end, rather than at the (more conventional) beginning of the survey instrument. Although counterintuitive to some, Dillman recommended the strategy because he believed that recipients probably begin answering a questionnaire (rather than discard or set it aside) if the initial questions are perceived to be interesting and socially useful ones (Dillman, 1978, p. 127). Once having invested time in answering these more appealing items, questionnaire recipients are more likely to continue and complete the more ordinary (demographic) questions than if the demographics had been placed at the beginning of the questionnaire. Notwithstanding the lack of empirical testing, Dillman's (1978) technique has influenced the social work literature substantially. Indeed, we found that seven of the nine most recently published social work research textbooks--Grinnell (1997), Marlow (1997), Monette, Sullivan, and Dejong, (1999), Royce (1999), Rubin and Babbie (1997), and York (1997, 1998)--recommended placing demographic questions at the end of the questionnaire. Although authors of two texts did not make a specific recommendation (Fortune & Reid, 1998; Yegidis, Weinbach, & Morrison-Rodriquez, 1999), none of the authors encouraged survey researchers to place demographics at the beginning of the questionnaire or suggested that demographic placement is irrelevant to either response rate or amount of missing data. To test Dillman's proposition empirically, we conducted an experimental study, in conjunction with a series of larger statewide surveys, to evaluate whether survey participants who received mailed questionnaires with demographic questions placed at the end of the research instrument (form E) would produce a higher response rate and less missing data than recipients who received a parallel form of the questionnaire in which the same demographic questions had been placed at the beginning (form B). METHOD Procedure Through the use of a recent 1999 NASW Virginia Chapter membership roster, systematic random sampling (Grinnell, 1997) was used to select 1,100 social workers from the 3,374 members of the chapter to participate in one of three jointly administered mail surveys. Because each of the research questions investigated in the larger series of surveys required the same demographic variables, the variables were measured in precisely the same way in all surveys. …