A Proof of Concept Analysis of Decision-Making with Time-Series Data.

Behavior analysts make dozens of practice-related decisions every day. Past research has extensively examined practice-related decision making by medical and other healthcare professionals, but decision making by practicing behavior analysts has garnered little research attention. The purpose of this proof of concept study was to begin a translational research agenda toward understanding what variables influence behavior analyst decision making and precisely how those variables do so. Behavior analyst decision making often involves data and the most fundamental characteristics of data are those indicating data trustworthiness—validity, reliability, and accuracy. To isolate a single independent variable, we used a lengthening data path procedure to parametrically assess how reducing data accuracy changed decisions to continue or modify an intervention in 30 students of behavior-analytic masters or doctoral programs. When data accuracy was 100%, most participants waited 9–10 trials before intervening. When data accuracy was below 60%, most participants waited 4–6 trials before intervening. To begin exploring potential behavioral processes underlying the influence of data accuracy on practice-related decisions, we also examined how well one popular description of decision making described participant choice—probability discounting. Probability discounting described the pattern of choices well for 16 of the 30 participants, suggesting other analytic frameworks should be explored. Nevertheless, data accuracy systematically influenced the majority of participants choices to continue or modify an intervention, although the degree of influence differed on an individual basis. [ABSTRACT FROM AUTHOR]