Volunteer Subjects

Volunteer subjects (individuals who volunteer to participate in a research investigation) became a focus of attention in the 1960s and early 1970s, stemming from findings that volunteers may respond differently from how nonvolunteers in the general population would respond. Ralph L. Rosnow and Robert Rosenthal (1970) suggested that SYSTEMATIC ERRORS resulting from volunteer bias, although not easily discerned, may exert a subtle influence on experimental and nonexperimental social science research. Through experimental studies, as well as an integrative review of extant literature, Rosenthal and Rosnow (1975) showed that volunteers differed from nonvolunteers in at least 17 characteristics (including education, social class, intelligence, sociability, and approval motivation), and that volunteer subjects are overly sensitive and accommodating to DEMAND CHARACTERISTICS.

According to Rosnow and Rosenthal (1997, pp. 89–91), these findings have at least three practical implications for the social scientist. First, using only volunteer subjects may serve to weaken the explanatory power of the scientific model under study, because the [Page 1187]volunteer effect is usually not addressed explicitly as a variable in the model. A second implication has to do with the GENERALIZATION of research results to a population that typically includes nonvolunteers. A third implication pertains to ethical considerations as set out in the Belmont Report, developed in 1974 by a national commission in the United States charged with the task of formulating guidelines that would protect the rights and welfare of human research participants in biomedical and behavioral research (National Commission, 1979). These guidelines specified that only volunteer subjects could be used in most research investigations. In effect, this development implied that the volunteer problem could not be eliminated in RESEARCH DESIGN, only controlled or corrected for. Even so, precise estimation of the extent of BIAS remains a daunting task, because population parameters are rarely known. The best that the scientist can do in most instances is to speculate on the direction of the bias and interpret results accordingly. Fortunately, our knowledge of reliable characteristics of volunteers, and the reliability of volunteering for various types of studies, listed by Rosnow and Rosenthal (1997, pp. 95–98), can be used in this speculation.

An issue related to the volunteer subject is that of the pseudovolunteer, a verbal volunteer who fails to show up for a research appointment. Early studies had shown that research conclusions, including those examining differences between volunteers and nonvolunteers, depend significantly on whether pseudo volunteers are included with the volunteer or the nonvolunteer group. Through a meta-analysis of research reports, Rosenthal and Rosnow (1975) found that, on average, approximately a third of those volunteering did not keep their appointments; more recent studies have confirmed that the MEDIAN pseudo volunteering rate has not changed noticeably since 1975.