The purpose of this study was to extend evaluation of rater accuracy using "Direct Behavior Rating--Single-Item Scales" (DBR-SIS). Extension of prior research was accomplished through use of criterion ratings derived from both systematic direct observation and expert DBR-SIS scores, and also through control of the durations over which the target behavior was displayed in each clip. Rating targets included "academically engaged," "disruptive," and "respectful" behavior. Undergraduate participants ("n" = 113) viewed video clips of classroom instruction and subsequently rated target students' behavior using DBR-SIS. Results indicated that both rater bias and rater error were present to some degree based on the behavior target, wording, and duration displayed, with greater bias found for the target involving respectful behavior. In addition, positive wording for academically engaged behavior resulted in improved accuracy, yet negative wording resulted in improved accuracy for disruptive and possibly respectful behavior. Implications for research and practice are discussed.