There have been repeated calls for studies in behavioral science and human-computer interaction (HCI) research to measure participants’ actual behaviors. HCI research studies often use multiple constructs as perceived measures of behavior, which are captured using participants’ self-reports on surveys. Response biases, however, are a widespread threat to the validity of self-report measures. To mitigate this threat to validity, we propose that studies in HCI measure actual behaviors in appropriate contexts rather than solely perceptions. We report an example of using movements that reflect both actual behavior and behavioral changes measured within a health care IS usage context, specifically the detection and alleviation of neuromuscular degenerative disease. We propose and test a method of monitoring mouse-cursor movements to detect hand tremors in real time when individuals are using websites. Our work suggests that analyzing hand movements as an actual (rather than perceptual) measure of usage could enrich other areas of IS research (e.g., technology acceptance, efficacy, fear, etc.), in which perceptions of states and behavior are measured post hoc to the interaction and subject to the threats of various forms of response bias.