On the Trustworthiness of Behavior Measurement

In general, scientific measurement is considered to be only an estimate of the phenomenon studied and the goal of any investigation is to arrive at the most accurate measure possible. Still, any experiment containing measurement errors risks misleading theory, the direction of future investigations, and science-based practice. For these reasons avoiding such errors should be of core interest for any science. Questions about the validity of behavioral data are relatively rare in behavior analysis compared to other branches of psychology. This is often related to the fact that many behavior analytic studies find a way to measure the behavior of interest directly and thereby reduce the influence of errors related to whether the units in an experiment represent the higher-order concept in a meaningful way. Nevertheless, choices made with regards to measurement procedures might yield unwarranted or invalid representation of the behavior and such issues cannot be denied on the basis of direct measurements. In addition, some studies will have to rely solely or in part, on indirect measures, for example because of limited resources or because of the nature of the phenomenon of interest. In such cases concept validity should be a central part in the planning, implementation and evaluation of the study. Finally, accuracy, reliability and believability as important indicators of trustworthy measurement will be accounted for. Some examples of assessment of these issues will be presented and their relative contribution to the quality of behavioral measurement will be discussed.