In the literature, the term nonreactive measurement is used in at least two senses, in a dichotomous as well as a continuous sense. Textbooks often think of nonreactive measures as representing a distinct set of procedures sharply different from reactive methods. Actually, there is a comparatively stable core of measures that are commonly subsumed under this heading. In social psychology, for instance, one of the most cited and consequently most prototypical nonreactive method is the so-
called lost letter technique (Milgram, Mann, & Harter, 1965), which is used as an indirect measure of attitudes. To implement this technique, stamped letters are distributed in specified residential areas, appearing to the chance observer to be lost by someone. They are addressed to different organizations, representing the attitude objects. For example, if researchers are interested in the relative approval of religious groups like Christian and Muslim, not biased by social desirability concerns, they could "lose" those letters addressed to either a church or a mosque. In actuality, the letters are addressed to the researcher via a post office box. The number of letters the researchers receive is taken as an indicator of the prevailing attitude toward the respective group (i.e., Christian or Muslim) in a specified area (for a test of the technique's convergent validity, see Cherulnik, 1975; for an advanced version, see also the lost e-mail technique by Castelli, Zogmaister, & Arcuri, 2001; also applied in Vaes, Paladino, Castelli, Leyens, & Gio-vanazzi, 2002). Although the interpretation of this method might be flawed by some problems (see following text), the technique guarantees nonreactive measurement insofar as the participants are not aware of their participation in an attitude test. This nonawareness is the most important criterion of nonreactivity that was identified by Webb et al. (1966) in their classic work on unobtrusive measures. This term is applied to those measures in which the studied individual "is not aware of being tested and there is little danger that the act of measurement will itself serve as a force for change or elicit role-playing that confounds the data" (p. 175). Unobtrusiveness and nonreactivity are often used synonymously. However, the term nonreactivity should be differentiated from unobtrusiveness in two ways. First, nonreactive measurement should be defined in a more comprehensive way than has been done for unobtrusive techniques (cf. Folger & Belew, 1985), not only focusing on whether and how research subjects perceive the act of measurement but also keeping in mind the researcher and her potential contribution to instrument reactivity (experimenter or observer effects; e.g., Rosenthal, 1976). Ideally, not only the subject but also the researcher should not be aware of the measurement when it occurs. As the second aspect of definition, it should be made clear that nonreactivity refers to the ultimate outcome of measurement and not to the means by which this end is achieved (i.e., that the subjects are not aware of being tested). Hence, we define nonreactive measures simply as those measures in that participant's behavior that are not influenced by social interaction with the researcher, because both directions of this interaction are perceived as potential sources of distortion. However, although measures like the lost letter technique are commonly perceived as nonreactive, in many of these measures an interaction between researcher and research subject occurs that might influence the participant's behavior in an indirect fashion. With regard to our example, the letters are not really lost but are placed in precise locations by researchers. Time and place of the letter distribution as well as the choice of addressee follow intentional action by the researcher that might subtly influence the participant's behavior in the sense of "cuing" (Folger & Belew, 1985). One might be able to counteract these potential distortions by balancing time, place, and research assistant according to the variables assumed to be confounding like the economic background of the sample, social status, or political attitudes of the assistants, and so on. Although these additional steps would be likely to reduce reactivity, they are not a unique part of the technique itself. Furthermore, the additional techniques that reduce a measurement's reactivity are incorporated into most psychological methods of data acquisition. Therefore, instead of thinking of reactive and nonreactive measures in a dichotomous way, a continuous definition of nonreactivity might actually be more suitable.
Was this article helpful?