4. Name the three threats to measurement accuracy and reliability. Define each, give examples of each, and summarize how we can minimize these threats.
-Poorly Designed Measurement System: When measurement systems are unorganized and difficult to use, this can reduce the accuracy and reliability of data collection. For example, if a teacher is taking data on five different data sheets, her data may not be as accurate as having one data sheet for data collection. Simple to use measurement systems, reduction of individuals being observed simultaneously and decreasing the duration of the sessions for observation can all help the data collector to have accurate and reliable findings.
-Inadequate Observer Training: When observers are inadequately trained, this can result in a threat to accuracy and reliability of their data collection. The observers must be able to discriminate between the presence or lack of target behavior and how to record that (using the correct symbols). For example, if an observer is inadequately trained, they may record an incorrect response as correct if they use the incorrect symbol, which completely alters the results and accuracy of the data collection. For these reasons it is vital observers are given preliminary exams prior to becoming a data collector to analyze whether that individual would be able to sufficiently take data. Additionally, before the observer is allowed to take data in actual ABA settings, they should be required to meet a certain criteria, such as taking data with 90% accuracy over several trials.
-Unintended Influences on Observers: Observer expectation and observer reactivity can affect the accuracy and reliability of data collected. For instance, if the teacher or observer expects the target behavior should be successful while using a visual schedule with the student, this may result in less recording of screaming behavior, as the teacher expects the student will be doing better with this format. Additionally, when observers are under the impression they are being monitored or that others will be taking data along with them, the observer may react and score how they believe the supervisor, for instance, would take data. It is essential to use naïve observers in order to reduce measurement bias, or to have data taken from another room using video for instance, to reduce observer reactivity.