Immo Fritsche and Volker Linneweber
Participants in social science research usually think. That is, they interpret the actions of a researcher and relate it to their own beliefs, emotions, and intentions. Accordingly, the behavior they exhibit during investigations is controlled by what they think is appropriate, depending on their interpretation of the study situation and their motivation to comply with these assumed requirements. Hence, what is often measured is not a "natural state" but the participant's intentional presentation. Imagine inviting people to participate in a study on helping behavior in which you are interested in the individual inclination to help others in need. You ask the subjects to indicate whether they would be willing to donate blood for a charitable organization. In many cases, however, the answer you receive might not say a lot about the participant's actual behavior or corresponding intentions, but rather about the participant's proper understanding of the study's demand characteristics (Orne, 1962) or about the relevant social desirability norms.
However, even if we adopt the more optimistic view that research participants do not think (even not unconsciously), the validity of the data obtained in investigations specifically designed to record some variable of interest might still be doubtful. That is because we can at least expect thinking on the part of the researcher. More specifically, we may assume that the researcher arranges the research setting in such a way to allow the practical and convenient measurement of the construct of interest. Unfortunately, even the nonthinking participant's willingness to donate blood does not necessarily say a lot about his daily conduct. We can suppose that throughout the course of his daily life, your subject will never be confronted with appeals for help like the one you were kind enough to present in your investigation.
Participant behavior under the influence of thinking by both the participant and the researcher can be called "reactive" if it is a subject's reaction to a specific situation, intentionally created for research by a researcher. Here, bias occurs not only by fault of the participants (subject bias) but also due to the influence of the researcher (experimenter or observer bias1) as well. Reactive measurement restricts what Brunswik (1947) called the data's "ecological validity." Although not yet ultimately defined in the psychological science, high ecological validity indicates that the results from an investigation may predict the item's behavior in its ordinary context.
Psychologists have tried to cope with the shortcomings of reactive methods by introducing various measures aimed at preventing subject and experimenter biases. Probably the most radical approach has been the proposal of "nonreactive" or "unobtrusive" measures, first stated in the groundbreaking work of Webb, Campbell, Schwartz, and Sechrest (1966). They systematically start discussing what has been previously known as "reactive effect of measurement" and "reactive arrangement bias"
'For the dynamics of experimenter effects, see also research on what has been called the "Rosenthal effect" in psychological experimentation (Rosenthal, 1976) and the "Pygmalion effect" in education (McNatt, 2000).
(p. 13). The table of contents reflects what was of interest when psychology started discussing alternatives to traditional measurement. Nonreactive measures suggested by Webb et al. (1966) include an examination of physical traces such as natural erosion and accretion and archives such as those found in various public and private records. Finally, they elaborated on the potential utility of simple and contrived observation methods for unobtrusive measures. Even though these unobtrusive measures refer explicitly to overt behavior and its products, it is often a goal of researchers to measure not only the behavior in a nonreactive manner, but internal variables such as attitudes, emotions, and abilities as well. More than 10 years later, in a reader edited by Lee Sechrest (1979), Bochner (1979) added some considerations concerning unobtrusive field experiments in social psychology, thus demonstrating that further developments occurred, particularly with respect to the issue of social desirability and in applied research in general.
Was this article helpful?