Reactive Arrangements

Reactive arrangements are an example of a threat to the internal validity of a research design. Reactivity occurs when the results of an experiment are due, at least in part, to behaviors of participants that artificially result from their participation in the experiment. Thomas D. Cook and Donald T. Campbell and colleagues, in their works defining threats to validity in research design, distinguish between artifacts due to the use of measures in a study and other reactions that occur because participants are aware that they are in a study. When these reactions become a functional part of the treatment or independent variable, then reactive arrangements are present.

Reactivity has occurred when the “meaning” of a treatment includes the human reactions to being in a study. Reactions to the study procedures themselves may occur in several different ways:

• Participants may respond favorably after receiving a nonactive drug used as a control (the placebo effect).
• Study subjects may guess or ascertain the expected outcome of an experiment and behave in ways they feel will please the researcher.
• Participants may perform higher or lower on skill or achievement measures because of increased motivation or anxiety-induced interference.
• Apprehension of being evaluated by a “scientist” may encourage some subjects to produce expected responses.

Protections against the threat of reactive arrangements center on hiding the true purpose of a study by making all control treatments appear to be authentic, measuring the outcome surreptitiously, and designing pretest measures so as not to give cues for expected outcomes. A researcher examining the effectiveness of a psychological therapeutic intervention, for example, may replace a no-treatment group with a group that receives another approach to therapy that is believed to be weaker or ineffective. Many studies use this approach with their treatment-as-usual group. Another approach is for researchers to make the measurement of outcomes less obvious by not assessing the outcome at the immediate conclusion of the treatment or procedures but, if possible and still theoretically meaningful, assessing outcomes on some delayed basis.

Participants will almost always make their own hypotheses as to the purpose of a study, however, so complete protection against this threat may not be possible. There are also ethical requirements to give enough information about the purpose of a proposed study so that recruits can give free and informed consent to take part. Some extreme ethical positions might even require that the researcher share the study hypotheses with all recruits. Most researchers, however, control for the threat of reactive arrangements through design choices or, at the least, measure variables in ways that allow for a determination of the presence of reactive arrangements.

Reactive arrangements are a particularly tough threat to compensate for, because they are among those threats to validity that are not eliminated or lessened by random assignment to different groups. Random assignment can control for preexisting differences between members of different groups, but in the case of reactive arrangements, the differences between groups do not exist until the study begins. While group comparison is the basis of modern experimental research, one consequence of this analytic technique is the possibility of reactive arrangements.

...